What Drives Voluntary Audit Adoption in Small German Companies?

The purpose of this study is to validate the drivers of voluntary audit in small companies identified in previous research and uncover additional determinants related to agency conflicts with owners. For our research we use the German institutional setting, documented in the literature as being very different from its Anglo&#8208;Saxon equivalent. Based on a random sample of 405 small companies responding to a postal questionnaire survey, we find that the proportion of owners not involved in management, the subsidiary status of a company, a company's legal form, and the importance of financial statements' information to management activities all increase the likelihood of voluntary audit. In contrast, firms that outsource accounting tasks to an external expert are less likely to opt for voluntary audit, suggesting that an external expert's involvement substitutes for an external audit. In addition, owing to the absence of a statutory audit history for small companies in Germany, we find that voluntary audits are less common compared with findings from previous studies.

previous research studies having been gathered primarily from data based on publicly listed companies or Anglo-Saxon accounting settings. According to the synthesis of prior auditing research on private companies, Langli and Svanström (2014) argue that: The differences that exist between private and public companies are so large and fundamental that without careful consideration we cannot rely on findings for public companies when we want to understand the role of auditing in private companies.
Responding to this call for additional evidence on voluntary audit drivers and accounting processes in private companies, we gathered survey questionnaire data from 405 usable responses by small private companies in Germany regarding financial statements' information for the financial year ending 2011. The data centered around questions examining voluntary audit drivers that relate to: (i) agency conflicts with owners; (ii) agency conflicts with lenders; (iii) the importance of financial statements for management activities; and (iv) the outsourcing of accounting tasks.
Our data show that voluntary audits are considerably less common among German firms than they are with firms in other countries.
While previous studies have found voluntary audit ranging between 26% and 80% (see Table 5) of businesses, we find that only 12% of our sample companies opt for a voluntary audit. Our results indicate that managers from countries where they have been subject to mandatory audit regimes value the cost and benefits resulting from a voluntary audit very differently from those who have not, demonstrating that the voluntary audit decision is likely to be influenced by previous habits (Niemi, Kinnunen, Ojala, & Troberg, 2012;Oliver, 1991).
Regarding the drivers of voluntary audit, in line with previous research, we find that the proportion of company owners who are not involved in management (e.g., Collis, 2010Collis, , 2012Seow, 2001;Tauringana & Clarke, 2000) and the importance that managers place on accounting information for management accounting purposes (Collis, 2010(Collis, , 2012Collis, Jarvis, & Skerratt, 2004;Niemi et al., 2012) increase the likelihood of an auditor being hired voluntarily. In contrast with previous studies on voluntary audit, we cannot find support for the status as a family firm (e.g., Collis, 2010;Collis et al., 2004), ownership dispersion (Dedman, Kausar, & Lennox, 2014), or leverage (Carey, Simnett, & Tanewski, 2000;Dedman et al., 2014;Tauringana & Clarke, 2000) impacting on a firm's voluntary audit decision. However, extending previous research, we do find evidence that the legal form in which a company operates, the status as a subsidiary, and outsourcing are further factors impacting on a manager's voluntary audit decision. By further examining the professional qualifications of those to whom accounting tasks are outsourced, we provide evidence that outsourcing accounting tasks to an external tax advisor decreases the likelihood of a voluntary audit, whereas outsourcing accounting tasks to an external auditor increases the likelihood of a voluntary audit. Subject to the professional qualifications of those to whom financial accounting tasks are outsourced, this result suggests that auditing can play a substitutive or a complementary role.
Our study contributes to the literature as follows. First, it validates earlier research findings from other jurisdictions in an institutional setting that has been documented in the literature as being very different from the Anglo-Saxon regime (Alexander, Britton, Jorissen, Hoogendoorn, & van Mourik, 2014) and which lacks a statutory audit history for small companies. Second, our study takes a more profound approach than earlier studies (e.g., Niemi et al., 2012) in examining the link between the voluntary audit decision and outsourcing accounting tasks. Considering the different professional qualifications of those to whom accounting tasks are outsourced, our study is the first to discuss the complementary versus substitutive role of auditing in a small company setting. Third, our study contributes by examining drivers of voluntary audit related to agency conflicts with owners (the legal form of the company, the existence of a supervisory board, and the status as a subsidiary) that have not yet been documented in prior research. By investigating the impact of nonmandated supervisory boards on voluntary audit decisions in small companies, we also add to the literature (on listed firms) discussing the complementary/substitutive relationship between governance mechanisms and external auditing.
The remainder of our paper is organized as follows. In Section 2 we provide a brief description of the institutional and regulatory setting (Germany). We discuss relevant prior literature and develop our hypotheses in Section 3, followed in Section 4 by a description of the data and models used in our empirical tests. Results from these tests are reported in Section 5. Thereafter, our paper concludes with a brief summary of the main findings and implications in Section 6.

