Analyst Private Information Acquisition and Stock Price Synchronicity ‐A Regulatory Perspective from China Abstract: This study investigates whether analysts’ private information acquisition influences the extent to which firm-specific information is capitalized into stock prices, as measured by stock price synchronicity in Chinese stock markets. We also study the moderating effect of the restrictions on selective disclosures imposed by the CSRC Directive 40 on the relationship between analyst private information and synchronicity. We find that synchronicity correlates negatively with analyst private information content, and also moves in a direction opposite to analyst private information acquisition over our sample period. This finding supports the information role of analysts’ private information seeking activities. We uncover that analyst private information acquisition reduces substantially after the 2007 regulation, which partially supports the effectiveness of Directive 40 in its objective of ‘levelling the information playing field’. However, the synchronicityreducing effect of analyst private information acquisition is attenuated after Directive 40, implying an impaired firm-specific information flow attributable to the restriction on selective disclosures. Keywords: Analyst private information acquisition; stock price synchronicity; selective disclosures; regulation; China 1 1. Introduction This study aims to investigate the effect of analyst private information acquisition on the stock price synchronicity of Chinese listed companies, and the way this relationship is shaped by the regulatory restriction on selective disclosures imposed by the Chinese 2007 regulation on selective disclosures. By conducting stock valuation using their specialized knowledge, and then disseminating information via earnings forecasts and stock recommendations, analysts provide valuable input for investors to use in estimating firms’ future earnings and in decision making (Barron et al., 2002a). It is recognized that both publicly available and privately acquired information are important to financial analysts (e.g., Arnold & Moizer, 1984; Barker, 1999; Pike et al., 1993). The importance of private information acquisition is well addressed in the literature (e.g., Barker, 1998; Holland, 2006). Although most of the extant evidence is from developed stock markets, in line with this argument, recent Chinese studies demonstrate that Chinese analysts exhibit better performance when they have more access to a firm’s indirect information and conduct more company-level surveys (Hu et al., 2008). With their access to macro-economic, firm-specific information and specialized knowledge, securities analysts play an important role in improving market efficiency, and more accurate analyst forecasts should reduce information asymmetry and improve the efficiency of equity valuation (Frankel et al., 2006). We use stock price synchronicity (hereafter synchronicity) to measure firms’ information environment because synchronicity is a measure of the relative amount of firm-specific information impounded into stock price. Prior literature provides evidence that firms with a better information environment have 2 lower synchronicity (Jin & Myers, 2006; Morck et al., 2000; Wurgler, 2000).1 From a market perspective, analyst activity, as a whole, can make more fundamental information capitalize into stock prices, and thus lead to lower stock price synchronicity and higher firm specific information content. However, using a sample of 45 emerging markets, including China, Chan & Hameed (2006) find that more analyst coverage leads to an increase in stock price synchronicity, indicating that analysts following emerging market stocks are primarily involved in the production of industry-wide and/or market-wide information rather than the costly acquisition of private information that is idiosyncratic to a firm. Piotroski & Roulstone (2004) and Ayers & Freeman (2003) find a similar positive relation between analyst coverage and synchronicity using US data. One problem that plagues the above analyst-synchronicity studies is that their measure of analyst coverage does not speak to the nature, public or private, of the information that analysts use in their forecasts. For example, analyst reliance on public or private information for analysis and forecasts may have a distinct effect on the capitalization of firm-specific information into stock prices. Thus, our study advances this topic by focusing on analyst private information acquisition. Specifically, we use analyst private information acquisition as a direct measurement of analyst effort in seeking private information, following Barron et al. (1998) to investigate the relation between this acquisition and firms’ stock price synchronicity. We posit that analyst private information acquisition reduces the 1 Lately, there is some doubt regarding synchronicity as a measure of firm-specific information incorporated into share prices in the international context (Alves et al., 2009; Ashbaugh-Skaife et al., 2005). Many of the concerns are related to other confounding factors that cannot be easily controlled in a cross-country study such as the size of the country, investor property rights, corruption levels and the role of the government. To check the validity of synchronicity as a proxy for firm information environment in a single country-setting, Gul et al. (2010) conduct an Earnings Responses Coefficients (ERC) test. Their analysis confirms that that synchronicity is a robust proxy for the usefulness a firm’s information in the Chinese context. 3 synchronic component of stock returns and thus improves the information environment for firms in China. The China Securities Regulatory Commission (CSRC) promulgated and put into effect ‘The Listed Companies Information Disclosures Administrative Rules’ (hereafter Directive 40) on 30 January 2007. The purpose of Directive 40 is to restrict listed companies’ selective disclosures of material information to financial analysts and institutional investors before their public disclosures in order to curb insider trading, ‘level the information playing field’, and protect individual investors. In spirit, Directive 40 is largely similar to the Regulation Fair Disclosure (hereafter Reg FD) that was issued by the US Securities Exchange Commission (SEC) in 2000. Due to the restriction of Directive 40 on private information supply, it is important to inquire whether there is an information gap as a result of this restriction, and whether the public information supply is sufficient to fill the potential information gap resulting in unchanged firm-specific information incorporated into stock prices post-regulation. Thus, our second research question investigates whether and how 2007 Directive 40 modifies the association between analyst private information content and synchronicity. Additionally, we also examine whether private information acquisition and synchronicity differs between affiliated and non-affiliated analysts, and whether this relationship is dependent on the institutional development of the province where a listed company resides. The distinct features of the Chinese institutional environment make Chinese data particularly suitable for our research questions. First, China’s widespread social connection network eases analysts’ private information seeking. ‘Guanxi’, translated as ‘personal connection’, facilitates private meetings and other flexible communication between analysts and management (Jing et al., 2012). It is reported that Chinese local analysts have 4 information advantages, measured as greater earnings forecast accuracy, due to their physical proximity to the firms or their affiliated relationships (Bartholdy & Feng, 2013). Directive 40, aiming to eliminate private information communication between firms and their analysts, may be difficult to implement owing to these social connections and cultural expectations. Furthermore, the investment service industry is at an early stage of development. Chinese analysts are less competent than their counterparts in developed countries. They often rely on private information obtained informally for stock valuation, earnings forecasts, and recommendations. Empirical studies in China have also reported uninterrupted private communication between firms and professional market participants post-Directive 40, and thus the effectiveness of regulatory restriction on private information acquisition is limited. Second, the information asymmetry level of Chinese listed companies is high. The truthfulness of financial information is impaired by opportunistic reporting and accounting manipulation (Chen & Yuan, 2004; Ding et al., 2007; Kimbro, 2005; Wang et al., 2008). Also, the voluntary disclosures of firms are deficient. There exists a high degree of information asymmetry between listed companies and investors (Piotroski & Wong, 2012). In this context, analysts’ private information is important in meeting the information demand of investors. How the restriction on selective disclosures impacts on analysts’ performance and the information environment of firms will be of interest to regulators and market participants. Our study will be informative in this regard. Lastly, Directive 40 in China presents a laboratory setting for drawing theoretical and regulatory inferences on the efficiency of regulatory restrictions on private information acquisition in an institutional environment that differs considerably from that of the US. Reg FD is implemented in the US, where there is stringent legislation and penalties against accounting and financial fraud, resulting in comparatively high quality accounting information. Also, US firms have greater transparency levels, as is evidenced by the greater extent of their public disclosures and their willingness to 5 make voluntary disclosures. In this context, US analysts can obtain sufficient reliable and relevant information from public sources for their analysis post-Reg FD (Kross & Suk, 2012). In contrast, Chinese analysts rely mainly on private information-seeking and, owing to the aforementioned reasons, may face an enlarged information gap post-Direction 40. Our inquiry into the economic consequences of Directive 40 in relation to analysts’ private information acquisition will shed light on the efficacy and efficiency of this regulation in a distinct institutional environment. We find that in general, synchronicity correlates with analyst private information content negatively, and moves in a direction opposite to analyst private information content during our sample period. This finding supports the informational role of analysts’ private information-seeking activities. In addition, we document that analysts’ private information acquisition reduces substantially after Directive 40. Furthermore, this reduction is more pronounced for non-affiliated than for affiliated analysts, indicating a continuous availability of other private communications between firms and their affiliated analysts. We further reveal that the synchronicity-reducing effect of analysts’ private information acquisition is attenuated after Directive 40 taking effect in 2007, implying impaired firm-specific information attributable to the restriction on selective disclosures. We also test whether the moderating effect of Directive 40 on synchronicity-analyst private information varies between affiliated and non-affiliated analysts. We find that the impairment of firm-specific information post-regulation is significantly lower for firms followed by affiliated analysts possibly due to their continuous access to private information post-regulation. Lastly, our additional analysis shows a significant modifying effect on the above relationship from the level of financial market development in the province where a firm domiciles. Overall, the results suggest efficacy of Directive 40, as evidenced by the decrease in analyst private information acquisition in recent years. However, our findings also pinpoint a lack of 6 efficiency of Directive 40 from the market perspective, in that synchronicity is increased along with the decrease of analyst private information. Our findings highlight that regulation to restrict selective disclosures that is effective in a low information asymmetry and more transparent public disclosures (e.g., the US) context, is not necessarily applicable to countries where the information asymmetry level is high, public disclosure level is low, and the financial analyst industry is still in its infancy. The next section discusses the relevant literature, describes the characteristics of the Chinese analyst industry and followed by the development of hypotheses. Section 3 describes the research methodology. Section 4 explains the test conducted, and reports the empirical findings. Section 5 concludes the paper. 2. Literature survey, China analyst industry and development of hypotheses 2.1 Literature survey The role of financial analysts is to analyze publicly available information (e.g., in respect of the macro economy, industry and company) and collect information through particular channels (e.g., company visits, meetings with company management). After synthesis and analysis, analysts transfer their information output to the public. Thus, analysts serve as an important mechanism to distribute information and improve capital market efficiency. It is recognized that accounting information, especially annual reports and direct contact with a company, are the most important and useful sources of information to financial analysts, even though there is a clear shift in the relative importance of these sources over time (e.g., Arnold & Moizer, 1984; Barker, 1999; Pike et al., 1993). 