Audit Market Segmentation – The Impact of Mid-tier Firms on Competition

Abstract

This analysis’ goal is to characterise the impact of mid-tier auditors on competition and average supplied audit quality. The major result of our analysis is the following: Auditors’ flexibilities with respect to clients’ complexity determine the offered audit quality levels and thus the average supplied audit quality. Defining a model-endogenous quality measure, we show that for some instances the mid-tier auditor improves average audit quality, but for other instances fails to do so. The result could be of special interest for standard setters, for example, the European Commission, which is currently revising EU audit regulation. Further, the analysis may serve as an instrument to analyse economic consequences of future changes of regulation.

Acknowledgements

We thank Anne Chwolka, Joel Demski, Eva Eberhartinger, Michael Ebert, Dirk Hachmeister, Hermann Jahnke, Robert Knechel, Carolyn Levine, Rainer Niemann, Paul Polinski, Christian Riegler, David Sappington, Jack Stecher, Jeroen Suijs, René Thamm, Alfred Wagenhofer, and Stefan Wielenberg for their helpful comments on earlier versions of this paper. Further, we thank the associate editor and two anonymous referees for helpful comments. We also thank the workshop participants at the Fisher School of Accounting/University of Florida (USA), Georg-August-University of Göttingen (Germany), Karl-Franzens-University of Graz (Austria), Leibniz-University of Hanover (Germany), the University of Economics and Business Vienna (Austria), and the University of Paderborn (Germany). Further participants of the annual conferences of the American Accounting Association in New York (USA), the European Accounting Association in Tampere (Finland) and the VHB in Nürnberg (Germany) provided helpful comments. Dirk Simons acknowledges funding from SFB 504 by the German research foundation (DFG). Nicole Zein acknowledges funding from the German Academic Exchange Service (DAAD).

Notes

¹Cf. Zijl and Karim (2008), Fan and Wong (2005), DeFond, Francis, and Wong (2000), DeFond, Wong, and Li (2000), Narasimhan and Chung (1998), and Lee (1996) for studies covering Asian countries. Empirical evidence for continental European countries is collected, for instance, in Bigus and Zimmermann (2008), Piot and Janin (2007), Stefani (2006), Ashbaugh and Warfield (2003), Willekens and Achmadi (2003), Piot (2001), Buijink, Maijoor, and Meuwissen (1998), Marten and Schultze (1998), Weets and Jegers (1997), and Christiansen and Loft (1992). Anglo-American data are analysed by McMeeking (2007), Baskerville and Hay (2006), Oxera Consulting Limited (2006), Pong and Burnett (2006), Beattie, Goodacre, and Fearnley (2003), Pong (1999), Choi and Zéghal (1999), and Narasimhan and Chung (1998).

²Empirically the differences between the then Big-5, the Middle-4, and the local auditors are shown in Choi, Kim, Kim, and Zang (2010) with respect to number of offices, average audit fee per office, and number of clients for the US market. The recent world survey by IAB (2013) suggests a segmentation into Big-4, Next-10, and others.

³However, empirically the existing gap in size between the Big-4 and the Next-10 is significant. Comparing the median Next-10 firm to the average Big-4 size shows the following: in terms of revenues (staff), the median Next-10 reaches 11% (15%) of the average Big-4 size. For more details, see IAB (2013).

⁴For instance, in Germany, the fraction of revenues that is allowed to be earned from a single client is capped at 15% or 30% of the auditor's total revenue depending on the client's listing status, see §§ 319 III No. 5, 319a I No. 1 HGB.

⁵The cited empirical evidence covers the segment of listed client firms. However, reputational concerns in a broader sense should apply for the private firm segment, too.

⁶Cf. Chan (1999) for a Hotelling-setting in the context of low-balling. Cf. Bleibtreu and Stefani (2012) for a circular-city model focusing on the interdependencies between market structure and non-audit services.

⁷High-quality levels are indicated by small values of i_j.

⁸Without loss of generality, the fixed costs for performing an audit which just meets minimum standards can be neglected, as they are identical across auditors. Hence only customisation costs matter.

⁹An interpretation of θ will be offered at the end of this section.

¹⁰For technical convenience, the cost function is symmetric. The major results can be obtained from other specifications, for example, with cheaper customisation in the case of downsizing, see Appendix 9.

¹¹Note that only costs of mismatch between the auditor's standard audit plan and the client's attributes are regarded, see Chan (1999, p. 616).

¹²Lower refers to the benchmark level underlying the auditor's standard audit plan.

¹³A compelling analogy can be taken from a chemistry lab: assume a chemist owns a tool case containing several detection devices (i.e. the toolkit represents the standard audit approach). In the first case (representing downsizing), you have to deal with chemical substances, which are diluted with water (representing low quality of the accounting data). Your detection device (representing your combination of analytical and substantive testing), although designed to identify the chemical substance at hand (the auditor has experience with business models similar to the client's one), does not work, because the concentration of the substance is too low. Hence, you have to look for other ways of detection (i.e. you have to adjust your audit plan), which causes costs. In the second case (representing upsizing), you have to deal with chemical substances you are not prepared for (the auditor has to become familiar with the client's business model, because his standard approach is not applicable for this complexity). Hence, you need to look for new devices causing additional costs again.

¹⁴An alternative model specification, where adjusting the standard audit approach to the needs of a less demanding client (r > i_j, downsizing) is cheaper than adjusting for a more demanding client (r < i_j, upsizing), is presented in Appendix 9.

¹⁵See Appendix 3.

¹⁶See Appendix 4.

¹⁷In general, for a Hotelling setting with three competitors, an equilibrium solution cannot be guaranteed, see Economides (1993) or Eaton (1972). Setting i₃ = 1 circumvents this problem.

¹⁸From Lemma 3.1 we get r_1,2 < r_1,3 < r_2,3.

¹⁹The subscript of Ψ indicates the number of considered tiers.

²⁰If θ₂ would become smaller than θ₁ then the auditors would just change positions with the old mid-tier auditor now being the first-tier auditor.

²¹The decimal numbers result from numerical operations with