| INSTITUTIONAL AND REGULATORY SETTING
Auditing regulation in the member states of the European Community (EC) or the EU has to comply with EC/EU directives. The Fourth Directive 78/660/EEC according to Article 51.1(a) in conjunction with Article 1 required limited liability companies to have their accounts audited by one or more persons authorized to carry out such audits.
However, according to Article 51.2, member states of the EC were permitted to provide an option enabling qualifying companies to forego the statutory audit. According to Article 11 in conjunction with Article 12, such qualifying companies were those that did not exceed two out of three size thresholds-only the maximum thresholds that member states could set were specified-regarding total assets, turnover, and the average number of employees in two consecutive financial years. In 2013 the Fourth Directive was replaced by Directive 2013/34/EU. This new directive according to Article 34.1 in conjunction with Article 1 requires limited liability companies that are medium-sized, large, or of public interest to have their accounts audited. As with requirements from the Fourth Directive, there could also be companies that do not need to have their accounts audited, therefore, unless they are of public interest or, according to Article 3.2 in conjunction with Article 3.10, exceed two out of three size thresholds regarding total assets, turnover, and the average number of employees in two consecutive financial years. As can be seen, depending on member states' interpretations of the directives, different settings can exist within the EU, first due to various definitions of the size criteria for small companies and second due to a possible option for qualifying companies to forego a statutory audit.
Currently, according to § 267 (1) of the German Commercial Code (HGB) companies are classified as small if they do not exceed two out of three size thresholds in two consecutive years (total assets: €6,000,000; turnover: €12,000,000; average number of employees: 50  (Collis, 2010) and a statutory audit requirement may be associated with significant costs as well as limited benefits for small companies (Directive 2013/34/EU), the audit exemption regime is of great economic importance.
Literature states that, owing to a higher demand for control mechanisms, the accounting profession developed earlier in countries where equity financing plays a major role, as opposed to countries where banks are the major providers of finance for business activities, and that the accounting profession's impact on local financial reporting practices is accordingly larger in countries where the accounting profession is well organized than in countries with smaller and later developed professional organizations (Alexander et al., 2014). Owing to its traditional emphasis on the importance of banks for business financing, Germany's accounting profession is not comparable with that in the UK (Haller, 2003;Nobes & Parker, 2016). 2 As opposed to the UK, the German auditing profession has a rather short tradition (Alexander et al., 2014;Haller, 2003) and is comparably weak (Nobes & Parker, 2016). In Germany, a statutory annual audit was first introduced for AGs and KGaAs only in 1931, whereas GmbHs have been subject to a mandatory audit regime only since 1985, when EU requirements were implemented in German law (Eierle, 2005). For GmbH & Co. KGs a statutory audit requirement was implemented only as late as 2000 (Eierle, 2005). Moreover, audit exemptions have always been very generous in Germany. Small companies in the legal form of the GmbH or the GmbH & Co. KG (the major legal forms for small and medium-sized firms with limited liability in Germany; see Table 1) have never been subject to a mandatory audit regime, and small AGs/KGaAs have been exempted from a mandatory audit for over 30 years (Eierle, 2005;Haller, 2003). Regarding the exemption criteria for small companies, Germany has always implemented the maximum size criteria possible under the EU accounting directives.
Accordingly, the audit exemption regime in Germany is very different from those covered by previous studies on voluntary audit, such as in the UK (e.g. Collis, 2012;Dedman et al., 2014) or Finland (Niemi et al., 2012;Ojala, Collis, Niemi, Kinnunen, & Troberg, 2016), where many small companies have only recently been relieved from the onus of a mandatory audit regime after an increase in size criteria.
As institutional theory emphasizes the role of habit on organizations' attempts to obtain stability and legitimacy (Oliver, 1991), firms' decisions to opt for voluntary audit may be driven by previous practices and resulting expectations from their stakeholders (Niemi et al., 2012). Furthermore, unlike the UK, Germany has always implemented the maximum size criteria possible under the EU accounting directives and has, compared with Finland (total assets: €100,000; turnover: €200,000; average number of employees: 3 [Ojala et al., 2016]), still considerably higher thresholds for small companies' audit exemptions.
In this respect, Germany provides a very different setting from that in the UK and Finland, for example, and is, moreover, free from the customary statutory audit history present in other countries.
Furthermore, unlike the UK's financial environment, accounting and taxation in Germany have traditionally been very strongly linked (Alexander et al., 2014;Schildbach, 2009), and consequently smaller companies usually prepare only one set of financial statements for financial accounting and tax purposes (Haller, 2003;Loitz, 2014).
Owing to this strong linkage, tax advisors play a crucial role in providing accounting services (Loitz, 2014). Accordingly, it is important to note that the audit and tax professions are strongly regulated in Germany, and to provide tax advisory services one must first pass a special state examination, § 35 StBerG (tax advisory act). In addition, financial statements prepared for tax purposes are subject to the tax enforcement regime, which is the case even if companies prepare only one set of financial statements for financial accounting and tax purposes. Since these factors might impact on a company's demand for voluntary audit, this is a further reason why Germany provides an important setting for studying voluntary audit.