7 The importance of private information acquisition is well addressed in the literature. Barker (1998) contends that analysts and fund managers can draw substantial value from meeting directly with companies because such meetings offer an opportunity to assess a company’s strategies and management abilities. Thus, analysts rank them as the most important source of information. Holland (2006) investigates how fund managers utilize information obtained from companies directly in order to overcome problems with publicly available information. For instance, financial reports are perceived to be too complex and overwhelming by many users. In all, the literature suggests that information sources, both public and private, have significant impact on the performance of analysts. With respect to the effect of analyst activities on synchronicity, the literature provides inconclusive findings. On the one hand, contrary to the conventional wisdom that security analysts specialize in the production of firm-specific information, several studies find that securities which are covered by more analysts incorporate greater (lesser) market-wide (firmspecific) information. Piotroski & Roulstone (2004) find that in the US, although the presence of insiders and large institutional investors has the net effect of increasing the amount of firm-specific information that is incorporated into stock prices, security analysts decrease that amount. In other words, in the US, security analysts do not have an advantage over insiders and institutional investors in accessing firm-specific information. Ayers & Freeman (2003) also report that price synchronicity is associated with analyst forecasting activities positively and inversely associated with insider trading. Similarly, Chan & Hameed (2006) demonstrate that more analyst coverage leads to an increase in stock price synchronicity in 45 emerging markets including China, indicating that analysts are primarily involved in the production of industry-wide and/or market-wide information rather than in the costly acquisition of private information that is idiosyncratic to a firm. Their argument is that the payoff to analysts who produce firm-specific information may be too low in emerging 8 markets because it is impossible to arbitrage on them. Also, there is difficulty associated with collecting firm-specific information in emerging markets. One important point not discussed in Chan & Hameed (2006) is that the analysts included in their sample are retrieved from and covered by I/B/E/S International, which collects information from analysts who agree to provide their estimates in return for free use of IBES products. Thus, IBES’s choice of analysts may introduce biases. For instance, analysts from major brokerage houses and foreign analysts are the main population of analysts covered, whereas analysts of smaller brokerage houses from developing countries and regional exchanges may be ignored. 2 Chan & Hameed’s (2006) finding of a positive effect of (foreign) analysts on synchronicity is important in the sense that this finding indicates foreign analysts mainly disseminate industry and market-wide information. On the other hand, analysts’ access to quality information and their analyzing abilities are expected to improve firms’ information environment and result in low synchronicity. Bushman et al. (2004) show greater idiosyncratic variation (low synchronicity) in countries with better developed financial analysis industries. Jin & Myers (2006) study 40 stock markets from 1990-2001 and find countries’ R²s (resulting in high synchronicity by computation) are correlated with opaqueness positively, as measured by the analyst forecast dispersion, Global Opacity Index, Global competitiveness report, and an index of accounting standards. This finding indicates a negative relation between analyst forecast performance (low forecast dispersion) and synchronicity. Kim & Shi (2008) investigate whether voluntary IFRS adoption reduces synchronicity and also study the role of analyst following and institutional infrastructure in shaping the relation between IFRS adoption and R². They find 2 Although IBES does not provide information on the identity of analysts and broker agencies, evidence suggests that the majority of analysts covered by IBES are foreign analysts in the case of China. For instance, Bae et al. (2008) report that in their sample derived from IBES, there are 74 analysts following Chinese firms and 67 of them are foreign analysts. Thus, Chinese domestic analysts conducting stock valuation for most of Ashare firms may have been excluded in Chan & Hameed (2006) as they use IBES as their analyst data source. 9 that the synchronicity-reducing effect of IFRS adoption is muted (strengthened) for firms with high (low) analyst following, and is stronger (weaker) for firms in countries with poor (good) institutional infrastructures. These findings indicate that analysts have already been playing an important role in improving firms’ information environments and reducing synchronicity prior to IFRS adoption. Thus, introducing IFRS in these firms has a minor impact on firms with good analyst coverage. One problem that plagues the above analystsynchronicity studies is that their measure of analyst coverage does not speak to the nature, public or private, of the information that analysts use in their forecasts. For example, an analyst’s reliance on public or private information for analysis and forecasts may have a distinct effect on the capitalization of firm-specific information into stock prices. Instead of using the analyst following, our study takes a more direct approach by focusing on analyst private information acquisition. Due to concerns over insider trading and selective disclosures of material information to market professionals and certain shareholders, US SEC passed Reg FD in 2000 in an effort to level the information playing field among all investors. Even in the US, literature on the effect of Reg FD on US corporations’ information environment is controversial. 3 Studies report that there is substantial evidence that Reg FD is effective in levelling the market participants’ playing field because post-Reg FD firms have reduced selective disclosures, increased their public disclosures and thus ameliorated information asymmetry (e.g., Dong et al., 2012; Sinha & Gadarowski, 2010). Frazzini et al. (2008) examine the effect of Reg FD on analyst behaviours and report that analysts’ ability to exploit social ties in order to gain informational advantages has virtually disappeared in the post-Reg FD era. Studies also find that analysts show increased reliance on public disclosures because firms use more earnings Fisch (2013) provides a detailed summary of the literature on the effect of the US Reg FD on firms’ information environment. 3 10 guidance as a substitute for selective disclosures post-Reg FD (Charoenrook & Lewis, 2007; Kross & Suk, 2012). Nevertheless, critics of Reg FD show that information asymmetry persisted after Reg FD took effect, and selective access to management continues to provide investors with trading advantages via private meetings, etc (Solomon & Soltes, 2013). For instance, Bushee et al. (2013) find that post-Reg FD, some investors’ trading size has increased following private meetings with company management, and this effect increases if the management involved is the CEO. Unusually large trading size normally indicates an investors’ material information advantage, which raises concern about selective communication between firms and certain types of investors. In addition, studies report that post-Reg FD, the ability of institutional investors to identify mispriced stocks is discounted (Cook & Tang, 2010), analysts’ forecast accuracy is reduced (Agrawal et al., 2006), and analysts’ ability to predict earnings surprises is weakened (Palmon & Yezegel, 2011). Gintschel & Markov (2004) examine whether Reg FD has reduced the informativeness of analysts’ information outputs. They find that in the post-Reg FD period the absolute price impact of information disseminated by financial analysts is reduced by 28%, and this drop in price impact varies systematically with brokerage house and stock characteristics related to the level of selective disclosure prior to Reg FD. In all, the countervailing evidence on the effect of the US Reg FD on firms’ information environments makes drawing a definitive conclusion difficult. 2.2 China analyst industry and development of hypotheses The Chinese financial analyst service industry was underdeveloped in the early years, and only limited analysis was conducted by Chinese securities firms. Often their research reports did not contain any quantitative forecasts and in most cases whatever material existed was treated as internal (Li & Fleisher, 2004). During the 1990s, most formal and public 11 analyses were performed by international securities firms on Chinese B-shares. Local groups of analysts appeared over the 1990s and in 1999 a licensing system was adopted. In July 2000, as a self‐disciplinary committee, the Securities Analysts Association of China (SAAC) was established, marking a milestone in the development of China’s securities analysis services. The SAAC is to direct the healthy development of the securities investment consultation industry and to help promote rational investment in China’s securities market (Asian Securities Analysts Federation, 2004). Recent developments in Chinese capital markets have necessitated the development of a financial analyst industry and the importance of financial analysts is increasing considerably. China opened its capital market gradually, resulting in a rapid increase in institutional shareholding. In addition, the introduction of the QFII (Qualified Foreign Institutional Investors4) and QDII (Qualified Domestic Institutional Investors5) programs has stimulated the growth of investment analyses. In addition, after the share split-regulation’s relaxation on the trading restriction of state owned shares in 2005, investors’ demand on professional stock valuation services increased in order to facilitate trading. In 2005, financial analysts got a formal name and ethics regulations were issued. There were about 20 securities firms with more than 700 analysts, which had increased to 106 firms by 2010 (Bartholdy & Feng, 2013). Most Chinese financial analysts work in three types of organisations - securities firms (brokerage firms), fund management firms and consultancy firms. Making investment recommendations is the main service provided by financial analysts irrespective of the departments or organisations in which they work (Wang et al., 2011). 4 QFII is a program that was launched in 2002 to allow licensed foreign investors to buy and sell Chinese currency-denominated A shares in the Shanghai and Shenzhen stock exchanges, which were previously closed off to foreign investors in order for government to exercise tight capital controls and restrict the movement of assets in and out of the country. 5 The QDII program started on 13 April 2006. It allows Chinese domestic financial institutional investors to invest in foreign securities markets via certain fund management institutions, insurance companies, securities companies and other assets management institutions which have been approved by CSRC. 12 Wang & Ahammad (2012) find that Chinese financial analysts use both publicly and privately available information to validate the truthfulness of each source. Specifically, their interview data reveals that company visits and private meetings are regarded ‘useful’ to ‘extremely useful’ for gathering information because they provide price-sensitive information including product cost structure and possible assets injection from the listed company’s principle shareholders. Although public information was weighted higher than private level information by Chinese analysts (Wang et al., 2011), it is found that a private source provides a more efficient way to obtain ‘first hand’ information about the future development of the company (Wang & Ahammad, 2012). Evidence from interviews with sample analysts shows that private information seeking is highly valued in the industry, in that analysts are required by their employers to conduct site visits from twice quarterly to once a year, with key companies from the industry allocated to them. One of the main objectives of the company visit is to search for price sensitive information that is unavailable to the public at that time. Also, the ‘effectiveness’ of the conversation largely depends on the relationship or ‘Guanxi’ between analysts and managers. In addition, verifying the truthfulness of published financial information is another important aim of company visits, especially when analysts are sceptical of the quality of publicly available financial information (Wang & Ahammad, 2012). Furthermore, verifying the real asset value is also an objective of company visits because the historical values of non-current assets in those state ownership enterprises (SOEs), that is recorded during privatization, do not reflect the true financial position of the company in a fast growing economy. Therefore, analysts often make judgements about the true value of those assets during visits (Wang & Ahammad, 2012). Lastly, other information that analysts normally seek includes firms’ long-term investments, financing plan, management abilities, business 13 strategies, as well as their access to ‘political lending’ and government connections that are highly valued in the Chinese business context. After collecting information, Chinese analysts analyze data using a module that is similar to the one being used in developed countries. This module is designed firstly to assess the macro-economic information, then the industry-based information, and lastly the firmspecific information. Interview evidence from Wang & Ahammad (2012) suggests that company-based analysis is based on integrated information from every aspect. They also illustrate how private meeting information such as investment strategies and financing plans are used for analyst stock valuation. In line with the above interview evidence, empirical studies show Chinese analysts tend to exhibit greater information comprehension and better job quality or performance when they have more access to a firm’s indirect information and conduct more companylevel surveys (Hu et al., 2008). Studies also report a local analyst forecast advantage evidenced by lower earnings forecast errors of local analysts compared with those of nonlocal analysts (e.g., Li et al., 2011).6 Bartholdy & Feng (2013) study Chinese A-share listed companies over the period 2002-2009, and find that affiliated securities firms, defined as securities firms acting as investment banker or underwriter, provide better earnings forecasts than un‐affiliated firms. Also, forecast errors produced by local securities firms and star analysts are smaller. Those findings on the analysts’ home advantage may be related to the information advantage of private information acquisition. 