Blockholders’ Ownership and Audit Fees: The Impact of the Corporate Governance Model

Abstract

This paper examines how two prominent corporate governance models, namely the shareholder and stakeholder models, have different effects on the relation between agency conflicts and the supply, and demand of audit services. Shareholder (stakeholder) countries rely heavily on public (private) information to reduce information asymmetry for outside investors in the context of high (low) litigation risk. We expect audit fees to reflect the level of agency conflicts in shareholder countries as well as the needs for information of the major blockholders in stakeholder countries. Using a sample of 7982 firm-year observations from 19 countries, we find a U-shaped relation between controlling shareholding and audit fees for shareholder countries and an inverted U-shaped relation between controlling shareholding and audit fees for stakeholder countries. These results are consistent across different firm-level governance arrangements.

Acknowledgements

The authors thank the editor, Luzi Hail and the anonymous reviewer for their insightful comments and guidance through the review process. We are also pleased to thank W. Alissa, D. Beneish, C. Beuselinck, V. Capkun, D. Hay, G. Hilary, C. Mangen and participants at the HEC-INSEAD joint workshop (2010), AFC (French Accounting Association, 2010), INTACCT workshop (Paris, 2010), Concordia University workshop (2015), Wien University (2015), Padova University (2016), EAA (European Accounting Association, 2013) as well as two anonymous reviewers at the AAA Auditing section (2011) for their useful comments on earlier versions of this paper.

Notes

1 Along with Ball et al. (2000), we acknowledge the diversity in litigation risk within common-law countries. For instance, Ball et al. (2000) classify the United States as the country with the highest litigation risk, followed by Australia and Canada, and classify the United Kingdom as having the lowest litigation risk. In an experimental paper, Hwang and Chang (2010) show that in the United States and Hong Kong – two common-law countries – auditors are more accommodating to client’s needs as the litigation environment weakens. This finding aligns with those of other studies using samples from both common and code-law countries (Choi et al., 2009; Francis et al., 2008).

2 In code-law countries, civil litigation is comparatively rare and the size of awards is relatively small.

3 See for instance the International Standards on Auditing – ISA 315, Understanding the Entity and its Environment and Assessing the Risks of Material Misstatement. All audit methodologies developed by large audit firms across countries (especially by Big 4 and second-tier audit firms) follow similar audit standards decomposing the audit risk into inherent risk, control risk and detection risk.

4 Despite the existence of pyramidal structures and double types of shares in practice, we simplify our argument assuming that control increases with the increase of cash flow rights. While these governance structures affect the slope of the control function, the basic reasoning remains unchanged.

5 Wingate (1997) ranks the litigation risk faced by the auditors according to the level of insurance premium paid by the audit firms over the world. On a scale of 1 to 15, common-law countries mostly rank between 10 (e.g. the United Kingdom) and 15 (the United States, corresponding to the maximum risk for auditors), while code-law countries mostly rank between 3 and 6 (e.g. 6.22 for France or Germany). We use this index in a robustness test (see Section 4.3).

6 For instance, in the United States, Reg FD (Regulation Fair Disclosure) forces management to disclose any material information at the same time to all shareholders. However, shareholders owning 5% or more of the shares or holding board seats are considered insiders and other rules apply. Therefore, in this monitoring phase, these larger shareholders act as a monitoring element for all other shareholders, reducing the need for additional audit effort and assuming part of the litigation risk.

7 Without mean-centered transformation, the condition on the linear term would have been: −2β₁< β₂< 0 for H1, and 0 < β₂< −2β₁ for H2.

8 This design is similar to Choi et al. (2008). However, we also test alternative clusters. Country clustering is not possible given the lower number of countries compared to the number of variables. Industry clustering (4-digit SIC code) gives similar results to the main analysis.

9 The impact in percentage of the audit fees generated by a change from x% ownership to (x  + Δ)% is given by the following transformation: % in audit fees = exp (Δ*(β₂+  β₁(2x +  Δ –  2m))) where m represents the mean of ownership for the shareholder or for the stakeholder sample.

10 For the stakeholder countries, main results continue to hold for most thresholds up to 40%.

11 This result should be regarded with caution, as the Worldscope field (#18370) used to extract the blockholding data explicitly mentions cash flow shareholding higher than 5%. Hence, shareholdings below 5% – specifically at 0% – should not be regarded as reliable (e.g. Faccio & Lang, 2002), as it may correspond to data that are not or only partially fulfilled, while blockholders with less that 5% may actually exist.

12 According to these references, countries in our sample that are market-based countries are the United Kingdom and the United States, and bank-based countries are Germany, France, and Japan. We ran the main test on these two subsamples. We obtain similar results to the main analysis.

Additional information

Funding

Cédric Lesage acknowledges the financial support of the Fondation HEC (Project F1103). Most of this work was done while he was at HEC Paris.