Reducing Risk by Understanding the Interplay of Critical Audit Matters and Culture

Abstract

Prior research has called for the need to investigate how national cultural values can affect accountants’ interpretations of accounting standards. We focus on CAMs because they serve an important monitoring function, instilling credibility, trust, and confidence in the financial statements of emerging and international entities. Using a robust fixed effects regression on a sample of 1387 CAMs from companies who are listed on a US stock exchange with a non-US audit firm (2019–2021), we find that the number of CAMs reported is positively (negatively) associated with individualism and uncertainty (power distance). We also find a positive association between the strength of investor protections and the number of CAMs. Next, we find similar relationships between language (as proxied by the FOG Index) and the three measures of national culture. These findings contribute to the literature focusing on how the interplay of CAMs and culture can help investors manage investment risk.

1. Introduction

Auditing is an important component of the multi-layered structure of institutions, laws, and regulations designed to facilitate well-functioning capital markets. External auditors play an important role in capital markets by providing an important governance mechanism (Smith et al., 2018), but the only outcome of the external audit process that is observed by investors is the audit report. The Public Company Accounting Oversight Board (PCAOB) was established by Congress in the Sarbanes–Oxley Act of 2002 to “protect investors and further the public interest in the preparation of informative, accurate, and independent audit reports” (PCAOB, 2024). The inclusion of critical audit matters (CAMs) in an auditor’s report is the most significant change to the audit report in more than 70 years, a change that has the potential to lead to “better auditing and/or to better communication between the auditor and financial statement users” (Carcello, 2014, p. 73). A CAM is “any matter arising from the audit of the financial statements that was communicated or required to be communicated to the audit committee and that: (1) relates to accounts or disclosures that are material to the financial statements and (2) involved especially challenging, subjective, or complex auditor judgment” (PCAOB, 2017). The focus on “challenging, subjective, or complex auditor judgment” makes CAMs distinct from “Key audit matters” [KAMs], which are defined by the International Audit Standards as “those matters that, in the auditor’s professional judgement, were of most significance in the audit of the financial statements of the current period” (ISA 701, IAASB, 2015). The inclusion of CAMs took effect for large-accelerated filers with the fiscal year ending on or after 30 June 2019, and for the audits of other filers with the fiscal year ending on or after 15 December 2020. The reduction in firm risk leads to a higher stock value and a boost in confidence among managers and employees (Abraham et al., 2024). Auditors who increase transparency by clearly communicating complex situations in CAMs reduce firm risk. This study focuses on how understanding the interplay of CAMs and culture can increase transparency in CAMs, which, in turn, helps investors better understand firm risk.

The inclusion of CAMs is designed to improve communication between an auditor and financial statement users (Carcello, 2014) by increasing transparency. The inclusion of CAMs allows auditors to exercise substantially more professional judgment during the financial reporting process. Gao and Zhang (2019) defined an auditor’s professional judgment as the ability to assess the audit risk and audit resources accordingly. Prior research has shown that an auditor’s judgment depends not only on knowledge, experience, and training, but is also a function of the national culture (Patel, 2003; Doupnik & Riccio, 2006; Gierusz et al., 2022). Still, other research has concluded that “audits are conducted differently in different countries” (Eierle et al., 2021, p. 303). It is, therefore, relevant to analyze how culture influences the likelihood of CAMs disclosures, as well as the readability of such disclosures. According to Cowperthwaite (2010), “Every culture develops in response to a unique set of circumstances; every culture is unique just as each person has a unique personality” (p. 178). We focus on CAMs in audit reports because they serve an important monitoring function, instilling credibility, trust, and confidence in the financial statements of emerging and international entities. The basic questions addressed in this study are as follows: Do differences in culture cause auditors in different countries to disclose more/less information through CAMs? Do differences in culture cause auditors in different countries to affect the readability of CAMs? We are the first to attempt to answer these questions in terms of cross-listed firms on US stock exchanges. We are aware of one other paper which has explored the possible differences in the reporting of audit matters across jurisdictions. Using two sets of Canadian firms (domestic firms which use KAMs and US cross-listed Canadian firms that report CAMs), Liu et al. (2024) found no difference in the value relevance or the ability to predict future cash flows. While both studies examine companies who are listed on a US stock exchange with a non-US audit firm when CAMs reporting was voluntary, 2019–2021, we use a sample which encompasses more countries and, specifically, we examine how culture and readability (as opposed to the length of the audit report and litigation risk, as in Liu et al. (2024)) are associated with the quality of the disclosure of CAMs. We find substantial support for the notion that culture affects the clarity of communications between auditors and the users of financial statements through the number of disclosures and the readability of the disclosures. Differences in culture can lead to differences in disclosure decisions. These findings should be of significant interest to the PCAOB as it broadens its oversight across international borders and sets policy for non-US companies.