6 Most of the studies on local analyst advantages are conducted by Chinese scholars and many publications are in the Chinese language. Due to the underdevelopment of the financial analyst profession in mainland China, early Chinese analyst studies focus mainly on H and B shares, thanks to the availability of financial analyst data, more mature security markets, and the greater transparency of H- and B-share firms. The published evidence in the English language about A-share markets is scant. For a review of the analyst literature on H- and B-share markets, please refer to Bartholdy & Feng (2013). 14 Although the international literature is inconclusive on the association between synchronicity and analyst activities normally measured as analyst following or forecast dispersion, it is reported in China that analyst activity, as a whole, can generate more fundamental information capitalized into stock prices and, thus, lead to lower stock price synchronicity and higher firm specific information content (Zhu et al., 2007). Given the importance of analyst private information in the Chinese context as aforementioned, we predict a negative effect of analyst private information acquisition on synchronicity. H1: Analyst private information acquisition moves in a direction opposite to synchronicity. Directive 40 issued on 30 January 2007 introduced the discipline of ‘fair disclosure’, along with the requirement that information disclosure should be ‘truthful, accurate, complete, and timely’. Article 2 states that information disclosure should be ‘public and simultaneous’. Article 4 states that ‘No person with knowledge of inside information shall, prior to a lawful disclosure of inside information, make public or disclose such information or conduct insider trading with such information’. Shortly after this regulation, empirical evidence tends to draw inferences on its effectiveness. Studies published in the Chinese language report controversial evidence on the effectiveness and efficiency of Directive 40. On the one hand, studies report reduced selective disclosures by firms to analysts and other institutional investors, and larger analyst forecast dispersion and errors post-regulation. This increase in forecast errors deteriorates with the years post Directive 40. Additionally, this increase in analyst forecast errors post-regulation is most severe for the firms with inferior public disclosures (Liu & Peng, 2012). Thus, as a result of Directive 40, the information flow in the market is reduced (Zhu & Wang, 2009). On the other hand, evidence has shown a continuous supply of private information by firms to analysts and other market participants, 15 especially affiliated financial analysts. It is reported that there still exists material earnings information leakage before pre-earnings announcements in the post-regulation era (Yang, 2010), and insider trading based on such private information still exists (Yang, 2012). Literature on the effect of Directive 40 in English is absent, with the exception of Gong (2012) which studies the impact of the 2007 regulations on the information environment of financial analysts in Chinese A-share markets. He finds that the forecast error and forecast dispersion of local analysts increased significantly in the post-regulation period, indicating a deteriorated information environment for local analysts and an increased level of information asymmetry between companies and local financial analysts post regulation. However, the forecast error and forecast dispersion of foreign analysts did not show any change in the postregulation period. Based on the mixed evidence, we develop the following hypothesis in the null form. H2a: Analyst private information acquisition is not changed after the implementation of Directive 40. We investigate not only the effect of reform on analyst private information flow, but also the economic consequence of the change in analyst private information content brought about by Directive 40. Gintschel & Markov (2004) investigate whether Reg FD affects the informativeness of analysts’ information output, and report that in the post-Reg FD period, the absolute price impact of information disseminated by financial analysts is lowered by 28%. Their measure of price impact is the absolute return on or around the announcement of analyst earnings forecasts or stock recommendations. We argue that the effect of Chinese Directive 40 may be inefficient from a market information perspective for the following reason. Due to the restriction on selective disclosures, analysts may receive less private information from firms directly. Meanwhile, 16 firms’ public disclosures are insufficient and less useful because of the lack of public disclosures, impaired truthfulness of accounting information resulting from rampant earnings management, and related party transactions (Piotroski & Wong, 2012). In contrast, the US Reg FD prompts analysts’ reliance on firms’ public disclosures in that US analysts rely more on firms’ earnings announcements, management forecasts and conference calls for their analysis post-Reg FD and, as a result, their forecast dispersion and errors decline post-Reg FD (Kross & Suk, 2012).7 Therefore, Chinese analysts may face an even greater information asymmetry problem after regulatory restriction on selective disclosures. Studies report an increase in analyst forecast dispersion and errors after Directive 40 (Gong, 2012). Additionally, this increase in analyst forecast errors post-regulation is most severe for firms with inferior public disclosures (Liu & Peng, 2012). Therefore, we argue that this enlarged information gap will reduce firm-specific information content in stock price leading to larger synchronicity post-regulation. H2b: Directive 40 mitigates the synchronicity-reducing effect of analyst private information acquisition. In addition, we argue that the effect of analyst forecasts on synchronicity is dependent on the type of analysts concerned. Xu et al. (2013) compare the effect of the star analyst on stock price synchronicity with the effect of non-star analysts. They find that star analyst coverage actually decreases synchronicity, whereas there is a positive association between analyst coverage and synchronicity for non-star analysts. Their findings suggest the heterogeneity of analysts in China. The analysis further attributes star analysts’ better performance to their firm-specific experiences as proxied by the number of years and the number of forecasts made by the analyst concerned. We focus on affiliated analysts vs. non7 As reviewed in the literature survey section, the US evidence is also inconclusive regarding whether Reg FD increases or decreases analyst forecast errors. 17 affiliated analysts, because the former have more direct access to information via site visits, private meetings and other communication channels, owing to their economic relationship with firms. In addition, recent Chinese analyst forecast literature finds that private information communication opportunities between firms and their affiliated analysts remain post-Directive 40 (Yang, 2010). We argue that firms followed mainly by affiliated analysts have relatively higher private information content in their forecasts and thus, those firms should show relatively lower stock price synchronicity post-Directive 40 in comparison to the firms followed mainly by unaffiliated analysts. The following hypothesis is formulated accordingly. H3: Affiliated analysts’ private information acquisition serves to reduce firms’ synchronicity post-Directive 40. Morck et al. (2000) contend that poor property right protection could discourage informed risk arbitrage because expropriation risk can increase the cost of collecting firmspecific information. The reduction in informed trading in the countries with poor investor protection can impede the capitalization of firm-specific information into stock prices resulting in a high stock price synchronicity. Using 20 non-US countries, Jiang et al. (2013) find that stock price informativeness (SPI) reduces with the detachment of voting rights from cash flow rights, and this SPI-reducing effect is attenuated for firms with high analyst following and in countries with strong country-level institutions. Kim & Shi (2012) find that synchronicity decreases with the strength of a country’s institutional infrastructures. Hasan et al. (2013) argue and report that synchronicity is a function of the institutional development in China as measured by property rights, the rule of law, and political pluralism. In addition, a stream of literature reports that legal and financial reporting environments also affect financial analysts forecast characteristics and accuracy (Barniv et al., 2005; Chang et al., 18 2000). Lang et al. (2004) find that analysts’ impact on stock valuation is also modified by investor protection in a country. Therefore, we posit that the effect of analyst private information on synchronicity is conditional upon institutional factors. Specifically, among the institutional factors, we focus on the strength of financial market development only, because the financial analyst industry is an integral part of the financial market development. The Financial Development Index is a sub-index of the Chinese Provincial Institutional Development Index constructed by Fan et al. (2011). The testable hypothesis is formulated as follows. H4: The synchronicity-reducing effect of analyst private information acquisition is conditional on financial market development in the province where a firm resides. 3. Empirical methodology 3.1 Data and sample description The financial analysts’ earnings forecasts for the A-share sample are retrieved from the CSMAR (China Stock Market Trading Database). Specifically, the CSMAR Analysts Forecasts database provides analysts’ forecasts solely for A-shares and is a primary source of broker forecasts, stock recommendations, dates and names of underwriters for IPOs and SEOs and analysts’ rankings. We also retrieve stock trading volume, daily stock returns, daily market returns and annual financials, ownership structures, pre-announcement of earnings and management earnings forecasts information from CSMAR. The MSCI (Morgan Stanley Capital International) World index is obtained from DataStream. The institutional development index is from Fan et al. (2011). 19 Our sample is based on all shares listed on the Chinese A-share market (i.e., A, AH, AB and AHB shares). The sample period is from 2003 to 2012. In July 2000, a professional committee called the Securities Association of China was formed, signalling the analyst industry’s establishment and self‐discipline. However, during the period between 2000 and 2002, a limited number of firms were followed by analysts. For instance, only 46 firms were followed by analysts in 2001, and most of them, by only one financial analyst. So, we can't calculate the variable of private information acquisition (PRIVATE) which requires data from multiple analysts. As a result, our sample starts from year 2003. The sample selection procedure is provided in Table 1. Financial companies are excluded. Shares with high delisting risk (classified as ST shares8 by the CSRC) are removed from our sample. To be included in our sample, a firm must have been continuously trading during the period 20032012. For PRIVATE calculation, we need at least two forecasts. Therefore, 7869 observations are deleted because of an insufficient number of forecasts for PRIVATE’s computation. The number of deleted observations for this reason is much greater in the early years than in the latter years because there is an increasing supply of analyst services in recent years. For R² estimation, we require firms to have at least 200 trading days. For this reason, 1710 firm-year observations are eliminated. Owing to missing data required to calculate the control variables, we further exclude 525 observations. INSERT TABLE 1 ABOUT HERE Panel B of Table 1 reports the industry distribution, showing that the largest industry in our sample is Machinery, Equipment and Instrument (C7), and it accounts for 17.38% of sample observations. The next two largest industries are Petroleum, Chemical, Rubber, 8 ST stands for Special Treatment, which applies to a Chinese listed firm when it reports losses for two consecutive years. Firms with an ST symbol face some trading restrictions, including a daily price fluctuation bracket of more than 5 percent, while the normal price fluctuation allowed for a listed company is 10 percent. If the ST firm continues to suffer a loss in the third year, it will be signified by a “*ST” and suspended from trading. A further loss in the following quarter will de-list the firm. 20 Plastic (C4) and Metal and Nonmetal (C6), contributing 10.28% and 9.74% of total observations respectively. Finance Industry (I) is excluded from the sample. 