The Demand for Audit in Private Firms: Recent Large-Sample Evidence from the UK

Abstract

Although theory suggests that companies would rationally select into audit even if it were not a legal requirement, many countries impose mandatory audits. This is arguably due to an audit having elements of a public good, which may result in not enough audits being purchased without regulatory intervention. The mandatory nature of public company audit has created problems for researchers wishing to investigate the demand for voluntary audit. Recent events in the UK, however, have provided such an environment. In the UK, private companies must publicly file financial statements and, until recently, they had also to be audited. However, this requirement has now been relaxed for many private companies. We are therefore able to examine the determinants of voluntary audit in a large sample of companies for which we have financial statement data. We analyse a sample of 6274 recently exempt companies, following them for three years post-exemption. We use agency theory and prior evidence to generate our hypotheses and examine them using a more comprehensive set of explanatory variables than has previously been available in the literature. Our results indicate that companies are more likely to purchase voluntary audits if they have greater agency costs, are riskier, wish to raise capital, purchase non-audit services from their auditor, and exhibited greater demand for audit assurance in the mandatory audit regime. We also document a trend away from audit over time. Overall, our results strongly support the idea that companies choose to be audited when it is in their interests to do so.

Acknowledgements

This paper has benefited greatly from the input of seminar participants at the Universities of Bath, Warwick, Exeter and Cardiff. Useful comments were also received from academics attending the BAFA and AFAANZ conferences. We are particularly grateful to Vivienne Beattie, Peter Carey, Mark Clatworthy, Andy Stark and Teri Yohn for sharing their thoughts with us. The efforts of the associate editor, Ann Vanstraelen, and two anonymous reviewers helped us to improve the paper substantially. Remaining errors are our own.

Notes

1 The UK regulator at the time was the Department for Business, Enterprise and Regulatory Reform, which was superseded in 2009 by the Department for Business, Innovation and Skills.

2 Section 161(1), which required the auditor of a company to be a ‘member of a body of accountants established in the United Kingdom and for the time being recognised for the purposes of this provision by the Board of Trade’, did not apply to exempt private companies. Section 161(2)(b), which precluded ‘a person who is a partner of or in the employment of an officer or servant of the company’ from conducting the company audit, also did not apply to exempt private companies.

3 On the other hand, an argument against mandatory audits is that it prevents companies from using voluntary audit to signal their type (Titman and Trueman, 1986; Lennox and Pittman, 2011).

4 Companies Act 1985 (Audit Exemption Regulations) 1994 (SI 1994/1935).

5 Companies Act 1985 (Audit Exemption Regulations) 2004 (SI 2004/14).

6 None of the companies in our sample had year-ends falling between 1 and 29 January 2004. Henceforth, we therefore refer to the voluntary audit regime as starting in 2004.

7 Examples of incorporated societies include the New Zealand Rugby Union and the New Zealand Automobile Association.

8 While listed companies may direct such a signal to public capital markets, private companies are more likely to wish to provide assurance to trade suppliers (regarding liquidity) and to their banks (regarding asset values and income, inter alia).

9 The other exemptions available under PASE relate to economic dependence and an audit partner joining a client company.

10 We thank an anonymous reviewer for pointing this out.

11 To be classed as small by the regulator, companies must satisfy two out of three size criteria over a period of two years, the first two criteria being the same as the turnover and total asset audit exemption thresholds and the third being a maximum average number of employees of 50.

12 Our results are also robust to estimations using a random-effects model. Wooldridge (2002) points out that there is no fixed-effects probit model.

13 To check whether our NAS results are robust, we also compute NAS as the ratio of non-audit fees to total fees (%NAS) and re-run our probit model given in Equation (1). The results are similar to those reported in the paper, i.e. there is a significant positive relation between %NAS and the likelihood of voluntary audits.

14 One exception is Seow (2001) who finds a significant positive association between the number of non-director shareholders and voluntary audit in a small sample of 32 observations.

15 We thank an anonymous reviewer for pointing this out to us.

16 In untabulated tests, we replace ROA2003 with contemporaneous ROA. The relationship remains negative but less significant. (This may be due to differences in the reporting practices of audited and unaudited companies but investigating this is beyond the scope of this study.) Coefficients and t-statistics on ROA obtained in those tests are: for 2004 coeff. = −0.047 (t = −0.54); for 2005 coeff. = −0.096 (t = −0.99); for 2006 coeff = −0.34 (t = −2.93). In the pooled test the coefficient is negative (−0.139) and significant (t = −2.39). The signs and significance of the other variables are not altered significantly.

17 The Pseudo R²s in Table 5 range from 9.49% to 11.57% indicating that there are other unmodelled factors that affect the voluntary demand for auditing. However, this level of explanatory power is similar to prior research explaining a company's decision to have an external audit compared to no audit. For example, the study by Carey et al. (2000) reports a Pseudo R² of 14.1%; Hay and Davis (2004) report a Pseudo R² of 10.8% for their model of auditor choice (0 or 1). In addition, the Pseudo R²s in probit models tend to be much lower than the R²s in OLS models because the continuous dependent variable in probit models is unobserved (Veall and Zimmerman, 1996). Further we report the more common McFadden (1973) Pseudo R² which is generally much lower than alternative Pseudo-R² measures (Veall and Zimmerman, 1996).

18 The size-matched sample is constructed by matching the companies that purchase a voluntary audit with companies that cease to be audited where the matching is based on total assets and year ends. The size-truncated sample consists of all companies that opt out of audit and the smallest 80% of firms that retain the audit in terms of their total assets.

19 The unexpected audit fee is computed using the residuals from a standard audit fee model employed in prior literature (e.g. DeFond et al., 2002). Details of the model are available from the authors upon request. The size-adjusted board measure is the residual from a regression of the number of current directors on the natural log of total assets. In untabulated tests, we replace ROA2003 with contemporaneous ROA. The relationship between contemporaneous profitability and voluntary audit remains negative and is significant in all years at 5% or better. The signs and significance of the other variables are not altered significantly.