| PREVIOUS LITERATURE AND HYPOTHESES DEVELOPMENT
The role of financial reporting differs between public and private companies. While public companies' financial reports serve the investment decision needs of financial markets, in private companies the main decisions relate to taxation and dividend distribution (Ball & Shivakumar, 2005). According to agency theory, the demand for auditing arises from information asymmetries and conflicts of interest between principals and agents (Jensen & Meckling, 1976;Watts & Zimmerman, 1986). Agency theory (Jensen & Meckling, 1976) suggests that audited financial reports play an important role in supporting relationships with investors as well as other principals who are distant from the actions of management and cannot, therefore, otherwise verify company information. In small companies, those principals include external shareholders, lenders, and suppliers (Power, 1997), who may also require audited accounts.

| Agency conflicts with owners
Conflicts of interest between owners (principals) and managers (agents), as well as information asymmetries arising from the separation of ownership and control, will give rise to agency costs (Jensen & Meckling, 1976;Watts & Zimmerman, 1986). Provision of financial reports can reduce these information asymmetries. However, owing to conflicts of interest, it cannot be guaranteed that management will report truthfully, and hence the need for an external audit is established (Gjesdal, 1981). For small businesses it can be argued that owners are often involved in management, thereby suggesting a lower level of agency costs resulting from the separation of ownership and control (Carsberg, Page, Sindall, & Waring, 1985;Collis et al., 2004;Collis & Jarvis, 2000); this in turn limits the need, therefore, for an independent assurance on financial statements (Directive 2013/34/EU). Nevertheless, one must be aware that small companies are heterogeneous in nature, and not all of them are owner-managed.
Thus, a complete absence of owner-related agency conflicts in small firms is very unlikely (Coppens & Peek, 2005;Page, 1984), which may, therefore, reveal a need for audited financial accounting information. In line with this argument, previous research analyzing small firms in the UK finds that a voluntary audit is more likely for small companies with shareholders who are not involved in management (Collis, 2010), have a higher number of nondirector shareholders (Seow, 2001), or lower managerial share ownership (Tauringana & Clarke, 2000). However, there are also studies that could not find a significant relationship between the demand for a voluntary audit and the existence of external owners without access to management information (e.g., Collis, 2010, for Danish companies) or which even found a significant negative relationship (e.g., Collis, 2012, for [non-micro] small companies in the UK). Based on our theoretical arguments herein and previous research findings from other countries, therefore, we propose the following hypothesis: H1a. Voluntary audit is more likely the higher the proportion of owners who are not involved in management.
Additionally, within the German context, Kaya (2010) argues that varying levels of information asymmetry and resulting agency conflicts between shareholders and management also depend on a company's legal form, which gives rise to differences in owners' information rights. In Germany, shareholders of GmbHs (and similarly GmbH & Co. KGs) have more comprehensive information rights than the shareholders of an AG (Kaya, 2010), which in the latter case may increase the demand for voluntary audit. This leads us to the following hypothesis: H1b. Voluntary audit is more likely for companies operating in the legal form of an AG.