3.2 Variables All variables are explained in the Appendix. Several main variables of interest are explained in detail as follows. Analyst private information acquisition (PRIVATE) To measure the average proportion of private information conveyed in analysts’ forecasts, we use the private information component (PRIVATE), common information component (COMMON) and analyst consensus construct (CONSENSUS) developed by Barron et al. (1998).9 COMMON and PRIVATE are estimated using observable properties of analysts’ forecasts (i.e., the variance of analysts’ forecasts, D; the squared error in the mean forecast, SE; and the number of forecasts, N) as follows (Barron et al., 1998; Proposition 1, p. 427): ( SE COMMON [(1 CONSENSUS D ) N 1 ) D SE ]2 N D , PRIVATE [(1 1 ) D SE ]2 N COMMON COMMON PRIVATE An alternative measure of analysts’ investment in private information acquisition includes the analyst coverage measured as the number of analysts following a firm, as used by Tong (2007) and Bushman et al. (2004). 9 21 Where N is the number of analysts’ forecasts. D is the sample variance of the analysts’ individual forecasts. SE is the squared error in the mean forecast, calculated as the squared difference between the mean forecast and actual earnings, i.e., (Actual EPS – Mean EPS Forecast)². Barron et al.’s (1998) measurement has been well used in financial analyst studies (Barron et al., 2002b; Barron et al., 2001; Byard & Shaw, 2003; Venkataraman, 2001). An increase in COMMON along with a decrease in PRIVATE is more consistent with a case levelling the playing field than any of the other cases. An increase in PRIVATE along with a decrease in COMMON is inconsistent with the ‘level playing field’ idea. Thus, distinguishing these two cases may provide some important insights into researched issues (Venkataraman, 2001). Stock price synchronicity For the dependent variable, stock price synchronicity, measurement we estimate the market model to decompose total return variations into two components: those related to common (market-wide and/or industry-wide) factors and those related to firm-specific factors. Following Gul et al. (2010), we use four different specifications because of the Chinese market’s segmentation in issuing different types of shares. Specifically, for all three types of share-issuing firms in our sample (i.e. firms with A-shares only, with A+B shares, and with A+H shares), we first estimate the following market model for each fiscal year: RETit 1MKTRETt 2 MKTRETt 1 3 INDRETt 4 INDRETt 1 it (1) Where, for firm i and t, RET represents the daily returns on A-shares traded on either the Shanghai or Shenzhen exchange; MKTRET and INDRET are the value-weighted A-share 22 market return10 and industry return respectively; and ε represents unspecified random factors. INDRET, the industry return is created using all firms within the same industry with firm i’s daily return omitted. In Equation 1, we also include lagged industry and market returns to tackle the concerns over potential non-synchronous trading biases that may result from the use of daily returns for estimating the market model (Scholes & Williams, 1977; French et al., 1987). To address the possibility that Chinese stock returns may be correlated with world market returns, in the following three models we also control for world market factors measured by the MSCI (Morgan Stanley Capital International) World index.11 In addition, to tackle Chinese segmental markets, we use equations 1a, 1b and 1c to estimate the share price co-movement for A shares only firms, A+B share firms and A+H share firms respectively, because returns on stocks of A+B (A+H) share firms are likely to co-move with B-share (Hshare) market factors in addition to A-share market factors in B (H) share markets. This approach has been used by both Yu et al. (2013) and Gul et al. (2010). RETit 1MKTRETt 2WORDRET it (1a) RETit 1MKTRETt 2 MKTRET B 3WORDRET it (1b) 10 A-shares are listed on either the Shanghai or Shenzhen stock exchange. This A-share market return is based on the composite (value-weighted) A-share index which reflects A-share price movements in both the Shanghai and Shenzhen exchanges. The value-weighted A-share market return equals the change in the composite valueweighted A-share indices from day t to day t – 1 deflated by the composite value-weighted A-share index on day t - 1. The composite A-share index data are extracted from the China Stock Market and Accounting Research (CSMAR) database. 11 The MSCI (Morgan Stanley Capital International) World index is a world market index that is based on stock prices of listed companies, representative of 22 stock markets in North America, Europe, and the Asia/Pacific region, and is weighted by the market capitalization of each constituent market. The index data are extracted from the Datastream database. 23 RETit 1MKTRETt 2 MKTRET H 3WORDRET it (1c) In estimating Equation 1 and Equations 1a, 1b and 1c, we require that daily return data be available for at least 200 trading days in each fiscal year. Stock price synchronicity is defined as the ratio of common return variation to total return variation, which is equivalent to R² of the market model used. To circumvent the bounded nature of R2 within [0, 1], we use a logistic transformation of R²: 2 R SYNCH i log( i 2 ) 1 Ri Where, SYNCH i is annual synchronicity for firm i. Ri2 estimated from Equation (1) is denoted as SYNCH (1), while the R²s estimated using Equations 1a, 1b and 1c are represented as SYNCH (1a, 1b, 1c). Institutional Development Provincial Index Given that administrative regions or provinces in China typically exhibit large variations in economic and institutional development, Fan et al. (2011) construct institutional development indices for each province to capture the strength of institutions at the provincial level within China. The larger this index is, the better the institutional development is. This measure incorporates six elements, namely overall market-institutions development; government intervention in markets; private enterprise development; regional protectionism; financial market development; and legal environment. The composite index is used as a control in our regression analysis on equations (2) and (3). For testing H4, we focus on the financial market development sub-index, because it is more directly pertinent to financial analyst activities. 24 4. Test and results 4.1 Descriptive statistics The variables’ descriptive analysis is presented in Table 2. All variables are winsorized at the bottom and top 1% points of their empirical distributions. Panel A shows that our mean values of PRIVATE, COMMON and CONSENSUS are 95.17, 57.47 and 0.53 respectively, which are all lower than those reported in Barron et al. (2002b) with mean values of 207.84, 118.28 and 0.63 over the period of 1986-1997 using US data. In addition, the sample variance of the analysts’ individual forecasts (D) (mean 0.41) and the squared error in the mean forecast (SE) (mean 1.01) are larger than their counterparts reported by Barron et al. (2002) with mean values of 0.067 and 0.638 respectively. The mean values of stock price synchronicity (-0.10, -0.34, -0.31 and -0.15 for SYNCH (1) and SYNCH (1a,1b,1c) and the mean values of the R²(1) and R²(1a,1b,1c) are in accordance with synchronicity studies using Chinese data. For example, the mean values of R² reported by Morck et al. (2000) is 0.453, while our R² values range from 0.42 to 0.47. The mean value of SYNCH reported by Gul et al. (2010) ranges from -0.316 to -0.232 using the sample period 1996-2003, which is comparable to our mean values of SYNCH (1), SYNCH(1a,b,c) that range from -0.34 to -0.10. It is noteworthy that Chinese studies have consistently reported higher R² and SYNCH than US studies. For example, using the same market model defined by Piotroski & Roulstone (2004), we find a much higher R² and SYNCH than their reported mean values, 0.193 for R² and -1.742 for SYNCH. This discrepancy in synchronicity and R² measures among countries is well addressed by Morck et al. (2000) and is interpreted as a significant difference in the information environment across countries. Panel B presents the comparison of variables between pre- and post-regulation periods. There is a decrease in both PRIVATE and COMMON, although the decrease in 25 COMMON is less severe than PRIVATE (from 247 to 64.06 and from 112.99 to 46.09 respectively). Thus, H2a is rejected, suggesting a sharp decline in analyst private information content post-Directive 40. In addition, Venkataraman (2001) suggests that a decrease in PRIVATE along with an increase in COMMON is consistent with the argument of ‘levelling the playing field’ in any test of analyst private information precision (PRIVATE). However, our data doesn’t show the increase in COMMON. Instead, both PRIVATE and COMMON decline sharply, suggesting deteriorated information content for analyst forecast postDirective 40. SYNCH1 has increased significantly from -0.40 to -0.04 due to a sharp increase in R² from the pre-regulation to the post-regulation period. The same increasing trends have been found for A-share only firms with increased SYNCH1a and R²1a post-regulation. Of note, this increasing trend in synchronicity has not been reported by recent Chinese synchronicity studies (Gul et al., 2010; Hasan et al., 2013). Forecast errors (actual EPS – forecast EPS) are more negative in recent years with a value of -2.41 compared to -0.97 preregulation, suggesting analyst forecasts are more optimistic post-regulation. The Institutional Development Index (INSDEV) has improved significantly, which is consistent with the reported statistics (Fan & Wang, 2001; Fan et al., 2009). The increased number of firms with affiliated analysts (AFFLATE), the increases in management earnings forecasts (MEF) and pre-announcement of earnings (PREANCE) are explained by the increasing analyst service and firm-level disclosures in recent years (Piotroski & Wong, 2012). Similarly, Brown et al. (2002), Bushee et al. (2004) and Heflin et al. (2003) report that the frequency and the price impact of public disclosures in the form of management forecasts, public conference calls, and pre-announcements of earnings have increased after Reg FD in the US. In general, most of the control variables have increased post-regulation except for three ownership structure variables, including state ownership (STATE), local government ownership (LOCGOV), and central government ownership (CENGOV). The split share structure regulation taking effect 26 in 2005 lifts the trading restrictions imposed on listed companies’ non-tradable shares (NTS), allowing NTS holders to sell and realize gains from stock price appreciation. With time, those previous NTS holders trade their shares in open markets. As a result of increasing trading activities, we observe a reduction in state ownership in recent years. Panel C reports the correlation matrix among variables. Synchronicity is negatively related to PRIVATE and COMMON, but positively associated with CONSENSUS. REG shows a positive correlation with synchronicity, but a negative correlation with PRIVATE and COMMON, suggesting greater synchronicity and lower PRIVATE and COMMON postregulation compared to pre-regulation. Most of the control variables are significantly correlated with synchronicity except for STATE. As proxies for public disclosures, PREANCE and MEF are both significantly and negatively correlated with synchronicity, which is consistent with the finding that firm-level transparency facilitates the incorporation of firm-specific information into stock prices (Ferreira & Laux, 2007). INSERT TABLE 2 ABOUT HERE 4.2 The trend in private information acquisition and synchronicity Over our sample period, the trends in analyst private information and synchronicity have been reported in Panel B of Table 2. Synchronicity (SYNCH1) increases from -0.40 preregulation to -0.04 post-regulation, and the same trend is shown for SYNCH1a. In contrast, there is a significant decreasing trend in analyst private information as evidenced by the tstatistic of -10.92. To provide direct evidence on the association between private information acquisition and synchronicity, we compare the mean synchronicity at different levels of private information: below 20%, 20~30%, 30~40%, 40~50%, 50~60%, 60~70% and over 70%. 27 Figure 1 depicts how synchronicity changes with analyst private information. With the increase in PRIVATE, SYNCH1 decreases. This relation is most obvious when PRIVATE is beyond 20-30% of its sample value. Thus, this evidence lends support to H1. INSERT FIGURE 1 ABOUT HERE 4.3 The effect of private information acquisition on synchronicity pre and post-Directive 40 To test the effect of private information acquisition on synchronicity in the pre and post-regulation periods, we estimate the following regression. SYNCH 0 1TIMEi ,t 2 PRIVATE i ,t 3 REGi ,t * PRIVATE i ,t k k CONTROLi ,t i ,t ....