CHAPTER 8 CONCLUSIONS, IMPLICATIONS AND LIMITATIONS . 239
4.2 S AMPLE SELECTION AND DATA
4.2.2 The criteria for data collection
In this study, the following criteria will be employed to guide the choice of the sample of companies: (i) the companies must be listed on the SGX Mainboard (for the case of Singapore), or the HOSE and the HNX (for the case of Vietnam);
(ii) financial firms and banks are excluded from the sample; (iii) the companies must be locally incorporated; (iv) the firms’ annual reports for the period of 2008–
2011 are available; and (v) the firms’ corresponding financial data for the period of 2008–2011, including market-based data and accounting-based data, must be available on Thomson One Banker (Worldscope database). The companies have
31 http://info.worldbank.org/governance/wgi/index.aspx#home
32 http://www.doingbusiness.org/
72
to meet the abovementioned five criteria to be included in the final sample due to the following reasons.
First, the basic reason for excluding firms listed on the UPCoM (Vietnam) and Catalist (Singapore) is that listing requirements of these two markets are different from those of main-board markets. For example, it is not compulsory for the Singaporean listing applicants on Catalist to meet any minimum quantitative entry criteria except for a sponsor’s acceptance. Whereas, a company can only list on the SGX Mainboard if it completely meets some strict requirements, including revealing pre-tax profits, market capitalisation, shareholding spread, operating track record, continuing listing obligations, accounting standards, and continuity of management33.
Similarly, Vietnamese listing applicants on the HNX or HOSE must fulfil several conditions stipulated in Decree No. 14/2007/NÐ-CP34, while firms trading on the UPCoM need not meet such conditions. The differences in listing requirements between the main-board and the unlisted markets may lead to different impacts both on corporate governance and financial performance variables. Hence, it is reasonable to separate companies listed on the UPCoM and Catalist markets from those listed on main-board ones when studying the relationship between corporate governance structure and financial performance (Haniffa & Hudaib, 2006).
Second, consistent with the previous literature, finance and banking industries are excluded from this study’s sample because their liquidity and governance can be
33 Further information can be seen at http://www.sgx.com
34 Decree No. 14/2007/NÐ-CP, dated January 19, 2007, issued by the Vietnam Prime Minister. This Decree provides detailed instructions to implement some articles of the Law on Securities 2006.
73
influenced by different regulatory factors (Bauer, Frijns, Otten, & Tourani-Rad, 2008; Dittmar & Mahrt-Smith, 2007; Schultz et al., 2010). For instance, financial firms and banks function under strict regulations that have different influences on corporate governance mechanisms, such as board structure (Yermack, 1996). In addition, they not only are governed by rules which do not apply to other commercial entities (Laing & Weir, 1999; Victoria, 2006), but are also subjected to specific accounting rules which may make the calculating of financial performance ratios difficult (Rose, 2007). Appendix 1 provides several illustrations of such differences in corporate governance regulations between financial industry and other industries in Vietnam and Singapore. Furthermore, given that many previous studies on corporate governance do not consider financial companies and banks (see e.g., Bauer et al., 2008; Dittmar & Mahrt- Smith, 2007; Haniffa & Hudaib, 2006; Mak & Kusnadi, 2005; Ntim, Opong, &
Danbolt, 2012; Schultz et al., 2010; Yermack, 1996 among others), excluding financial companies from the samples makes this study’s findings comparable.
Third, given the international characteristic of the SGX market, it is the sample selection criterion that publicly listed companies in Singapore must be locally incorporated. Foreign companies listed on the SGX market should be excluded from the sample since they may be subjected to various corporate governance practices. In addition, the institutional environments within which such foreign companies operate may have different effects on their corporate governance–
financial performance relationships. Therefore, this criterion facilitates a consistent comparison between the two countries’ domestic companies.
74
Fourth, the criterion of a four consecutive years’ dataset implies that the companies included in the research sample should have fully required information covering a four-year consecutive period (2008–2011). This criterion meets the requirements of proposed robustness analyses for panel data and helps to obtain a balanced panel dataset. In the presence of endogenous variables, a balanced panel dataset facilitates the estimation of this study’s empirical models. This is because the combination of panel imbalance and endogeneity may induce extreme difficulty in estimating and inferring (Flannery & Hankins, 2013)35.
Specifically, using the Monte Carlo simulation method, Flannery and Hankins (2013, pp. 13, 16) indicate that while the System GMM is likely “the most robust methodology for unbalanced panels with endogenous variables”, the root mean squared errors (RMSEs) of the endogenous variables are so much larger for unbalanced panels that it would be impossible to draw reliable inferences. For this reason, the choice of a balanced panel instead of an unbalanced one is an acceptable compromise between the sample representativeness and the estimation effectiveness, at least in this study.
However, one concern is that this criterion may introduce potential survivorship bias into the sample (Ntim et al., 2012). Taking this concern into consideration, this research explicitly reports the number of delisted and/or inactive companies discovered in the sample selection process. Table 4.2 and Table 4.4 show that the number of delisted and/or inactive companies for the Vietnamese market (3 out of 837) and for the Singaporean market (17 out of 773) accounts for relatively small
35 Flannery and Hankins (2013) used the Monte Carlo simulation to examine seven estimation methods under a variety of corporate finance dataset characteristics. The evidence and suggestions provided in their study may help empirical researchers in determining the most appropriate estimation method for the various features of datasets.
75
proportions of the initial population sizes. Moreover, the sample size of this study is far larger than that of previous studies36 which helps to eliminate the potential survivorship bias and ensure the generalisation of this study’s findings. For these reasons, it is plausible to argue that the potential survivorship bias, induced by the sample section criterion of the consecutive four-year period, may not be a problem in this study.
Finally, the year 2008 is selected because it is one year after the promulgation of the new corporate governance guidelines/regulations in both countries. More specifically, the revised Singaporean Code was issued on 14 July 2005 and came into effect from 1 September 2007. Whereas, the Vietnamese Code was first released and became effective in March 2007. Given that the new corporate governance guidelines/regulations in both countries affect their companies’ annual reports in the next financial year, 2008 is thus an appropriate point of time to collect data for the comparative purpose of the current research. The sample ends in 2011 since it is the most recent year for which data were available at the time this study was conducted. Moreover, the time frame is kept the same in both markets to facilitate the comparative purposes of this study.