Our first research question deals with the number of CAMs and Hofstede’s cultural indices. We are aiming to confirm that there is an established relationship based on culture in a new setting, using companies that are listed on a US stock exchange (New York or NYSE, American Stock Exchange, or Nasdaq) with a non-US audit firm during the voluntary adoption period or the first year of adoption. We use a unique sample of 1387 CAMs from companies who are listed on a US stock exchange with a non-US audit firm (2019–2021) drawn from Audit Analytics data and Compustat. The use of this type of dataset strengthens our empirical contribution to the literature. Using a fixed effects regression model, we examine the association between culture and CAMs for international firms that operated in the US during the initial years of the adoption of CAMs (the period of voluntary adoption and the first year of required adoption). We believe this period will best demonstrate how culture influences CAMs, since the audit process and the final product (as well as the individual traits of the audit firm) would be affected by the process of continuing to adapt to a new set of rules. We proxy for culture using Hofstede’s cultural indices and investor protections to study how culture influences the number and readability of CAMs.

A few studies have examined the association between Hofstede’s cultural indices and KAMs. There is, however, no comprehensive review of the impact of Hofstede’s framework on, or how it may influence the determinants of, CAMs disclosures for international firms listed on US stock exchanges. Kitiwong and Srijunpetch (2019) examined the association of Hofstede’s measures of masculinity and uncertainty avoidance with the number of KAMs disclosed. Examining a sample of industrial firms in Thailand, Malaysia, and Singapore over the 2016–2018 period, they found no association between the number of KAMs and Hofstede’s characteristics. Federsel and Hörner (2023) conducted a similar analysis (2017–2021 data) of the then 28 European Union countries. They included individualism, uncertainty avoidance, and power distance in their models, conducting a principal components analysis that combined these variables with “General Trust”, “Ethics”, and “Indulgence”. They found that the resulting factor was negatively associated with the number of KAMs.

Following the significance of individualism, uncertainty avoidance, and power distance in the auditing literature, we examine the interplay of culture and CAMs for cross-listed non-US firms that employ a foreign audit firm using the three cultural indices that the auditing profession and the literature has found to be relevant: 1. authority and power distance (which examines how inequality is dealt with); 2. uncertainty avoidance (which examines how society deals with uncertainty); and 3. individualism (which examines how individuals in a society relate to individuals both in and outside of their culture) (Cowperthwaite, 2010). CAMs in an auditor’s report are designed to provide incremental information in sufficient detail about accounts or disclosures that are material to the financial statements and involve especially challenging, subjective, or complex auditor judgment.

We find a positive association between auditors from countries with high individualism and countries with high uncertainty avoidance and the number of CAMs. We find a negative association between power distance and the number of CAMs. By understanding the interplay between cultures with differences in culture values, we add to the discussion of how auditors approach auditing under a new common set of requirements (in this case, CAMs) that increase transparency and clarity in communication, which, in turn, helps investors better understand firm risk.

Another way culture is manifested in a country is through its willingness to protect investors. Auditors’ incentives to detect insider expropriates differ across countries and are related to a country’s formal institutions. For example, higher auditor penalties and insider penalties lead to an increase in total investment levels (Newman et al., 2005). Leuz et al. (2003) found that strong investor protection affects financial reporting practices by reducing earnings management. Thus, we explore how the level of investor protection affects CAMs disclosures. J. R. Francis and Wang (2008) showed that countries with strong investor protection regimens have a greater level of financial transparency. We add to the literature by examining how the interplay between cultures (using the cultural classifications of Hofstede (1980, 2001) as well as investor protections) improves transparency through the use of CAMs. We find that the strength of investor protection is positively associated with the number of CAMs disclosed.

Finally, we add to the literature by examining the interplay of language and culture. Language reflects the values and beliefs of a culture. Hall and Oberg (1958) emphasized the need to understand the cultural context to be able to communicate properly within a culture. Thus, in order to effectively and efficiently understand annual report language, the reader must understand the culture in which that language was and is developed. How easy it is to understand the language of CAMs prompted us to question whether the readability of CAMs is also associated with culture; however, little research has been conducted to explore the relationship between readability and culture. Using the FOG index, we find that aspects of culture are predictably associated with CAMs, i.e., individualism and uncertainty avoidance are positively associated with CAMs’ readability, power distance is negatively associated with CAMs’ readability, while investor protection is positively associated with the readability of CAMs.

