CHINA: A SET THEORETIC ANALYSIS
3 EMPIRICAL STUDY: METHODOLOGY
3.4 Causal Conditions
Large controlling shareholder. The proportion of shares of the largest shareholder is the most used measure of ownership concentration (Huyghebaert & Wang, 2012; Jiang et al., 2010; Lin et al., 2009; Qiu & Yao, 2009; Wang & Xiao, 2011; Wu et al., 2011; Yeh et al., 2012). The full non-membership, crossover point and full membership are again calibrated according to the 5th, 50th and 95th percentiles. The values are 17.6%, 37.8% and 68%. Since the theoretical prediction is that a large controlling shareholder contributes to the absence of tunneling, these thresholds are consistent with results from past studies. Jiang et al. (2010) found that tunneling is highest when the largest block holder owns less than 30% of total shares, which falls comfortably below the chosen crossover point of 37.8%. Chen et al. (2011) found that the proportion of shares held by the largest shareholder is positively correlated with firm performance until it reaches a threshold of 48%, which falls well above the crossover point.
Large minority. This variable replicates previous studies that use the sum proportion of shares of the second to tenth largest shareholders as a measure of the presence of large minority shareholders (Berkman et al., 2009; Faccio et al., 2001; Huyghebaert & Wang, 2012; Qiu & Yao, 2009). The 5th, 50th and 95th percentiles were respectively 1.7%, 18.8% and 44.3%. The presence of this condition together with a large controlling shareholder indicates that the firm has a very high ownership concentration among the ten largest shareholders.
Float. The floating ratio is the proportion of shares that are negotiable on the stock exchange. The 5th, 50th and 95th percentiles were respectively 25%, 39.6% and 63.5%.
Independent board. Board independence is measured as the proportion of independent directors on the board of directors. According to the Guiding Opinion on the Establishment of the Independent Directors’ System by Listed Companies issued by the China Securities Regulatory Commission (CSRC) in 2001, independent directors cannot include any direct family member of anyone employed by or holding over 1% of the shares of the firm or of a subsidiary. The Opinion of the CSRC mandates that at least a third of the board must be independent as a minimum threshold. Following this, firms that display a proportion of independent directors lower than 33.3% are coded as 0. Those that barely comply with the 33.3% rule are coded as .25. Firms that display more than the mandated threshold of 33.3 % receive .51, meaning that they are slightly in the set. Following international best practices, including the rules of the NYSE and the conclusions from the Cadbury report in the UK as well as similar provisions in other countries (Australia, Brazil, Finland, etc.) which recommend that at least half of the board be independent, firms with at least 50% independent directors are coded as fully in (1). To complete the incremental scale, boards with at least 40% independent directors but less than 50% are coded as .75. Following this scale, the term “independent board” should be understood as boards consisting of more than a third of independent directors.
Proactive disclosure. There is no single best systematic measure of disclosure used in the literature, but the timeliness of release of the annual report has been shown to be a sensitive element of disclosure in listed firms (Haw, Qi, & Wu, 2000; Haw, Park, Qi, & Wu, 2006; McGee & Yuan, 2012; Park, Song, Yang, Hossain, & Koo, 2013). Following this
insight, the delay between year-end and the date of release of the annual report is used as a proxy for proactive disclosure. The scale is anchored according to the three reporting deadlines mandated by the Securities and Exchange Commission in the US for filing the 10- k form, which is a standardized version of the financial statements handed to the authorities: 60 days, 75 days and 90 days. While these forms are different in kind from full fledge corporate annual reports and despite the fact that the three deadlines are originally meant to apply to firms of different sizes, these dates are used as indications of proactive disclosure, which should intrinsically go above and beyond mere compliance. The CSRC in China indeed mandates a maximum deadline of 120 days for all listed firms. Consequently, a delay of 60 days or less between year-end and release was coded as fully in (1). 75 days or less was coded as in (.67). 90 days or less was coded as out (.33). More than 90 days was coded as fully out (0).
Independent audit. Two distinct elements combine to construct a measure of independent internal audit. Firstly, to be included in the set, firms must have set up an audit committee in the board of directors. Audit committees are involved in the preparation of annual reports, have access to financial officers, are responsible for recommending external auditors and have privileged access to the auditors. The presence of audit committees is generally expected to improve the overall audit quality (Stewart & Munro, 2007). Secondly, audit committees in China are supposed to include a majority of independent directors, preferably including the chairman of the committee. Moreover, it is strongly advised that an independent director with an accounting title sit on the committee. Ample literature discusses the benefits of independent audit committees. Such independence has been found to reduce earnings manipulation (Klein, 2002), instances of fraud (Abbott, Park, & Parker, 2000) and to enhance firm value (Chan & Li, 2008). While the composition of the audit committee is disclosed in the annual reports, including which members are independent directors, additional information was used to validate their independence. To be included in the set, the disclosed working address of the independent director with an accounting title must be different from the address of the firm. This resulted in a binary causal condition. Firms with an audit committee on the board in which the independent director with an accounting title has a different address than that of the firm were coded as 1. All other firms were coded as 0.
High leverage. Consistent with previous studies, leverage was measured as a ratio of total liabilities to total assets (Jiang et al., 2010; Wang & Xiao, 2011; Yeh et al., 2012). However, to account for the considerable variance in leverage across industries, an adjusted measure subtracts the leverage from the industry average, using the first order CSCR classification and second order classification in the case of manufacturing firms. This procedure is often used in the literature (Acharya, Amihud, & Litov, 2011; Berger & Ofek, 1995; Billett & Xue, 2007), including in studies of Chinese firms (Berkman, Nguyen, & Zou, 2011; Cheung, Jing, Lu, Rau, & Stouraitis, 2009). The 5th, 50th and 95th percentiles were respectively -32.5%, 0.2% and 42%.
Large firm. In line with many past studies, total assets are used as a measure of firm size (Chen et al., 2009, 2011; Liu & Lu, 2007; Yeh et al., 2012). The 5th, 50th and 95th percentiles were respectively 345 million Yuan, 1,518 million Yuan and 9,325 million Yuan. With such a calibration, the absence of large firm can also be interpreted as the presence of relatively small firms.
State control (SOE). This causal condition is coded as a simple binary variable. Firms ultimately controlled by a state agency at any level of government, as disclosed by the annual report, are coded as 1. Other firms are coded as 0. Considering the condition of state control in conjunction with large controlling shareholder provides an indication on the shareholding size of the state in SOEs.