Alternatively, discussion in prior literature regarding larger and listed companies looks at whether an external audit plays a complementary or substitutive role to internal corporate governance mechanisms (e.g., Hay, Knechel, & Ling, 2008;Knechel & Willekens, 2006;Zaman, Hudaib, & Haniffa, 2011). Since outside directors and audit committee members will be concerned about their personal exposure and financial loss that could result from litigation in the event of management fraud or some other scandal related to the organization, they may demand external assurance. In line with this argument, research covering listed companies finds a positive relationship between the strength of corporate governance mechanisms (such as the existence of an audit committee) and audit fees, thereby confirming the complementary role of an external audit (Carcello, Hermanson, Neal, & Riley Jr., 2002;Hay et al., 2008;Knechel & Willekens, 2006;Zaman et al., 2011). However, Knechel and Willekens (2006) argue that controls are complementary only as long as they are voluntary. Since mandated controls do not result from an endogenous demand for control, stakeholders will balance the externally imposed (nonefficient) requirement for internal control mechanisms against the endogenous demand for other forms of control (such as an external audit; Knechel & Willekens, 2006). In line with this argument, Knechel and Willekens (2006)  Prior research also suggests and finds for the UK that the number of owners impacts on agency costs and, therefore, increases the demand for a voluntary audit (Dedman et al., 2014). Companies with a larger number of owners are more likely to suffer problems of communication and coordination. This might lead to disputes among shareholders and result in more severe agency conflicts between the owners and management, which could be mitigated by an audit of financial statements. In contrast, if small companies are family owned, conflicts of interests between family members are assumed to be less pronounced due to a higher level of trust and consequently lower agency conflicts (Collis et al., 2004;Corten, Steijvers, & Lybaert, 2015). However, Corten et al. (2015) also argue that even in small and family-owned companies agency problems can exist because of an entrenchment effect (similar to Carey & Tanewski, 2013). In line with these conflicting arguments, empirical results from previous research are mixed. While Collis and coworkers (Collis, 2010;Collis et al., 2004) find for the UK that small companies that are wholly family owned reveal a lower demand for voluntary audit, this could not be confirmed by Collis (2012). In contrast to these findings, Corten et al. (2015) hypothesize and provide evidence for US private family firms (possibly because of entrenching) that the demand for reviews and compilations increases in the case of second-generation family firms.
Based on this discussion, we state the following hypotheses: H1f. Voluntary audit is more likely the greater the number of owners.

H1g. Voluntary audit is more/less likely if the company
is a family firm. Jensen and Meckling (1976) suggest that agency costs also arise from outside financing. Privately held companies do not have direct access to the capital markets, which makes them turn instead to debt financing (Berger & Udell, 1998). In basic terms, if a firm uses debt capital, managers acting in the interest of the firms' owners will have an incentive to undertake profitable business activities at the expense of outside investors (Jensen & Meckling, 1976). Because lenders anticipate such managerial behavior they price-protect themselves by, for instance, charging higher interest rates. Accordingly, the best solution for both parties (managers and lenders) is to engage an external auditor in order to reduce agency costs. This is especially relevant for small entities that are quoted in the literature as being more likely to suffer from credit rationing (Berger & Udell, 2006), partly because they are "often acutely informationally opaque" (Berger & Udell, 1998). In line with these arguments, previous literature provides evidence that small companies are more likely to demand a voluntary audit: (i) if financial statements are given to the bank (Collis, 2003); (ii) if banks require audited accounts (Collis, 2008(Collis, , 2012; (iii) if bank debt exists (Collis et al., 2004;Niemi et al., 2012); and (iv) if the company's leverage is high (Carey et al., 2000;Dedman et al., 2014;Tauringana & Clarke, 2000