(2) Where, for firm i and year t, TIME is the difference between the fiscal year and 2003 used to control for the trend in synchronicity. This approach is similar to Cohen et al. (2008), which investigates the effect of SOX on real and accrual earnings management controlling for the possible trends in earnings management activities. We have identified a discernible trend in SYNCH as shown in Panel B of Table 2. Thus, we expect the coefficient on TIME to be positive. PRIVATE represents analyst private information, and we expect its coefficient to be negative. REG is a dummy variable which is 1 for fiscal years over 2007-2012, and 0 otherwise. So, REG is the specific event based on the year. We don’t expect REG to affect synchronicity directly because Directive 40 is designed to restrict private information seeking, which then affects the firm-specific information incorporated into stock price. So, REG’s effect is indirect via its moderation on PRIVATE. In all the regression specifications, we employed cluster analysis in the panel data setting to control for cross-sectional variance that is not explained by the independent variables in our models (Peterson, 2009). 28 Following previous related research (Chan & Hameed, 2006; Gul et al., 2010; Piotroski & Roulstone, 2004) and referring to Chinese unique institutional features, we include a total of 16 control variables that may affect synchronicity. They are: annual trading volume turnover (VOL), firm size (SIZE), leverage (LEV), market-to-book ratio (MB), earnings volatility (STDROA), the number of firms in the industry to which a firm belongs (INDNUM), and industry size (INDSIZE). In addition, we include analyst forecast common information content (COMMON) and forecast consensus (CONSENSUS) in regressions to address the argument that both analyst’s private and common information contents benefit a firm’s information environment. State ownership (STATE) and whether a firm is controlled by central (CENGOV) or local government (LOCGOV) are controlled for, owing to the influence of ownership structures on synchronicity. We also use two proxies, preannouncements of earnings (PREANCE) and management earnings forecast (MEF) to control for the effect of public disclosures on synchronicity, because increasing trends in voluntary disclosure of other value-relevant information – for example, via conference calls or the release of pro-forma earnings – increase the market reaction to firms’ earnings (Collins et al., 2009; Francis et al., 2002). Year dummies are not included as TIME and REG are all measured based on year, and the trend in synchronicity is expected to be captured by the coefficient on TIME. Industry dummies are included to control for potential industry fixed effects. The appendix provides the definitions of all variables included in the regressions. INSERT TABLE 3 ABOUT HERE Table 3 presents multivariate regression results for equation (2) and equation (3). In this section, we firstly discuss the analysis results using equation (2), which examines the effect of private information acquisition on synchronicity pre and post-Directive 40. Equation (3) results are discussed in section 4.4. Regarding equation (2) results, the coefficients on 29 PRIVATE are negative and significant for both regressions using SYNCH1 and SYNCH1a as dependent variables (coefficients are -0.0001, t-statistics are -4.17, and -3.59 respectively, both having p < 0.00). This negative coefficient suggests that analyst private information acquisition serves as an important mechanism to convey firm-specific information to investors and, thus, reduces synchronicity before 2007. The moderating effects of Directive 40 are significantly positive for both estimations using SYNCH1 and SYNCH1a as dependent variables (coefficients are 0.0002, t-statistic are 5.84, and 6.63 respectively, both having p < 0.00). The results suggest that regulatory restriction on selective disclosures results in an underproduction of analyst private information thus reduces firm-specific information content in stock price, observed as a greater synchronicity post-regulation. Thus, the finding lends support to our H2b. As more than 90% of the total sample observations are from A-share only firms, and we conclude that the above findings are driven mainly by the A-share market. We also estimate equation (2) using SYNCH1b and SYNCH1c as dependent variables. However, untabulated analysis results for these two sub-samples do not support the hypotheses. Due to the limited observations in these two sub-samples (360 and 358 firm-year observations respectively), we are cautious about interpreting the results. With respect to the control variables, we find that the trend variable (TIME) is positive (coefficient 0.04, t statistic 5.39, p < 0.00) when SYNCH1 is used as the dependent variable, which is in accordance with the univariate analysis presented in Table 2 showing an increasing trend in synchronicity with time. This effect is more pronounced for all shares. Analyst common information content (COMMON) is associated with synchronicity negatively (coefficient -0.0001, t statistics -1.89, p < 0.05), whereas their consensus forecasts (CONSENSUS) is related to synchronicity positively (coefficient 0.09, and t statistics 4.61, p 30 < 0.00). Thus, the findings suggest that the common information content in analyst forecasts also facilitates firm-specific information to be incorporated into the share price, but consensus in forecasts increases the co-movement of a firm’s share price. The coefficient for VOL is significantly negative (coefficient -0.01, t statistic 5.73, p < 0.00) for A-share only firms. This suggests that active trading enhances the incorporation of firm-specific information into stock prices. Firm size (SIZE) is positively associated with synchronicity (coefficients 0.11 and 0.07, t statistics 11.47 and 6.65, p < 0.00), which is consistent with Piotroski & Roulston (2004) and Gul et al. (2010). It is argued that large Chinese firms tend to mirror the market to a greater extent than small firms do. In addition, large firms constitute a major proportion of firms included in the market and industry indices, so their performance affects market and industry indices returns to a greater extent (Gul et al., 2010). The coefficients of MB are significantly negative, implying greater firm-specific information content in the share price for high growth firms in that those firms may have more research and development, and major capital investment information being incorporated into their share prices. Both STDROA and INDNUM are significantly and negatively related to synchronicity, suggesting that firms with larger earnings volatility participating in the industry with more operators are more likely to have informative stock prices. Interestingly, STATE shows a negative association with synchronicity. In contrast, LOCGOV has positive association with synchronicity which is consistent with Gul et al. (2010) in that they also find synchronicity is higher when the largest shareholder is government-related in China. Our finding highlights that local government is the driving force of this synchronicity-increasing effect of government shareholding. Lastly, for those firms issuing only A shares (SYNCH1a as the dependent variable), it seems their public disclosure does not play a role in reducing stock price synchronicity. Instead, the pre-earnings announcement (PREANCE) increases the stock price co-movement 31 (coefficient 0.06, t statistic 3.28, p < 0.00). As expected, management earnings forecasts (MEF) reduces the level of share co-movement significantly (coefficient -0.07, t statistic 5.39, p < 0.00), suggesting that managers communicate firm-specific information via their earnings forecasts. Analyst forecast errors (ABSFE) have a significantly positive effect on synchronicity, which is consistent with the belief that owing to the analyst efforts in information seeking and research, the more precise an analyst forecast is, the more information about the fundamental value of the firm will be incorporated into the price (Lee & Liu, 2011). The signs on these control variables are highly consistent with prior literature, and the results, by and large, resemble analyses using SYNCH1 and SYNCH1a.. 4.4 The moderating effect of affiliated versus non-affiliated analysts We also test whether the above relationships are moderated by a particular type of analyst, affiliated analysts, owing to their close connection with firms, and the direct effect of regulation on them. Firstly, we compare analyst private information content between affiliated and non-affiliated analysts from the pre- to post-regulation eras in order to understand the changes. Table 4 Panel A demonstrates these comparisons. INSERT TABLE 4 ABOUT HERE Based on a full sample analysis, the affiliated analysts’ forecasts have more private information content (129.21) than the non-affiliated ones (94.99), although this difference is not statistically significant over the sample period of 2003-2012. Non-affiliated analysts’ private information reduces significantly from the pre-regulation period (247.00) to the postregulation period (63.64) (t-statistic -10.94, p < 0.00). In contrast, although the affiliated analysts’ private information content also declines post-regulation, the decline is not statistically significant (t-statistic -0.58). As a result, there is a discernible difference in 32 forecasts’ private information content between affiliated (122.61) and non-affiliated analysts (63.64) post-regulation (t-statistic 1.29, p < 0.10). In comparison, the PRIVATE of the two types of analyst does not differ in the pre-regulation period (264.54 for affiliated analysts versus 247.00 for non-affiliated analysts). The findings suggest that post-Directive 40, the private information gap between affiliated and non-affiliated analysts has increased due to the decrease in non-affiliated analysts’ private information. Thus, the results imply that Directive 40 may have forced firms to reduce the selective disclosures to general analysts but not to their affiliated analysts. Post-regulation, affiliated analysts still have private information acquisition channels, even if there is restriction on the selective disclosure of material information from management to certain market participants. Recent Chinese studies corroborate this inference (Yang, 2010). If analysts have less private information post-regulation, their forecast accuracy will reduce, resulting in larger forecast errors. To draw a more definite inference, we also compare the forecast errors between two types of analysts over pre- and post-regulation periods. The results are reported in Panel B of Table 4. We find that affiliated analysts have lower forecast errors (0.41) than those of non-affiliated ones (2.27) over our sample period. However, the difference in forecast errors between the two types of analyst is not significant in the pre-regulation period (t-statistics -0.77). Post-regulation, although both types of analysts experience larger forecast errors than before, the increase in forecast errors is more prominent for non-affiliated analysts (t-statistic 8.09, p < 0.00). As a result, post-regulation there is a smaller forecast error for affiliated analysts than for non-affiliated ones (2.02 versus 2.58, t-statistic -2.48, p < 0.00). This result is in accordance with the Bartholdy and Feng (2013) finding that affiliated securities firms issue more accurate earnings forecasts than non‐affiliated firms over the period of 2002-2009 in China. In conjunction with the evidence reported in Panel A of Table 4, we infer that affiliated analysts have more accurate forecasts 33 especially in the post-regulatory period possibly due to their continuous access to superior firm-specific information that may not be available to the public. Then, we argue that since affiliated analysts have more private information than their non-affiliated counterparts, the stock price of firms followed by affiliated analysts should have more firm-specific information content than the firms followed by non-affiliated analysts post-regulation. Thus, we expect a negative effect of affiliated analysts on synchronicity post-regulation resulting from their superior access to private information and greater PRIVATE as conjectured in H3. The following equation, as an expansion of equation (2) is used to test this proposition. SYNCH 0 1TIMEi ,t 2 PRIVATE i ,t 3 REGi ,t * PRIVATE i ,t 4 AFFLATE i ,t * REGi ,t * PRIVATE i ,t CONTROL k k i ,t i ,t .................................................................................(3) Where, AFFLATE is a binary variable with a value of 1 if at least one of the firm’s analysts is affiliated. Equation (3) results are reported in Table 3. All other variables are defined as before. Our variable of interest is the three-way interaction, AFFLATE*REG*PRIVATE. The coefficient on it is negative as expected for estimations using both SYNCH1 and SYNCH1a. Although this decremental effect is only marginally significant for SYNCH1 analysis (t statistic -1.64, p < 0.10), it is more prominent for SYNCH1a analysis (t statistic 3.46, p < 0. 00). This implies that affiliated analysts play more important information roles for firms issuing only A shares. We also estimate equation (3) for SYNCH1b, 1c. No definite inferences can be drawn from analysis.12 All other variables have shown largely consistent 12 B and H share markets have features distinct from the A-share market. With respect to the financial investment industry, analysts of the A-share market are fledglings, whereas analysts of B- and H-share markets are more sophisticated and have more incentives to compete and provide quality services to investors, especially foreign investors in these two markets (Barniv, 2009). In addition, information asymmetry in the A-share market is significantly greater than it is in B- and H-share markets, in that A-shares tend to have low levels of public disclosures, inferior financial reporting quality and poor corporate governance (Piotroski & Wang, 2012). 