Wang, Yicheng
(2022)
Detecting Tax Avoidance: Evidence from Quantitative and Linguistic Cues.
PhD thesis, University of Liverpool.
|
Text
Final Thesis_Yicheng Wang_201305701.pdf - Unspecified Download (2MB) | Preview |
Abstract
In this thesis, I examine tax avoidance detection by the use of quantitative and linguistic cues. This thesis provides a battery of approaches to detect tax avoidance, which contribute new tax avoidance means to the approaches already extant in the literature. All the approaches are based on the most used public disclosure of firms, the 10-Ks in the United States. I structure this thesis with three free-standing but theme-related papers. In the first paper, I focus on the quantitative cues to detect tax avoidance by exploring conforming tax avoidance. Omitting conforming tax avoidance would result in an incomplete picture of the extent to which firms avoid tax. Given the importance of Badertscher et al.’s (2019) attempt to develop the only existing measure of conforming tax avoidance, it is vitally important to tax authorities and policy analysts that their approach provides rigorous and consistent results. Therefore, I present two major issues with their approach, which I analyse in some detail and provide solutions to. I apply the refined measure to two scenarios where the prior findings might be incorrect because of the lack of conforming tax avoidance measures. Exploring the quantitative cues of tax avoidance could be inadequate, complex, and inefficient as financial measures are one-dimensional, misspecified, and/or not widely available. Taking 10-Ks as a repository of firms’ narratives and treating accounting as a compound of philosophy and mathematics, I expect to see more linguistic cues of tax avoidance in 10-Ks. However, this kind of research is very limited in prior literature. Thus, in the second and third papers, I use textual analysis to detect tax avoidance in 10-Ks. In the second paper, I focus on one specific section, Management Discussion and Analysis (MD&A) in 10-Ks. I use the existing well-established dictionaries to detect tax avoidance. I find that tax avoidance is significantly associated with the tone change of the MD&A section. In the third paper, I construct a tax-related dictionary to measure a firm’s ability to engage in tax avoidance. I apply the dictionary in the entire 10-Ks and find a significantly positive relation between the raw counts of words in this dictionary and the level of tax avoidance. Both papers provide incremental linguistic cues beyond traditional accounting variables to reveal and predict tax avoidance. The findings in all three papers together provide a set of multi-dimensional approaches for researchers, investors, and tax authorities to detect tax avoidance in 10-Ks in a more comprehensive, informative, and efficient way.
| Item Type: | Thesis (PhD) |
|---|---|
| Uncontrolled Keywords: | tax avoidance; conforming tax avoidance; tax-related words; textual analysis; linguistic cues |
| Divisions: | Faculty of Humanities and Social Sciences > School of Management |
| Depositing User: | Symplectic Admin |
| Date Deposited: | 14 Feb 2022 16:02 |
| Last Modified: | 18 Jan 2023 21:16 |
| DOI: | 10.17638/03146458 |
| Supervisors: |
|
| URI: | https://livrepository.liverpool.ac.uk/id/eprint/3146458 |
Dimensions
Dimensions