| Importance of financial statements for management activities
Regardless of a company's size, managers have an interest in the integrity of financial information because of its support in facilitating better decisions. To reduce costs resulting from poor decisions, managers establish and maintain internal and external control systems to provide assurance regarding the integrity of financial information not only for external stakeholders but also for themselves (Jensen & Payne, 2003). Since small companies often lack professional accounting competence and control mechanisms (Collis et al., 2004), engaging an external auditor may reduce the risk of misstatements and thus ensure the correctness of financial statement information (Abdelkhalik, 1993). Thus, an external audit is expected to substitute for poor internal controls (Wallace, 1984). The more important managers perceive financial information to be in operating and controlling the activities of their companies, the more they might value an audit.

| Outsourcing
It is well known that the smallest companies cannot afford in-house accounting expertise and typically outsource their accounting function (Berry, Sweeting, & Goto, 2006;Everaert, Sarens, & Rommel, 2007;Kirby & King, 1997). Owing to limited resources and lack of competence regarding financial accounting tasks (e.g., bookkeeping, preparation of financial statements, or preparation of tax returns) in small companies (Gooderham, Tobiassen, Døving, & Nordhaug, 2004), these services are often outsourced to external experts (Niemi et al., 2012).
Germany's accounting profession is not comparable to that in the UK or the USA (Haller, 2003), with accounting tasks therefore being outsourced to those qualified either as auditors or tax advisors. Both professions are strongly regulated in Germany, implying, for instance, that every auditor/tax advisor has to pass a professional state exami-  Collis, 2012;Collis et al., 2004;Niemi et al., 2012).
Capital market-oriented companies, firms belonging to the accounting, auditing, or tax consultancy industry, and companies serving as general partners in limited liability partnerships were excluded from the sample. We conducted the survey in summer (June and July) 2013.
In total, we received back 443 questionnaires, 24 of which had to be rejected since respondents indicated that they were subject to statutory audit. One questionnaire was rejected due to the company's status as a general partner in a limited liability partnership, and a further 13 were excluded from the analysis due to contradictory information.
Ultimately, 405 usable responses were received, giving a response rate of 6.75%. The sample development is shown in Table 2. The validity of Number of companies randomly selected 2,000 2,000 2,000 Questionnaires sent out 6,000 Questionnaires returned 443 Rejected questionnaires, due to: • a statutory audit 24 • general partner in a limited liability partnership 1 • contradictory information 13 Usable responses 405 (6.75%) Usable responses by stratum 112 152 141 responses is supported by the fact that 94.3% of the questionnaires were completed by the manager or the firm's head of accounting.
Owing to the imputation of some missing variables, as is customary in survey research (Allee & Yohn, 2009), all of the 405 usable responses were featured in our analyses. 8 A multiple imputation procedure was applied that provided five data samples. The analyses (i.e., the descriptive statistics, correlations, and binary logistic regressions) were carried out for each of these data samples. The overall estimates are the arithmetic means of estimates for each of these data samples (Equation 1). The corresponding overall variances (Equation 4) are a combination of the average within-imputation variances (Equation 2) and the between-imputation variances (Equation 3), with (1 + 1/N) as an adjustment for finite N (Little & Rubin, 2002;Rubin, 1987).