34 results with equation (2) analysis. Thus, our findings support the conjecture that affiliated analysts’ private information acquisition tends to lower firm’s stock price synchronicity postregulation. H3 is supported. It is noted that Bartholdy and Feng (2013) fail to find the market differential reaction to the stock recommendations made by affiliated and non-affiliated analysts over the period of 2002-2009, although affiliated analysts show lower forecast errors than their non-affiliated counterparts. Our finding of the significantly negative coefficient on AFFLATE*REG*PRIVATE suggests that earnings forecasts made by affiliated analysts accelerate the incorporation of firm-specific information into stock prices perhaps due to their continuously access to private information from management post-regulation. 4.4 The moderating effect of financial market development on the association between analyst private information acquisition and synchronicity To test H4, we compare the amount of analyst private information content (PRIVATE) between provinces with high and low financial market development using univariate analysis, as reported in Table 5. INSERT TABLE 5 ABOUT HERE The simple t statistic shows the significant higher (lower) analyst private information content in their forecasts in the provinces with a high (low) level of financial market development (t statistic 2.88, p < 0. 00). However, the difference in SYNCH1 is not statistically significant, although synchronicity is slightly lower in provinces with a high level of financial market development than it is in the provinces with a low level of financial development. Other factors influencing synchronicity need to be controlled in regression analysis in order to draw a definite inference. We thus conduct the following test reported in Table 6. Table 6 shows that during our full sample period, the effect of PRIVATE on 35 synchronicity is not significant. However, once the financial market development is taken into account, the coefficient on the interactive term FINDEV*PRIVATE is negative and significant across all four model specifications (t statistics range from -1.76 to -4.15). It suggests that for firms in provinces with a better developed financial markets (FINDEV takes value of 1), the analyst private information acquisition is an important factor lowering stock price synchronicity. Thus, the information usefulness of analyst private information acquisition seems strongest in provinces with relatively developed financial markets. Furthermore, although we have not developed a formal hypothesis, when we add the threeway interaction, REG*FINDEV*PRIVATE into our model to test the regulation effect, the synchronicity-reducing effect of analyst private information acquisition is mitigated (coefficients 0.0002, t statistics 5.04 and 5.16 respectively, p < 0.00). This result suggests that post-Directive 40, the firm-specific information content of analyst private information acquisition is reduced (synchronicity is increased) for those firms operating in the provinces with better developed financial markets. INSERT TABLE 6 ABOUT HERE 4.5 Sensitivity test To test the robustness of our results, several sensitivity tests are conducted. The financial crisis in 2008 may affect our expected associations, so we re-estimate our regressions controlling for a dummy variable, CRISIS, which is 1 for observations in 2008, and 0 otherwise. The results are reported in Table 7, which shows that synchronicity is greater during the crisis period in 2008 with a significant coefficient on CRISIS for all four specifications. In addition, as expected, the effect of private information acquisition on synchronicity is moderated positively by regulation, showing a significant coefficient on REG*PRIVATE, and the negative moderating effect of affiliated analysts holds continuously, 36 as evidenced by the negative coefficient on AFFLATE*REG*PRIVATE. However, the effect of private information on synchronicity before 2007 is not statistically significant, although the sign is negative as expected (t statistics range from -0.54 to -0.86). Alternatively, we reestimate equations (2) and (3) after excluding observations from the 2008 fiscal year. The results on our main variables of interest are similar to those reported in Table 3. Other sensitivity analyses include (1) an alternative cut-off point to define affiliated analysts used for H3 testing. AFFLATE is 1 if more than half of the analysts following the same firms are affiliated and, 0 otherwise. The results for H3 testing using equation (3) are qualitatively unchanged; (2) we use an alternative measurement of management forecasts (MEF2) for regression estimations. The results are largely consistent with the results reported in Table 3. INSERT TABLE 7 ABOUT HERE 5. Summary and concluding remarks Financial analysts are the main information intermediaries serving to disseminate information and reduce information asymmetry between firms and investors. Studying the impact of analyst activities on stock prices and returns improves our understanding of the economic utility of analyst activities. Extant studies on the association between analyst coverage and synchronicity usually employ analyst following as a proxy for analyst effort and have produced mixed evidence. Our paper advances this topic by explicitly using analyst private information acquisition to proxy for analyst effort in private information seeking. We argue that analysts’ reliance on public or private information for analysis and forecasts may have distinct effects on the capitalization of firm-specific information into stock prices. Thus, using analyst following to measure the effect on synchronicity is a crude approach to drawing 37 inferences on analysts’ information advantage, and the economic consequences of analyst information acquisition. To the best of our knowledge, our paper is the first to link the nature of analyst information to firm’s stock price synchronicity directly, which is a major contribution of our study. In addition, we argue that analyst private information-seeking is exceptionally important in Chinese security markets because Chinese listed firms generally have a low level of, and poor quality public disclosures, resulting in high information asymmetry between firms and market participants. Analyst private information acquisition in this context is an important supplement to public information in overcoming information asymmetry and conveying firm-specific information to the wider investment community. We document a negative correlation between synchronicity and analyst private information acquisition, and both move in opposite directions during our sample period. Regarding the restriction on selective disclosures promulgated as Directive 40, we unveil that the CSRC Directive 40 promulgated in 2007 has reduced analyst private information content in their forecasts significantly, resulting in an increased synchronicity in the post-regulation era from 2007 to 2012. However, this reduction in private information acquisition is more pronounced for non-affiliated than affiliated analysts, indicating an ongoing availability of other private communications between firms and their affiliated analysts. Furthermore, we find that the above relations are modified by the extent of the financial market development in the province where a firm operates. Overall, although our finding supports the effectiveness of Directive 40 in its objective of ‘levelling the information playing field’ as is evidenced by the reduced analyst private information content in their forecasts, we also find a lack of efficiency of Directive 40 from a market perspective. That is, the synchronicity-reducing effect of analyst private information acquisition is ameliorated post-Directive 40. Overall, our findings provide new insights into the economic consequences of regulatory changes made to 38 curb insider trading and selective disclosures in an emerging market where public information is less than optimal and the analyst industry is in its infancy. 39 Table 1 Sample selection and industry distribution Panel A: Sample selection Year Listed companies in China Less: Financial companies Less: ST companies Less: Delisted companies Less: PRIVATE Total Sub-sample 1 Less: Trading days less than 200 days Total Sub-sample 2 Less: Control variables Total Sample 2003 1268 9 106 60 866 227 2004 1356 9 111 51 837 348 2005 1352 9 101 36 692 514 2006 1435 15 127 29 599 665 2007 1549 27 160 18 478 866 2008 1603 27 146 18 387 1025 2009 1694 30 149 10 299 1206 2010 1919 36 165 10 229 1479 2011 2048 40 147 4 200 1657 2012 2115 41 104 5 321 1644 Total 19745 263 1518 409 7869 9686 43 59 18 137 135 90 103 227 173 101 1710 184 48 136 289 13 276 496 29 467 528 34 494 731 45 686 935 48 887 1103 60 1043 1252 67 1185 1484 66 1418 1543 63 1480 8597 525 8072 Panel B: Industry distribution of the sample observations CSRC Industry Classification Firm-year Obs. Percentage Distribution 1-A Farming, Forestry, Animal Husbandry and Fishery 152 1.88% 2-B Mining and Quarrying 256 3.17% 3-C0 Food and Beverage 348 4.31% 4-C1 Textile, Clothing, Fur 268 3.32% 41 0.51% 6-C3 Papermaking, Printing 148 1.83% 7-C4 Petroleum, Chemical, Rubber, Plastic 830 10.28% 8-C5 Electronic 415 5.14% 9-C6 Metal, Nonmetal 786 9.74% 1403 17.38% 519 6.43% 84 1.04% 13-D Production & Supply Of Power, Gas & Water 334 4.14% 14-E Construction 186 2.30% 15-F Transportation, Storage 388 4.81% 16-G Information Technology Industry 520 6.44% 17-H Wholesale and Retail Trades 442 5.48% 19-J Real Estate 420 5.20% 20-K Social Services 267 3.31% 68 0.84% 5-C2 Timber, Furniture Industry 10-C7 Machinery, Equipment, Instrument 11-C8 Medicine, Biologic Products 12-C9 Other manufacturing 21-L Transmitting, Culture Industry 22-M Integrated Total 197 2.44% 8072 100.00% 40 Table 2 Descriptive statistics Panel A: Descriptive statistics Variables Analyst private PRIVATE information COMMON acquisition D/∣actual EPS∣ SE/∣actual EPS∣ N CONSENSUS Stock price SYNCH1 synchronicity R²1 (SYNCH) and SYNCH1a R² R²1a SYNCH1b R²1b SYNCH1c R²1c Other variables FE INSDEV AFELATE VOL SIZE STDROA MB INDNUM INDSIZE STATE LOCGOV CENGOV MEF PREANCE Mean 95.17 57.47 0.41 1.01 41.39 0.53 -0.10 0.47 -0.34 .42 -0.31 0.42 -0.15 0.46 -2.17 54.45 0.12 5.15 22.00 0.02 3.54 4.68 27.19 0.34 .38 0.18 0.57 0.30 Median 4.43 4.42 0.05 0.07 20 0.59 -0.11 0.47 -0.33 0.41 -0.32 0.41 -0.12 0.46 -0.40 54.25 0 4.23 21.84 0.02 3.00 4.68 27.29 0.33 0 0 1 0 Panel B: Difference in main variables’ mean values pre and post-regulation Variable Pre-regulation Post-regulation PRIVATE 247.00 64.06 COMMON 112.99 46.09 CONSENSUS 0.46 0.54 SYNCH1 -0.40 -0.04 R²1 0.41 0.49 SYNCH1a -0.57 -0.30 R²1a 0.37 0.43 FE -0.97 -2.41 ABSFE 1.07 2.51 INSDEV 48.57 55.66 AFFLATE 0.07 0.13 VOL 3.84 5.42 SIZE 21.86 22.03 STDROA 0.023 0.028 MB 0.002 0.003 INDNUM 102.32 143.20 INDSIZE 26.34 27.37 STATE 0.37 0.34 LOCGOV 0.48 0.35 CENGOV 0.21 0.17 Std. Dev. 363.48 246.89 1.26 3.87 54.39 0.33 0.56 0.13 0.58 .13 0.56 0.13 0.64 0.14 5.95 14.04 0.32 3.64 1.19 0.02 2.69 0.70 1.06 0.17 0.48 0.38 0.49 0.46 Differences (t-statistics) -10.92*** -6.87*** 6.96*** 23.47*** 23.58*** 15.64*** 16.35*** -11.96*** 11.96*** 20.56*** 8.88*** 19.64*** 5.44*** 7.68*** 24.98*** 20.35*** 43.50*** -4.97*** -8.51*** -2.86*** 41 Variable Pre-regulation Post-regulation Differences (t-statistics) MEF 0.37 0.61 16.89*** PREANCE 0.09 0.35 27.27*** Note: *** ** * , , represent statistical significance at a probability of <0.01, <0.05, and <0.01 respectively (two-tailed test). 42 Panel C: Correlation matrix SYNCH1 PRIVATE COMMON CONSENSUS REG VOL SYNCH1 1 PRIVATE -.041** 1 COMMON -.030** .217** 1 CONSENSUS .033** -.341** .182** REG .240** -.189** -.102** .088** VOL -.098** .001 .045** .031** .164** SIZE .238 ** -.045 ** ** ** .032 ** -.011 LEV ** -.066 .001 SIZE LEV STDROA INDUM STATE CENGOV LOCGOV PREANCE MEF 1 -.097 .000 1 .055 -.016 1 -.373** 1 -.046** .455** 1 STDROA -.048** -.062** -.018 .081** .074** .063** -.044** -.022* 1 INDUM -.053** -.088** -.065** .088** .165** .010 -.057** -.017 .006 1 STATE -.006 .012 -.021 -.031** -.058** -.072** .210** -.043** .011 -.049** 1 CENGOV .029** .005 .001 -.049** -.033** -.084** .248** .137** .012 .023* .126** LOCGOV .093** .071** .042** -.074** -.097** -.073** .156** .151** -.035** -.114** .167** -.370** 1 PREANCE -.055** -.088** -.103** .081** .215** .096** -.255** -.268** -.039** .116** .086** -.100** -.234** 1 MEF -.112** -.108** -.071** .113** .184** .171** -.197** -.073** .139** .112** .001 -.064** -.196** .338** 1 AFFLATE -.020 -.006 .003 .033** .094** -.017 -.045** .031** .011 .042** -.008 -.040** .061** .019 .007 AFFLATE 1 Notes: *, ** and *** represent correlation coefficient significance at the 0.10, 0.05 and 0.01 levels (two-tailed) respectively. 43 1 The percentage of PRIVATE and their dividing values 0 -0.02 -0.04 <0.4548 1.0695 2.1615 4.4337 9.0903 19.01 >19.01 0-20% 20-30% 30-40% 40-50% 50-60% 60-70% 70100% -0.06 -0.08 SYNCH1 -0.1 -0.12 -0.14 -0.16 Percentage of PRIVATE Dividing values of PRIVATE SYNCH1 0-20% <0.4548 -0.0603 20-30% 30-40% 40-50% 50-60% 60-70% 1.0695 2.1615 4.4337 9.0903 19.01 -0.0496 -0.0811 -0.082 -0.0882 -0.1326 70-100% >19.01 -0.1505 Figure 1. The relationship between stock price synchronicity, analyst private information acquisition 44 Table 3 The effect of private information acquisition on synchronicity pre and post-Directive 40 Equation (2) Variables Sign SYNCH1 SYNCH1a Coefficient (t-stat) -0.25 (-0.27) 0.04*** (5.39) -0.0001*** (-4.17) 0.0002*** (5.84) Coefficient (t-stat) -1.02 (-1.11) 0.01 (1.14) -0.0001*** (-3.59) 0.0002*** (6.63) Equation (3) SYNCH1 SYNCH1a Coefficient Coefficient (t-stat) (t-stat) Constant -0.28 -1.10 (-0.31) (-1.20) TIME + 0.04*** 0.01 (5.37) (1.11) PRIVATE - (H1) -0.0001*** -0.0001*** (-4.17) (-3.59) REG*PRIVATE + (H2b) 0.0002*** 0.0002*** (6.01) (7.32) AFFLATE*REG*PRIVATE - (H3) -0.0001* -0.001*** (-1.64) (-3.46) COMMON -0.0001** -0.00004 -0.0001** -0.00004 (-1.89) (-1.55) (-1.89) (-1.55) CONSENSUS + 0.09*** 0.15*** 0.09*** 0.15*** (4.61) (6.45) (4.60) (6.41) VOL -0.0005 -0.01*** -0.0005 -0.01*** (-0.27) (-5.73) (-0.24) (5.68) SIZE + 0.11*** 0.07*** 0.11*** 0.07*** (11.47) (6.65) (11.52) (6.80) LEV + -0.16*** -0.04 -0.16*** -0.04 (-2.79) (-0.57) (-2.82) (-0.65) MB -0.03*** -0.04*** -0.03*** -0.04*** (-6.73) (-6.89) (-6.73) (-6.89) STDROA -0.68*** -0.69** -0.68*** -0.69** (-2.54) (-2.38) (-2.56) (-2.42) INDNUM -0.002*** -0.002*** -0.002*** -0.002*** (-5.91) (-5.37) (-5.93) (-5.45) INDSIZE -0.09** -0.03 -0.09** -0.03 (-2.37) (-0.78) (-2.34) (-0.73) STATE ? -0.30*** -0.38*** -0.30*** -0.38*** (-6.58) (-8.10) (-6.53) (-8.00) CENGOV ? 