The introduction of CAMs gives auditors an opportunity to convey more clearly how material accounts or disclosures that are “especially challenging, subjective, or (require) complex auditor judgment” (PCAOB, 2017) are handled by their firm. This is especially important for international firms that are doing business in a highly litigious environment such as the US (Cowperthwaite, 2010). These findings should be of significance to the PCAOB as it sets policy for non-US-based companies that trade on US securities exchanges.

The next section provides a survey of the relevant literature and formulates our hypotheses. Section 3 describes our data sources, descriptive statistics, and models. Section 4 reports our results concerning the association between our proxies of culture in connection with the number of CAMs and the readability of CAMs. The last section summarizes and concludes our study.

2. Review of the Literature and Hypotheses Development

We examine how language (number of CAMs and readability) interacts with culture (cultural indices and the legal environment) to affect the audit report, in particular, the CAMs. There is, however, no comprehensive review of the impact of Hofstede’s framework on whether or how it may influence the determinants of CAMs disclosures among foreign firms who cross-list on American stock exchanges. Nonetheless, following Federsel and Hörner (2023) and many others in the auditing literature, we use Hofstede’s indices of individualism, uncertainty avoidance, and power distance to study how the culture of the home country of the company influences the number of CAMs.

2.1. Number of CAMs and the Role of Culture

Cultural norms and beliefs are influential forces that affect people’s perceptions, dispositions, and behaviors. As such, national culture has long been recognized as a critical environmental characteristic that underlies systematic differences in management behavior. The cultural classifications of Hofstede (1980, 2001) represent the most influential national cultural framework in the business literature. Hofstede’s framework has inspired many empirical studies (Kirkman et al., 2006), including studies concerning national culture and accounting theory. Many previous studies have used Hofstede’s (1980, 2001) indices of cultural values to categorize national culture along the following dimensions: individualism (IND), power distance (PD), uncertainty avoidance (UA), masculinity, and long-term orientation. The auditing literature has focused on individualism (IND), power distance (PD), and uncertainty avoidance (UA). Cowperthwaite (2010) discussed the influence of national culture on the implementation of international accounting standards and identified these three dimensions of Hofstede’s cultural variables as most relevant to the auditing profession.

Individualism refers to the tendency to prioritize oneself and those closest to oneself (e.g., family). As Hofstede explained, “people in an individualistic society are more like atoms flying around in a gas while those in collectivist societies are more like atoms fixed in a crystal” (Hofstede, 2024). Individualism is one extreme along a continuum, with the opposite characteristic being collectivism. Collectivistic societies prioritize the in-group over individuals within a group, or those outside of the in-group. Highly individualistic cultures emphasize individual achievement, self-orientation, and autonomy (Hofstede, 2001). While CAMs provide auditors with the opportunity to communicate to stakeholders regarding the issues that arose during the audit and which required challenging, subjective, or complex judgment, the prior literature is mixed on the interpretation of the information contained in CAMs.

On the one hand, CAMs disclosures are associated with reduced auditor liability, since the disclosure forewarns users of reporting risks and protects auditors (Brasel et al., 2016; Kachelmeier et al., 2020). On the other hand, Christensen et al. (2014) found that nonprofessional investors are more likely to change their investment decisions because of the mere presence of a CAM. The CAM paragraph effect on investment is, however, reduced when a CAM is followed by a paragraph offering its resolution. Klevak et al. (2023) found that firms with more extensive CAMs are associated with perceived uncertainty, more volatile stock prices, and dispersed analyst forecasts. This is consistent with market misinterpretation. This might occur because when the related CAMs are disclosed, jurors could conclude that the auditor was aware of the risk and should have foreseen a negative outcome (Gimbar et al., 2016a). Similarly, any unrelated CAMs may trigger jurors to doubt the quality of the audit because the auditor detected an item that was different from the issue ultimately associated with the litigation (Gimbar et al., 2016b). CAMs thus might increase the potential for litigation against auditors.

We posit that auditors within a highly individualistic society might have the incentive to report a greater number of CAMs to reduce their liability risk. In countries with high individualism, laws and rights are the same for all, which means that the same standards need to be used consistently when assessing any individual actions. Therefore, societies with high individualism will be less tolerant of corruption and accounting errors (Davis & Ruhe, 2003; Caban-Garcia et al., 2017), given their emphasis on individual achievement, self-orientation, and autonomy. In highly individualistic societies, there is a belief in individual decisions, as opposed to collective decisions, so that auditors feel responsible for their decisions instead of the company being responsible for the decisions of the individual auditor (Cowperthwaite, 2010). We propose the following hypothesis in alternative form:

H1: 

The strength of individualism is positively associated with the number of CAMs disclosed.