| Models
To test our hypothesized relationships between the decision for voluntary audit and agency factors (resulting from agency conflicts with owners and lenders), the importance of financial statements for management activities, and outsourcing, we use the following logistic regression model (model 1): where In addition, as the variables BOARD and NONMANDBOARD are expected to be highly correlated, we run an additional model (model 2) to test H1d where we replace the variable BOARD by the variable NONMANDBOARD with all other variables remaining the same.
All variables used in the tests are described in   We augment both logistic regression models (model 1 and model 2) with five control variables. The first represents company size, which in previous literature is often used as a proxy for the separation of ownership and control (Chow, 1982;Tauringana & Clarke, 2000) or a measure of wealth value at risk (Abdel-khalik, 1993;Collis, 2010), measured by the natural logarithm of total assets (SIZE

| Descriptive statistics
Descriptive statistics and univariate analyses are reported in Table 4.
Means, medians, and standard deviations are shown for the total sample and two subsamples regarding the decision to have a voluntary audit (n = 49) or not (n = 356). The descriptive analysis shows that our sample covers a wide range of small companies with balance sheet totals ranging from €2,600 to €72,000,000 (with a mean value of €1,893,000 and a median of €1,000,000) and with less than five owners on average (median 2). Nine (2.2%) of the companies included in our analysis operate in the legal form of an AG and 396 (97.8%) in the legal form of a GmbH/GmbH & Co. KG. In our sample, the proportion of AGs is somewhat higher than in the total population of small companies (see Table 1). This can be explained by the stratified sampling technique, since companies choosing the legal form of an AG are usually also larger.

| Voluntary audit ratio
It is worthwhile emphasizing the extremely low proportion of small companies that opt for voluntary audit in our sample compared with voluntary audit ratios documented in studies from other countries.
While we find that only 12% of the companies examined (49 from 405) report having a voluntary audit, 10 the voluntary audit ratios found in previous research from other countries ranged from 26% to 80% (see Table 5). 11 The low audit ratio in Germany is noteworthy for two reasons.
First, in other studies (see Table 5) the size criteria of companies selected are considerably lower than those in this study. Taking into account research findings showing that size is one of the main drivers of the voluntary audit decision (e.g., Collis, 2012;Collis et al., 2004;Niemi et al., 2012), one would expect a much higher voluntary audit ratio for the companies in our study. Second, owing to our stratified sampling procedure, larger companies are overrepresented in our sample, which results in the voluntary audit ratio being even lower for the whole population of small companies in Germany. When analyzing the voluntary audit ratios for each stratum in our sample and weighing the results proportionally, we find an average audit ratio of 5.3% (see Table 6).
The considerably different results for Germany regarding the voluntary audit ratio show that findings from other countries regarding the decision to opt for voluntary audit cannot be transferred to Germany without reflection. One reason for the low audit ratio in Germany might be the absence of a mandatory audit history for small  While it can be seen that outsourcing to an external tax advisor Statistical (two-tailed) significance (p-values) better than .01, .05 and .1 indicated by ***, **, and *.

| Drivers of voluntary audit
For variable definitions, see Table 3. To examine the independence of the two subsamples (voluntary audit versus no voluntary audit) the chi-square test and the nonparametric Kolmogorov-Smirnov test were respectively performed in terms of the independent dichotomous categorical variables and the independent variables measured at least on an ordinal scale.    As all of the correlation coefficients are less than .7 (except for BOARD and NONMANDBOARD, which are, however, tested in two different models), there is no major overlap indicated in the independent variables' predictive power (multicollinearity) (Kervin, 1992).