0.04* 0.01 0.04* 0.01 (1.66) (0.32) (1.65) (0.27) LOCGOV ? 0.09*** 0.08*** 0.09*** 0.08*** (4.91) (3.96) (4.86) (3.86) PREANCE 0.02 0.06*** 0.02 0.06*** (1.39) (3.28) (1.43) (3.40) MEF -0.07*** -0.01 -0.07*** -0.01 (-5.39) (-0.97) (-5.40) (-1.03) ABSFE + 0.003*** 0.004*** 0.003*** 0.004*** (3.09) (4.01) (3.09) (4.02) INSDEV -0.0003 -0.00001 -0.0003 0.00004 (-0.58) (-0.02) (-0.55) (0.07) Ind. Dummies Included Included Included Included N 8072 7354 8072 7354 Adj. R2 0.13*** 0.14*** 0.13*** 0.14*** Note: ***, **, and * represent statistical significance at the 1%, 5% and 10% levels respectively (two-tailed test). See appendix for variable definitions. SYNCH1 is measured for all A-shares including A-share firms, AB-share firms, AH-share firms; SYNCH1a is for only A-share firms; SYNCH1b is for AB-share firms; SYNCH1c is for AH-share firms. All variables are defined in the Appendix. 45 Table 4 The comparison in analyst private information and forecast errors between affiliated and non-affiliated analysts and between pre- and post-regulation Panel A: Analyst private information PRIVATE Full sample periods Pre Post Affiliated analysts 129.21 264.54 122.61 Difference between preand post-regulation periods (t-statistics) -0.58 Non-affiliated analysts 94.99 247.00 63.64 -10.94*** Difference between affiliated and non-affiliated analysts 0.67 0.19 1.29* N/A Full sample periods Pre Post Affiliated analysts 0.41 0.79 2.02 Difference between preand post-regulation periods (t-statistics) 2.13** Non-affiliated analysts 2.27 1.08 2.58 8.09*** Difference between affiliated and non-affiliated analysts -2.06** -0.77 -2.48*** N/A Panel B: Forecast error. ABSFE Note: ***, **, and * represent statistical significance at the 1%, 5% and 10% levels respectively (two-tailed test). 46 Table 5 Comparison in mean value of PRIVATE and SYNCH1 for firms between high and low FINDEV provinces Provinces with high FINDEV Provinces with low FINDEV Difference in PRIVATE (SYNCH1) between high and low FINDEV provinces (t-statistic) PRIVATE 108.16 84.48 2.88** SYNCH1 -0.10 -0.09 -0.46 Note: ***, **, and * represent statistical significance at the 1%, 5% and 10% levels respectively (two-tailed test); FINDEV is defined in Appendix. It is a dummy variable presenting high or low levels of financial market development. It is 1 if a firm is located in a province where the financial market development index is above the sample mean value; 0 otherwise. 47 Table 6 The effect of private information acquisition on synchronicity conditional on financial market development Without regulation effect Variables Sign SYNCH1 SYNCH1a Coefficient (t-stat) 0.53 (0.75) 0.05*** (7.50) 0.0001 (0.72) -0.00006** (-2.08) Coefficient (t-stat) -0.40 (-0.53) 0.01*** (2.54) 0.0001 (1.60) -0.00005* (-1.76) With regulation effect SYNCH1 SYNCH1a Coefficient Coefficient (t-stat) (t-stat) Constant 0.53 -0.41 (0.75) (-0.54) TIME + 0.04*** 0.01** (7.10) (2.17) PRIVATE 0.0002 0.0001* (0.81) (1.67) FINDEV*PRIVATE - (H4) -0.0002*** -0.0002*** (-4.11) (-4.15) REG* FINDEV*PRIVATE + 0.0002*** 0.0002*** (5.05) (5.16) COMMON -0.00004* -0.00003 -0.0001** -0.00004 (-1.60) (-1.24) (-1.93) (-1.56) CONSENSUS + 0.09*** 0.15*** 0.09*** 0.15*** (4.39) (6.33) (4.57) (6.50) VOL 0.001 -0.01*** 0.0002 -0.01*** (0.59) (-5.22) (0.13) (5.67) SIZE + 0.11*** 0.07*** 0.11*** 0.07*** (15.27) (7.83) (15.25) (7.86) LEV + -0.15*** -0.03 -0.15*** -0.03 (-3.06) (-0.56) (-3.04) (-0.56) MB -0.03*** -0.04*** -0.03*** -0.04*** (-6.81) (-7.06) (-6.82) (-7.05) STDROA -0.64*** -0.67*** -0.64*** -0.66** (-2.81) (-2.58) (-2.78) (-2.56) INDNUM -0.46*** -0.47*** -0.45*** -0.47*** (-6.03) (-5.92) (-5.93) (-5.81) INDSIZE -0.06* 0.01 -0.06* 0.01 (-1.83) (0.28) (-1.84) (0.28) STATE ? -0.30* -0.38*** -0.30*** -0.38*** (-8.33) (-9.60) (-8.36) (-9.66) CENGOV ? 0.05** 0.01 0.04** 0.01 (2.42) (0.55) (2.33) (0.45) LOCGOV ? 0.09*** 0.08*** 0.09*** 0.08*** (6.48) (4.95) (6.33) (4.81) PREANCE 0.02 0.05*** 0.02 0.06*** (1.53) (3.63) (1.48) (3.60) MEF -0.07*** -0.01 -0.07*** -0.02 (-5.51) (-1.06) (-5.53) (-1.09) ABSFE + 0.003*** 0.004*** 0.003*** 0.004*** (2.72) (3.64) (2.78) (3.71) Ind. Dummies Included Included Included Included N 8072 7354 8072 7354 Adj. R2 0.13*** 0.14*** 0.13*** 0.14*** Note: ***, **, and * represent statistical significance at the 1%, 5% and 10% levels respectively (two-tailed test). See appendix for variable definitions. SYNCH1 is measured for all A-shares including A-share firms, AB-share firms, AH-share firms; SYNCH1a is for only A-share firms; SYNCH1b is for AB-share firms; SYNCH1c is for AH-share firms. All variables are defined in the Appendix. 48 Table 7 Sensitivity test for the effect of private information acquisition on synchronicity pre and post-Directive 40 Expanded Equation (2) Variables Sign SYNCH1 SYNCH1a Coefficient (t-stat) -1.10 (-1.35) 0.04*** (6.66) 0.83*** (57.49) -0.0002 (-0.86) 0.0001*** (2.57) Coefficient (t-stat) -2.13*** (-2.58) 0.01* (1.82) 0.80*** (42.69) -0.0001 (-0.54) 0.0001*** (3.53) Expanded Equation (3) SYNCH1 SYNCH1a Coefficient Coefficient (t-stat) (t-stat) Constant -1.15 -2.22*** (-1.41) (-2.68) TIME + 0.05*** 0.01* (6.63) (1.79) CRISIS + 0.84*** 0.81*** (57.54) (42.97) PRIVATE - (H1) -0.0001 -0.0001 (-0.86) (-0.53) REG*PRIVATE + (H2) 0.0001*** 0.0001*** (2.89) (4.38) AFFLATE*REG*PRIVATE - (H3) -0.0002** -0.0004*** (-2.37) (-3.96) COMMON 0.0001 -0.00007 0.0001 -0.00004 (0.07) (0.31) (0.08) (-1.55) CONSENSUS + 0.03 0.08*** 0.03 0.08*** (1.46) (4.01) (1.43) (3.96) VOL 0.01*** -0.005** 0.01*** -0.004** (3.34) (-2.33) (3.38) (-2.25) SIZE + 0.13*** 0.09*** 0.13*** 0.09*** (14.39) (9.10) (14.47) (9.27) LEV + -0.25*** -0.13** -0.26*** -0.14** (-5.41) (-2.26) (-5.46) (-2.36) MB -0.02*** -0.03*** -0.02*** -0.03*** (-5.71) (-6.61) (-5.71) (-6.61) STDROA -0.95*** -0.98** -0.96*** -0.99*** (-3.98) (-3.83) (-4.01) (-3.88) INDNUM ? -0.0004 -0.0004 -0.0004 -0.0004 (-1.53) (-1.37) (-1.56) (-1.45) INDSIZE -0.08** -0.01 -0.08** -0.01 (-2.40) (-0.32) (-2.37) (-0.26) STATE ? -0.23*** -0.31*** -0.23*** -0.30*** (-5.50) (-6.95) (-5.43) (-6.83) CENGOV ? 0.02 -0.01 0.02 -0.01 (0.99) (-0.30) (0.97) (-0.37) LOCGOV ? 0.09*** 0.07*** 0.09*** 0.07*** (4.98) (3.79) (4.92) (3. 67) PREANCE 0.01 0.05*** 0.02 0.05*** (0.94) (3.05) (0.99) (3.20) MEF -0.08*** -0.02* -0.08*** -0.02* (-6.28) (-1.66) (-6.30) (-1.73) ABSFE + 0.003*** 0.004*** 0.003*** 0.004*** (3.17) (4.15) (3.18) (4.16) INSDEV 0.00002 0.0003 -0.00004 0.0003 (0.04) (0.49) (-0.08) (0.59) Ind. Dummies Included Included Included Included N 8072 7354 8072 7354 Adj. R2 0.33*** 0.31*** 0.33*** 0.31*** Note: ***, **, and * represent statistical significance at the 1%, 5% and 10% levels respectively (two-tailed test). See appendix for variable definitions. 49 SYNCH1 is measured for all A-shares including A-share firms, AB-share firms, AH-share firms; SYNCH1a is for only A-share firms; SYNCH1b is for AB-share firms; SYNCH1c is for AH-share firms. FINDEV is defined in Appendix. It is a dummy variable presenting high or low levels of financial market development. It is 1 if a firm is located in a province where the financial market development index is above the sample mean value; 0 otherwise. All variables are defined in the Appendix. Appendix Variable definitions Variables PRIVATE Definition Analyst private information acquisition (precision) measured as D PRIVATE [(1 COMMON 1 ) D SE ]2 N Where N is the number of analysts’ forecasts. D is the sample variance of the analysts’ individual forecasts. SE is the squared error in the mean forecast, calculated as the squared difference between the mean forecast and actual earnings, i.e., (Actual EPS – Mean EPS Forecast)². Analyst common information acquisition (precision) measured as ( SE COMMON D ) N 1 ) D SE ]2 N CONSENSUS Analysts’ consensus information [(1 COMMON COMMON PRIVATE Stock price synchronicity, measured using the three regressions. CONSENSUS SYNCH RETit 1MKTRETt 2 MKTRETt 1 3 INDRETt 4 INDRETt 1 it (1) Where, for firm i and t, RET presents the daily returns on A-shares traded on either the Shanghai or Shenzhen exchange; MKTRET and INDRET are the valueweighted A-share market returns and industry returns, respectively. The valueweighted A-share market return equals the change in the composite value-weighted A-share indices from day t to day t – 1 deflated by the composite value-weighted A-share index on day t - 1. The industry return is created using all firms within the same industry with firm i’s daily return omitted. Industry weighted daily returns are calculated using the weighting of each firm’s market capitalization as a percentage of all firms’ total market capitalization in a particular industry. Lagged MKTRET and INDRET are also included. 50 To address the possibility that Chinese stock returns may be correlated with world market returns, in the following three models, we also control for world market factors as measured by the MSCI (Morgan Stanley Capital International) World index.13 In addition, to tackle Chinese segmental markets, we use Equations 1a, 1b and 1c to estimate the share price co-movement for A shares-only firms, A+B share firms and A+H share firms respectively, because returns on stocks of A+B (A+H) share firms are likely to co-move with B-share (H-share) market factors in addition to A-share market factors in B (H) share markets. This approach follows (Yu et al., 2013) and (Gul et al., 2010). RETit 1MKTRETt 2WORDRET it (1a) RETit 1MKTRETt 2 MKTRET B 3WORDRET it (1b) RETit 1MKTRETt 2 MKTRET H 3WORDRET it (1c) In estimating Equation 1 and Equations 1a, 1b and 1c, we require that daily returns data be available for at least 200 trading days in each fiscal year. Stock price synchronicity is defined as the ratio of common return variation to total return variation, which is equivalent to R2 of the market model used. To circumvent the bounded nature of R2 within [0, 1], we use a logistic transformation of R²: 2 R SYNCH log( i 2 ) 1 Ri Where, SYNCHi is annual synchronicity for firm i. FE REG AFFLATE Control variables VOL The calculation using R² estimated from Equation (1) is denoted as SYNCH (1), while the calculation using R² estimated using Equations 1a, 1b and 1c are represented as SYNCH (1a, 1b, 1c). Forecast error measured as the difference between actual EPS and forecasted EPS, deflated by actual EPS. The Directive 40, measured as dummy variable with a value of 1 for observations from 2008 to 2012, and a value of zero for observations in early sample years. A dummy variable for affiliated analysts. It takes a value of 1 if more than half of the analysts following a firm are affiliated. Affiliated analyst is defined as an analyst from the securities firm acting as investment banker/underwriter services for a firm. In particular, an affiliated analyst issues an earnings forecast on a stock within five years of its IPO date or within two years of its SEO date. Annual trading volume turnover, computed as the total number of shares traded in a year, divided by the total number of shares outstanding at the end of that fiscal 13 The MSCI (Morgan Stanley Capital International) World index is a world market index that is based on the stock prices of listed companies representing of 22 stock markets in North America, Europe, and the Asia/Pacific region, and is weighted by the market capitalization of each constituent market. The index data are extracted from the Datastream database. 51 SIZE LEV STDROA MB INDNUM INDSIZE STATE LOCGOV CENGOV PREANCE MEF INSDEV FINDEV year. Firm size, measured as the log of total assets at the end of a fiscal year. Leverage measured as total liabilities divided by total assets at the end of a fiscal year. Earnings volatility measured as the standard deviation of a firm’s ROAs over the preceding five-year period, including the current year. Market to book ratio computed as the total market capitalization divided by the total net assets at the end of fiscal year. The number of firms in the industry to which a firm belongs. Industry size measured as the log of the year-end total assets of all the sample firms in the industry to which a firm belongs. The percentage of shares held by state owner(s) at year’s beginning. Local government shareholding measured as a dummy variable. It equals 1 if the firm’s largest shareholder is local government-related and 0 if the firm’s largest shareholder is non-local government-related. Central government shareholding measured as dummy variable. It equals 1 if the firm’s largest shareholder is central government related and 0 if the firm’s largest shareholder is non-central government-related. Public pre-announcements of earnings, which is a dummy variable. 