Content area
Full text
1 Introduction
Debt issuance and the exercise of employee stock options (ESOs) provide companies additional funding and tax benefits through resulting deductions. Given these common traits, firms potentially can substitute ESOs for debt financing. [2] Aier and Moore (2008) find that such substitution occurs and is influenced by firms' tax status. Continuing this line of research, we conjecture that this substitution may also vary based on the relative size of a firm's interest expense. Furthermore, the substitution effect may be impacted by the change in the accounting treatments of ESOs required by Statement of Financial Accounting Standards (SFAS) 123R. We investigate these possibilities by utilizing Censored Quantile Regression (CQR) analysis that allows for an assessment of the substitution effect across different levels of interest expense.
The seminal work of [41], [42] Modigliani and Miller (1958, 1963) suggests that the deductibility of interest expense makes debt more appealing than equity financing if transaction costs are assumed not to exist. [15] DeAngelo and Masulis (1980) advance the theory by proposing the substitution hypothesis that predicts a negative relationship between alternative tax shields (e.g. depreciation) and debt. The expected value of interest tax shield declines as total deductions grow and the probability of having zero or negative earnings increases.
Empirical studies of the substitution effect have focused on investigating the impact firms' tax status and improving the modeling of the dependent variable. Recognizing that the substitution theory is dependent on a firm's tax status, [37] MacKie-Mason (1990) finds that the negative relationship between debt issuance and the usage of investment-related tax shields only exists for firms shifting to a lower tax bracket. [13] Cloyd et al. (1997) refines the tax status evaluation further by exploring the substitution effect among three groups of firms: one group (tax-sensitive) that is sensitive to additional deductions due to a moderate income level and the other two groups that are not sensitive because the income is either very low (tax-exhausted) or very high (tax-insatiable).
To fulfill the [15] DeAngelo and Masulis (1980) assumption of constant before-tax earnings, [17] Dhaliwal et al. (1992) test the substitution hypothesis using the ratio of interest expense to operating earnings rather than a leverage ratio as the dependent variable. In addition, the dependent variable based on interest...