7 /304
7
臺大管理論叢
第
28
卷第
1
期
Several studies document evidence of pyramid-like structures in East Asia (e.g., La
Porta et al., 1999; Claessens et al., 2000; Claessens et al., 2002). Bebchuk et al. (2000),
Morck, Wolfenzon, and Yeung (2004), and Johnson, La Porta, Lopez-de-Silanes, and
Shleifer (2000) argue that corporate pyramid can be used to facilitate non-arm’s length
transactions. Gramlich, Limpaphayom, and Rhee (2004) document that keiretsu group
member firms have lower effective tax rates than stand-alone firms in Japan. Keiretsu
firms strategically shift reported income among affiliates in order to reduce overall
effective tax rates. With the control right disproportionate to the ownership, controlling
shareholders can easily expropriate wealth from minority shareholders and creditors
(Bebchuk et al., 2000; Morck et al., 2004). The resources of the lower layer firm can
easily be transferred out of the firm for the benefit of the controlling shareholders, a
practice known as tunnelling (Johnson et al., 2000). Thus, pyramidal structures shift the
central agency problem from a problem between outside investors and managers (Jensen
and Meckling, 1976) to a problem between controlling shareholders and other
stakeholders (Shleifer and Vishiny, 1997; La Porta et al., 1999).
2.2 Literature on Tax Avoidance and Firm Characteristics
Prior studies have identified many firm characteristics as being associated with tax
avoidance using the actual data of tax avoidance activities or a number of proxies based
on the financial reporting data (e.g., GAAP effective tax rate, cash effective rate, book-tax
difference, etc.). The characteristics include firm size, profitability, leverage, capital
intensity and foreign operations (e.g., Gupta and Newnerry, 1997). For example, Gupta
and Newnerry (1997) document that effective tax rates are negatively associated with a
firm’s leverage, PPE intensity (the ratio of net property, plant, equipment to total assets),
and positively associated with firm performance (ROA) and inventory intensity. Rego
(2003) finds that firms of greater pre-tax income have more incentives to engage in tax
avoidance activities and that firms operating in more dispersed geographic areas have
more opportunities to take tax avoidance activities.
Using confidential tax shelter and tax returns from the Internal Revenue Service,
Lisowsky (2010) develops a model to infer the likelihood that a firm engages in a tax
shelter. He finds that tax shelter likelihood is positively associated with subsidiaries
located in tax havens. However, Lisowsky (2010) suffers from three drawbacks. First, he
does not explore whether the practice of tax shelter can vary with investment structure. He
assumes that the organization structure is not a factor affecting tax shelter. Second,