| Logistic regression
The results of the two binary logistic regression models are shown in Tables 8 and 9 OUTSOURCE_TAXADV; and C: OUTSOURCE_AUD). Also reported are the variance inflation factors VIFs for the variables included in our binary logistic regression model as a further check for potential multicollinearity. As these VIFs are clearly below 10.0, we have no reason to suspect that results were affected by serious multicollinearity (Niemi et al., 2012;Ojala et al., 2016 increases, the greater the proportion of company owners who are not involved in management, which is in line with previous literature (Collis, 2010;Seow, 2001;Tauringana & Clarke, 2000). Extending previous research and consistent with findings from listed companies (Hay et al., 2008), we also find that voluntarily audited companies are more likely to be subsidiaries, confirming that the existence of a often rely on tax financial statements for their credit decisions, especially for smaller companies (Haller et al., 2008;Oehler, 2006) and may due to the strong alignment between financial accounting and taxation benefit from tax enforcement.  (Collis, 2010(Collis, , 2012Collis et al., 2004;Niemi et al., 2012).
Regarding H4a, the result for OUTSOURCE_EXP is only weakly significant (p < .10) with a negative sign, indicating that firms which outsource financial accounting tasks (such as bookkeeping, financial statement preparation, or tax return preparation) to an external expert (tax advisor and/or auditor) are less likely to opt for voluntary audit. This suggests that the involvement of an external expert provides additional assurance for the integrity of financial information and, therefore, substitutes for an external audit. When analyzing this substitution effect with more detailed differentiation between outsourcing to a tax advisor (H4b) or an auditor (H4c), we find for OUTSOURCE_TAXADV a strong significant negative association (p < .001) and for OUTSOURCE_AUD a strong significant positive association (p < .001). While the result for outsourcing accounting tasks to a tax advisor is in line with our prediction suggesting a substitution effect when an external expert is involved, the result for outsourcing accounting tasks to an auditor is in contradiction with our hypothesis.
If companies follow the independence requirements of German company law, auditors or other persons with whom they jointly exercise their profession are not allowed to perform an audit of the companies' financial statements if they are involved in bookkeeping or the preparation of these companies' financial statements (threat of self-review) according to § 319 (3) HGB. However, since we do not have access to the identities of individuals in our sample, we cannot exclude the possibility of this requirement being violated. Our finding for outsourcing accounting tasks to an auditor is in line with that from Niemi et al. (2012), who argue that problems of moral hazard will arise when accounting tasks are outsourced to an external auditor, thus inducing a demand for additional assurance. However, this argument would hold for outsourcing to a tax advisor as well as an auditor. An alternative explanation is that, for small companies, tax-related services are the most widely used business services provided (Leung, Raar, & Tangey, 2008), and owing to a tax advisor's specialization it is possible that they could have a competitive advantage over the auditor providing tax-related services (Niemi et al., 2012). Accordingly, when the manager of a small business has to decide whether to engage a tax advisor or an auditor, it is more likely that they will choose a tax advisor unless they value an auditor's more comprehensive professional qualification. This will be the case especially if they find financial statement information important for their management activities, as indicated by our descriptive statistics (Table 7). Thus, in line with the argumentation for listed companies on the complementary role of auditing (e.g., Hay et al., 2008;Knechel & Willekens, 2006;Zaman et al., 2011), one could expect that managers who value the additional professional qualification of an auditor when outsourcing accounting tasks also appreciate the additional assurance of an audit. In summary, regarding outsourcing as a driver of voluntary audit, our results support only our hypotheses H4a and H4b but not H4c.  Statistical (two-tailed) significance (p-values) better than .01, .05 and .1 indicated by ***, **, and *.
For variable definitions, see Table 3. and INDUSTRY_4) are weakly significant (p < .10) in only one of the three versions.

| Robustness check
Owing to our research method, we have to take into account a potential nonresponse bias. We test in two ways, therefore, to check whether or not the responsiveness of late respondents, as a proxy for nonrespondents (Lehman, 1963;Niemi et al., 2012;Wallace & Mellor, 1988), differs significantly from early respondents. In the first step we augment our binary logistic regression models with the variable RESPONSE_TIME, which measures the number of days elapsed before respondents returned the survey questionnaire. Results (untabulated) show that the regression coefficient RESPONSE_TIME is not significant in any model; moreover, other results are not qualitatively affected by this inclusion. In the second step, we separate our sample using RESPONSE_TIME from the median into early and late respondents; using a chi-square test (untabulated), we find no signifi-  Statistical (two-tailed) significance (p-values) better than .01, .05 and .1 indicated by ***, **, and *.
For variable definitions, see Table 3.
(untabulated) do not give grounds to suspect the influence of endogeneity to any significant extent as R 2 is considerably lower (.103, .100, and .110) and the regression coefficients that were previously significant are now insignificant (with the exception of NONOWNERMGT p < .10 and SIZE p < .10 or p < .05). We can conclude, therefore, that VOLAUDIT and IMPORTANCE are not driven by the same factors. Regarding the drivers of voluntary audit, in line with evidence from prior studies covering other countries we find that a higher proportion of company owners who are not involved in management increases the likelihood of an auditor being hired voluntarily. This finding is consistent with an interpretation that agency problems are present between the owners and outside managers of small companies.

