Sharing corporate tax knowledge with external advisers

Abstract

Tax knowledge is critical for companies to comply with tax laws and engage in tax planning and avoidance. Firms rely on external advisers in handling tax issues, however, sharing corporate tax knowledge with external advisers entails both opportunities and risks. We identify four relational factors that are associated with the decision of corporate taxpayers to share knowledge with external tax advisers. Survey data from 221 corporate taxpayers reveals a novel distinction between operational and strategic knowledge sharing. The operational dimension has a functional nature, whereas the strategic dimension has a more intentional character. Accessibility to, and a positive experience with, external advisers enables operational knowledge sharing. When firms perceive specific tax benefits in relation to sharing knowledge, they are more inclined to engage in operational knowledge sharing with external advisers but less prone to strategic knowledge sharing. Instead, strategic knowledge sharing is enhanced when firms have access to, and value the knowledge of their advisers, although this latter factor plays no significant role in explaining operational knowledge sharing. A positive experience with advisers also associates with strategic knowledge sharing. We link our results to other research and discuss implications for regulators considering, or requiring, firm disclosures of corporate tax strategy.

Keywords:

• corporate tax
• knowledge sharing
• tax advisers
• tax planning

Acknowledgements

We thank the Associate Editor (Martin Jacob), two anonymous reviewers, Mike Peel, and participants at the 2017 Tax Research Network Annual Conference, Bournemouth, and the 2018 British Accounting and Finance Association Annual Conference, London, for their detailed and helpful comments on earlier versions of this manuscript. We gratefully acknowledge financial assistance from the ACCA.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplemental data

Supplemental data for this article can be accessed at https://doi.org/10.1080/00014788.2018.1526058

Notes

1. Informal discussions with senior officials in two tax administrations suggest that complex taxpayers often simultaneously employ advisers from several different firms of advisers. They speculate the motive is to limit advisers’ knowledge of clients’ circumstances to well defined discrete aspects. Klassen et al. (2016) report that using one’s auditor represents the smallest share of potential sources of tax advice, i.e. the firm’s own auditor, another external adviser or internal source. While the authors interpret the decision not to use the audit firm as being made to protect (perceived) auditor independence, attempting to limit auditors’ access to tax related matters is an alternative interpretation. The application of legal professional privilege to communications between a client company and its lawyers while excluding similar communications with non-lawyers, e.g. accountants, distorts companies’ decisions on whom to employ to provide advice (Prudential 2013, ICAEW 2016) and arguably the information they choose to disclose to their adviser.

2. The ACCA has statutory recognition and is a U.K. based institute for professional accountants with membership via examination. At the time of the survey, there were 18,926 Corporate Sector panel members. The ACCA was not involved in the design of the survey nor did it have control over its content.

3. Of the fully completed 180 responses, 14 had missing values with respect to control variables hence the subsequent testing on the reduced sample of 166.

4. A principal component analysis with direct oblimin rotation was conducted to determine whether the scale was unidimensional. Bartlett’s test of sphericity is significant (χ²(3) = 235.271, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.63, suggesting an adequate factorability. The three items form a unidimensional scale: only one component has an eigenvalue above 1 (initial eigenvalue is 2.19), explaining 72.9% of the total variance.

5. A principal component analysis with direct oblimin rotation was again performed to determine scale unidimensionality. Bartlett’s test of sphericity is significant (χ²(6) = 284.255, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.71, suggesting adequate factorability. The four items form a unidimensional scale: only one component has an eigenvalue above 1 (initial eigenvalue is 2.55), explaining 63.7% of the total variance.

6. The analysis with direct oblimin rotation was conducted to determine whether the scale was unidimensional. Bartlett’s test of sphericity was significant (χ²(28) = 543.891, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.82, suggesting an adequate factorability.

7. ‘Sharing knowledge with the external tax adviser(s) enables the determination of the correct tax liability’ and ‘Sharing knowledge with the external tax adviser(s) enables a decrease in tax liability.’

8. Bartlett’s test of sphericity was significant (χ²(15) = 384.313, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.73, suggesting an adequate factorability. All of the six items have a primary factor loading of at least 0.4 and none of the items have a cross-loading above 0.32 (Tabachnick and Fidell 2007).

9. The analysis with direct oblimin rotation was conducted to determine whether the scale was unidimensional. Bartlett’s test of sphericity was significant (χ²(28) = 578.013, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.79, suggesting an adequate factorability.

10. ‘My organisation is motivated to share knowledge with the external tax adviser(s).’

11. The correlation between the two components is still beyond the 0.32 threshold (r = 0.44), indicating that the analysis is suitable for further interpretation. Bartlett’s test of sphericity was significant (χ²(21) = 511.491, p < 0.001) and the Kaiser–Meyer–Olking measure was 0.76, suggesting an adequate factorability. All of the seven items have a primary factor loading of at least 0.4 and none of the items have a cross-loading above 0.32 (Tabachnick and Fidell 2007).

12. We thank a reviewer for suggesting this approach.

13. In all regression models, the VIF values are below 10 (the highest VIF level is 2.99), which shows that there is no problematic collinearity in our data, see also footnote 15 and 16.

14. The addition of this variable significantly increases the adjusted R² as reported in Table 3. Together with a maximum VIF of 2.99 this result suggests the set of independent variables in model 3 does not exhibit problematic collinearity.

15. The addition of this variable significantly increases the adjusted R² as reported in Table 3. Together with a maximum VIF of 2.68 this results suggests the set of independent variables in model 3 does not exhibit problematic collinearity.

16. There is no evidence of endogeneity (simultaneity) in either version of model 2, i.e. with the dependent variable comprising Operational knowledge sharing or Strategic knowledge sharing respectively. In all cases the null hypothesis of the Wu-Hausman test cannot be rejected at acceptable significance levels. We thank a reviewer for raising this point.

17. ‘Non-engaged’ respondents were defined as respondents with a zero standard deviation of their responses across the attitudinal questions.

18. The extreme values of the two variables, Number of Jurisdictions and Number of Employees were defined after visually examining the data as values in excess of 48 and 98,000 respectively, resulting in the respective exclusion of 10 and 5 cases.

19. We do not report in the paper the three sets of regression analyses discussed, but they are presented in a separate appendix that is available via an email request to the corresponding author.

20. A general unwillingness to disclose is consistent with the observation that firms rarely voluntarily publish their compliance risk rating produced by HMRC’s ‘Business Risk Review.’ This even holds for firms classified by HMRC as being ‘Low (Compliance) Risk.’

Additional information

Funding

This work was supported by ACCA: [Grant Number The Management of Tax Knowledge].