1 is assigned to a firm if a pre-announcement of earnings is made and, 0 otherwise. Preannouncement of earnings is not compulsory in China. Management forecasts measured in two ways, MEF1 and MEF2. MEF1 is 1 if a firm’s management has forecasted their earnings, otherwise zero; MEF2 equals 1, if a firm discloses its management forecast, but CSRC didn't require the firm to disclose; it equals -1 if CSRC requires a firm to disclose, but companies didn't disclose the management forecast, 0 otherwise.14 MEF1 is used for the main analysis as tabulated, while MEF2 is used for one sensitivity test. Institutional development index, which is a composite index developed for each province in China by Fan et al. (2011). It includes six components, namely (1) overall market-institutions development; (2) government intervention in markets; (3) private enterprise development; (4) regional protectionism; (5) financial market development; and (6) legal environment. Financial market development measured as a dummy variable representing high or low levels of financial market development based on the fifth component of Fan et al.’s (2011) index. It is 1 if a firm is located in a province where the financial market development index is above the sample mean; and 0 otherwise. 14 CSRC requires companies to disclose management forecasts when: 1) companies turn loss into profit; 2) companies’ net profit is loss; 3) companies’ earnings are fluctuating, that is earnings grow at more than 50% from the previous year, or earnings decline more than 50% from the same period in the previous year. This is the so-called ‘Semi-mandatory Management Forecast policy’. 52 References Asian Securities Analysts Federation, 2004. Yearbook. ATC, Killarney Heights. Agrawal, A., Chadha, S., Chen, M., 2006. Who is afraid of Reg FD? The behavior and performance of sell-side analysts following the SEC’s fair disclosure rules. Journal of Business, 79, 2811. Alves, P., Peasnell, K., Taylor, P. 2009. The use of the R2 as a measure of firm-specific information: a cross country critique. Journal of Business Finance & Accounting, 37 (1-2), 1-26. Arnold, J., Moizer, P. 1984. A survey of the methods used by UK investment analysts to appraise investments in ordinary shares. Accounting and Business Research, 14 (55), 195-207. Ashbaugh-Skaife, H., Gassen, J., LaFond, R. 2005. Does stock price synchronicity represent firm-specific information? The international evidence. Working paper. MITSloan. Ayers, B.C., Freeman, R.N. 2003. Evidence that analyst following and institutional ownership accelerate the pricing of future earnings. Review of Accounting Studies, 8, 47-67. Bae, K., Stulz, R., & Tan, H. (2008). Do local analysts know more? A cross-country study of the performance of local analysts and foreign analysts. Journal of Financial Economics, 88(3), 581–606. Barker, R.G. 1998. The market of information-evidence from finance directors, analysts and fund managers. Accounting and Business Research, 29 (1), 3-20. Barker, R.G. 1999. Survey and market-based evidence of industry-dependence in analysis, preferences between the dividend yield and price-earning valuation models. Journal of Business Finance and Accounting, 26 (3-4), 393-418. 53 Barniv, R., Myring, M.J., Thomas, W.B. 2005. The association between the legal and financial reporting environments and forecast performance of individual analysts. Contemporary Accounting Research, 22 (4), 727-758. Barron, O.E., Byard, D., Kile, C., Riedl, E.J. 2002a. High-technology intangibles and analysts’ forecasts. Journal of Accounting Research, 10 (2), 289-312. Barron, O.E., Byard, D., Kim, O. 2002b. Changes in analysts' information around earnings announcements. The Accounting Review, 77 (4), 821-846. Barron, O.E., Kim, S.C.L., Stevens, D.E. 1998. Using analysts’ forecasts to measure properties of analysts’ information environment. The Accounting Review, 73 (4), 421434. Barron, O.E., Pratt, J., Stice, J.D. 2001. Misstatement direction, litigation risk, and planned audit investment. Journal of Accounting Research, 39 (3), 449-462. Bartholdy, J., Feng, T. 2013. The quality of securities firms’ earnings forecasts and stock recommendations: Do information advantages, reputation and experience matter in China? Pacific-Basin Finance Journal, 24 (C), 66-88. Brown, S., Hillegeist, S., & Lo, K. (2002). Regulation FD and voluntary disclosures. Working paper. Northwestern University. Bushee, B. J., Matsumoto, D. A., & Miller, G. S. (2004). Management forecasts, public conference calls, and pre-announcements of earnings have increased after Reg FD. The Accounting Review, 79(3), 617-643. Bushee, B.J., Jung, M.J., Miller, G.S. 2013. Do Investors Benefit from Selective Access to Management? Working paper. Wharton School of Business. Bushman, R., Piotroski, J., & Smith, A. (2004). What determines corporate transparency. Journal of Accounting Research, 42, 207-252. 54 Byard, D., Shaw, K.W. 2003. Corporate disclosure quality and properties of analysts' information environment. Journal of Accounting, Auditing & Finance, 18 (3), 355378. Chan, K., Hameed, A. 2006. Stock price synchronicity and analyst coverage in emerging markets. Journal of Financial Economics, 80 (1), 115-147. Chang, J.J., Khanna, T., Palepu, K. 2000. Analyst Activity Around the World. Retrieved from: http://ssrn.com/abstract=204570 Charoenrook, A., Lewis, C.M. 2007. Information, Selective Disclosure, and Analyst Behavior. Working Paper, Financial Market Research Center. Chen, K.C.W., Yuan, H. 2004. Earnings management and capital resource allocation: evidence from China's accounting-based regulation of rights issues. The Accounting Review, 79 (3), 645-665. Cohen, D., Dey, A., Lys., T. 2008. Real and accrual-based earnings management in the preand post-Sarbanes-Oxley periods. The Accounting Review, 83 (3), 757-787. Collins, D.W., Li, O.Z., Xie, H. 2009. What drives the increased informativeness of earnings announcements over time? Review of Accounting Studies, 14 (1), 1-30. Cook, D.O., Tang, T. 2010. The impact of Regulation FD on institutional investor informativeness. Financial Management, 39, 1273-1294. Ding, Y., Zhang, H., Zhang, J. 2007. Private vs. state ownership and earnings management: evidence from Chinese listed companies. Corporate Governance: An International Review, 15, 223-238. Dong, B., Li, E.X., Ramesh, K., Shen, M. 2012. The Effects of Regulation FD on Informal and Institutionalized Leakages of Information in Earnings Press Releases. Working Paper. University of Virginia. 55 Li, D., Li, X., & Zhang, B. (2011). China's Analysts Earning Forecast Errors and Home Advantage. Finance and Economics (in Chinese), 2011(3), 21-47. Fan, G., Wang, X. 2001. NERI Index of Marketization of China’s Provinces: 2000 Report (in Chinese). Economic Science Press, Beijing, PRC. Fan, G., Wang, X.L., Zhu, H.P. 2009. NERI index of marketization of China’s provinces (in Chinese). Economics Science Press, Beijing, PRC. Fan, G., Wang, X.L., Zhu, H.P. 2011. NERI index of marketization of China’s provinces (in Chinese). Economics Science Press, Beijing, PRC. Ferreira, M.A., Laux, P.A. 2007. Corporate governance, idiosyncratic risk, and information flow. The Journal of Finance, LXII (2), 951-989. Fisch, J. (2013). Regulation FD: an alternative approach to addressing information asymmetry. In S. Bainbridge (Ed.), Insider Trading: Edward Elgar Publishing. Francis, J., Schipper, K., Vincent, L. 2002. Expanded disclosures and the increased usefulness of earnings announcements. The Accounting Review, 77 (3), 515-546. Frankel, R., Kothari, S.P., Weber, J. 2006. Determinants of the informativeness of analyst research. Journal of Accounting and Economics, 41 (2), 29-54. Frazzini, A., Malloy, C.J., Cohen, L. 2008. Sell Side School Ties. Working Paper. Harvard Business School. French, K., Schwert, G. W., & Stambaugh, R. (1987). Expected stock returns and volatility. Journal of Financial Economics, 19, 3-30. Gintschel, A., Markov, S. 2004. The effectiveness of Regulation FD. Journal of Accounting and Economics, 37 (3), 293-314. Gong, R. 2012. The Impact of the 2007 Reforms on the Information Environment in the Chinese A-share Market. Ph.D. Thesis, University of Auckland, Auckland. 56 Gul, F. A., Kim, J.-B., Qiu, A. 2010. Ownership concentration, foreign shareholding, audit quality, and stock price synchronicity: evidence from China. Journal of Financial Economics, 95 (3), 425-442. Hasan, I., Song, L., Wachtel, P. 2013. Institutional development and stock price synchronicity: evidence from China. doi:http://papers.ssrn.com/sol3/papers.cfm? abstract_id=2314906 Heflin, F., Subramanyam, K. R., & Zhang, Y. (2003). Regulation FD and the financial information environment: early evidence. The Accounting Review, 78 (1), 1-37. Holland, J.B. 2006. Fund management, intellectual capital, intangibles and private disclosure. Managerial Finance, 32 (4), 277-316. Hu, Y.M., Lin, T.W., Li, S. 2008. An examination of factors affecting Chinese financial analysts’ information comprehension, analyzing ability, and job quality. Review of Quantitative Finance and Accounting, 30, 397-417. Jiang, L., Kim, J.-B., Pang, L. 2013. The influence of ownership structure, analyst following and institutional infrastructure on stock price informativeness: international evidence. Accounting and Finance, Forthcoming. Jin, L., Myers, S.C. 2006. R2 around the world: new theory and new tests. Journal of Financial Economics, 79 (2), 257-292. Kim, J.-B., Shi, H. 2008. Enhanced disclosures via IFRS and stock price synchronicity around the world: do analyst following and institutional infrastructure matter? Retrieved from: http://ssrn.com/abstract=1586657 Kim, J.-B., Shi, H. 2012. IFRS reporting, firm-specific information flows, and institutional environment: international evidence. Review of Accounting Studies, 17 (3), 474-517. 57 Kimbro, M.B. 2005. Managing underpricing? The case of pre-IPO discretionary accruals in China. Journal of International Financial Management & Accounting, 16 (3), 229262. Kross, W.J., Suk, I. 2012. Does Regulation FD work? Evidence from analysts' reliance on public disclosure. Journal of Accounting and Economics, 53 (1–2), 225-248. Lang, M.H., Lins, K.V., Miller, D.P. 2004. Concentrated control, analyst following, and valuation: do analysts matter most when investors are protected least? Journal of Accounting Research, 42 (3), 589–623. Lee, D.W., Liu, M.H. 2011. Does more information in stock price lead to greater or smaller idiosyncratic return volatility? Journal of Banking & Finance, 35 (6), 1563-1580. Li, L., Fleisher, B.M. 2004. Heterogeneous expectations and stock prices in segmented markets: application to Chinese firms. The Quarterly Review of Economics and Finance, 44 (4), 521-538. Liu, S., Peng, X. 2012. Fair disclosure and financial analyst forecast accuracy. Securities Market Herald (in Chinese), 2010 (3), 33-38. Morck, R., Yeung, B., Yu, W. 2000. The information content of stock markets: why do emerging markets have synchronous stock price movements? Journal of Financial Economics, 58, 215-260. Palmon, D., & Yezegel, A. (2011). Analysts’ Recommendation Revision. s and Subsequent Earnings Surprises: Pre-and-Post Regulation FD. Journal of Accounting, Auditing & Finance, 26 (3), 475-501. Petersen, M. (2009). Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches. Review of Financial Studies, 22, 435-480. 58 Pike, R., Meerjanssen, J., Chadwick, L. 1993. The appraisal of ordinary shares by investment analysts in the UK and Germany. Accounting and Business Research, 23 (92), 489499. Piotroski, J.D., Roulstone, D. 2004. The influence of analysts, institutional investors, and insiders on the incorporation of market, industry, and firm-specific information into stock prices. The Accounting Review, 79 (4), 1119-1151. Piotroski, J.D., Wong, T.J. 2012. Institutions and Information environment of Chinese listed firms. In: J. Fan, R. Morck (Eds.), Capitalizing China. National Bureau of Economic Research and University of Chicago Press. Scholes, M., & Williams, J. (1977). Estimating betas from nonsynchronous data. Journal of Finance, 5, 309-328. Sinha, P., Gadarowski, C. 2010. The efficacy of Regulation Fair Disclosure. Financial Review, 45 (2), 331-354. Solomon, D., Soltes, E. 2013. What are we meeting for? The Consequences of Private Meetings with Investors. Working Paper. University of Southern California. Tong, H. (2007). Disclosure standards and market efficiency: evidence from analysts' forecasts. Journal of International Economics, 72, 222-241. Venkataraman, R. 2001. The impact of SFAS 131 on financial analysts’ information environment. Ph.D. Thesis, The Pennsylvania State University. Wang, J., Ahammad, M.F. 2012. Private information acquisition and stock evaluation by Chinese financial analysts. International Journal of Management, 29 (1), 117-131. Wang, J., Haslam, J., Marston, C. 2011. The appraisal of ordinary shares by Chinese financial analysts. Asian Review of Accounting, 19 (1), 5-30. Wang, Y., Chen, S.K., Lin, B.-X., Wu, L. 2008. The frequency and magnitude of earnings management in China. Applied Economics, 40, 3213-3225. 59 Wurgler, J. 2000. Financial markets and the allocation of capital. Journal of Financial Economics, 58, 187-214. Xu, N., Chan, K.C., Jiang, X., Yi, Z. 2013. Do star analysts know more firm-specific information? Evidence from China. Journal of Banking & Finance, 37 (1), 89-102. Yang, S. 2010. Is information disclosure of listed companies fair? Finance and Trade Research (in Chinese), 2010 (5), 113-119. Yang, S. 2012. Has information leakage of listed companies decreased? Based on comparison pre and post implementation of fair disclosure of information. Finance and Trade Research (in Chinese), 2012 (2), 143-150. Yu, Z., Li, L., Tian, G., Zhang, H. 2013. Aggressive reporting, investor protection and stock price informativeness: evidence from Chinese firms. Journal of International Accounting, Auditing and Taxation, 22, 71-85. Zhu, H., He, X., Tao, L. 2007. Does financial analysts following improve the stock price synchronicity? Journal of Financial Research (in Chinese), 2007 (2), 110-121. Zhu, H., Wang, H. 2009. The economic consequences of the fair disclosure of information. Management world (in Chinese), 2009 (2), 23-35. 60