| SUMMARY AND CONCLUSIONS
Extending previous research, we find that voluntarily audited companies are more likely to be subsidiaries, supporting evidence from listed companies that the existence of a major shareholder increases demand for external auditing services. Furthermore, our finding that companies in the legal form of an AG are more likely to opt for voluntary audit provides evidence that information asymmetries between managers and shareholders depend on a company's legal form. In contrast to prior research (Dedman et al., 2014), we do not find evidence that firms opt for voluntary audit of their financial statements if they have a larger number of owners.
Additionally, we cannot corroborate findings that the status as a family firm impacts on a firm's voluntary audit decision. This is in contrast to results from Collis et al. (2004) and Collis (2010), but in line with the later study of Collis (2012). Furthermore, extending prior research on listed firms, we cannot provide support for the argument that the general existence of either a supervisory board or a nonmandated supervisory board having an impact on the voluntary audit decision. Nor can we confirm the claim that leverage could be a voluntary audit driver among small companies in Germany. This finding is significant, in that it differs only slightly from that of Dedman et al. (2014), who from UK data find weak support for leverage as a voluntary audit driver, but is consistent with Niemi et al. (2012), who do not find support using Finnish data. Put together, these results suggest that in traditional bank-dominated economies, such as Germany and Finland, banks have different monitoring mechanisms for their small and private enterprise customers that either substitute for or are more cost efficient than a voluntary audit. However, in this context, the number of lenders might be another factor impacting the need for a voluntary audit. Literature suggests that borrowing from a single lender improves a lender's control (Petersen & Rajan, 1994), which in turn could decrease demand for an audit. As we do not have any data on the number of lenders used by our sample firms, we leave it for future research to examine in more detail the impact of a firm's relationship with its lenders on the demand for a voluntary audit. Consistent with other studies (Collis, 2010(Collis, , 2012Collis et al., 2004;Niemi et al., 2012) in the field of voluntary audit, we find that an audit could be interpreted as a check Our study provides three major contributions. The first lies in validation of earlier research findings from other jurisdictions in a very different institutional setting. In the absence of a statutory audit history for the companies considered, their decisions, unlike those in other studies, are not influenced by a previously mandated audit regime. Our second contribution is taking a more profound approach than earlier studies (Niemi et al., 2012) in examining the link between a voluntary audit decision and the outsourcing of accounting tasks.
Considering the different professional qualifications of those to whom the accounting tasks are outsourced, we add to existing literature regarding the complementary or substitutive role of auditing. As a third contribution, our study examines drivers of voluntary audit related to agency conflicts with owners (the legal form of the company, the existence of a supervisory board, and the status as a subsidiary) that have not been documented in prior research. Investigating the impact of nonmandated supervisory boards on voluntary audit decisions in small companies, we add to the literature (on listed firms) discussing the complementary/substitutive relationship between governance mechanisms and external auditing.
database possibly overestimate voluntary audit ratios (Collis, 2010;Collis et al., 2004;Dedman et al., 2014). As these studies use for their sample selection the amount of turnover-for instance, disclosed in companies' financial statements, which is a voluntary disclosure-they automatically eliminate the vast majority of small companies that do not opt to report such data. Since the voluntary disclosure is likely to be positively correlated with other types of voluntary behavior (such as a voluntary audit), the audit ratios derived might be biased (Collis, 2012) and they could, therefore, actually be lower.

Extract of translated questionnaire showing variables analyzed
3.4 Was the company included in a 2011 consolidated financial statement as a parent company or as a subsidiary? (A company is assumed to be a subsidiary if another company holds at least 50% of the voting shares.) ○ No, the company was not included in a consolidated financial statement ○ Yes, the company was included in a consolidated financial statement as a parent company.
○ Yes, the company was included in a consolidated financial statement as a subsidiary