Uncertainty avoidance refers to a preference (or distaste) for taking risks versus knowing and planning future events and exerting control over those future events. “Uncertainty avoidance … has to do with anxiety and distrust in the face of the unknown, and conversely, with a wish to have fixed habits and rituals, and to know the truth” (Hofstede, 2024). Rigid institutions and established social norms evolve in high UA countries to alleviate uncertainty through law and formal rules (Bik & Hooghiemstra, 2017). In societies with high uncertainty avoidance, individuals tend to want to avoid risk and unpredictability. As a result, work environments in such societies will try to provide stability and certainty through clear rules and instructions. More precise rules and instructions influence the auditor’s role and behavior, since auditors can be shielded from criticism and litigation if they followed the rules (Pinto et al., 2020). In cultures with a high uncertainty avoidance index, there is a fear of personal failure and a strong affinity for technological solutions as opposed to innovative solutions. Auditors in cultures with a high uncertainty avoidance index gravitate to more rules-based systems to reduce ambiguity. They seek clarity and structure (Cowperthwaite, 2010). Based on this desire to reduce ambiguity, we expect that societies with high UA will be associated with a greater number of CAMs being disclosed. This leads to our second hypothesis (in alternative form):

H2: 

The strength of uncertainty avoidance is positively associated with the number of CAMs disclosed.

Power distance refers to the tendency to accept an unequal sharing of power and responsibility within a society by the individuals within the society. “Power Distance is the extent to which the less powerful members of organizations and institutions (like the family) accept and expect that power is distributed unequally” (Hofstede, 2024). Countries with high power distance scores tend to be more autocratic, with less questioning of the actions of authorities or those holding power. In societies with high power distance, people are more likely to view power inequality as good and acceptable. As a result, auditors in high power distance countries are more likely to be secretive and less transparent, resulting in less information sharing (Hope et al., 2008). They may also be more likely to hire employees who are relatively less educated and therefore, less likely to question managers’ directions and judgments. This could result in a less effective internal control system, more material accounting errors (Chen et al., 2007), and more unresolved issues during an audit. According to Lewellyn (2017), in high power distance cultures, individuals expect to receive instructions from the authorities. Therefore, in these societies, individuals are unlikely to deviate from the instructions of their superiors. Innovations would need support from supervisors who rely on formal rules. The need to have approval from supervisors leads to less disclosures from subordinates (Cowperthwaite, 2010). Thus, the characteristics of high power distance cultures leads to our third hypothesis (in alternative form):

H3: 

The strength of power distance is negatively associated with the number of CAMs disclosed.

Finally, we control for culture by examining the resultant legal environments through investor protections. Prior research has found a compelling link between strong investor protections and financial transparency (J. R. Francis & Wang, 2008). Strong investor protections often reduce earnings management (Leuz et al., 2003) and lead to an increase in investment levels (Newman et al., 2005). With respect to CAMs, any variation may be due to the culture, the result of a culture attempting to apply the PCAOB rules formed in another culture, or the accounting quality that various cultural indices proxy for. Since country characteristics play an important role in financial transparency, we employ three country-level metrics of investor protections that have been used in prior research: the law system, rule of law, and control of corruption. According to J. R. Francis and Wang (2008), countries with a common-law legal tradition tend to have stronger investor protections through corporate law and securities laws, characteristics that have been correlated with audit quality. The RULE OF LAW INDEX is a measure of investor protection created by Kaufmann et al. (2011) that computes agents’ perceptions of how much they trust and follow the laws of society, specifically the effectiveness of the enforcement of contracts, property rights, the police, and the courts, as well as the likelihood of crime and violence. This measure is a proxy that captures the perceptions of the extent to which public power is used for personal gain, including both small-scale and large-scale types of corruption.

We expect that auditors in countries with strong investor protections will disclose a greater number of CAMs because the stronger protections reduce the incentives to mask firm performance. This leads to our fourth hypothesis (in alternative form):

H4: 

The strength of investor protections is positively associated with the number of CAMs disclosed.

2.2. Readability of CAMs and the Role of Culture

Next, we examine how the language used by an auditor to inform stakeholders about the challenges or complexity of the audit process differs across jurisdictions due to national culture. As with the number of CAMs, the complexity of the CAMs should be of significant interest to the PCAOB as it broadens its oversight across international borders.

The influence of language is everywhere and provides the source for representing human beliefs and values. Several studies have shown that annual report texts are written in language too difficult for most readers to understand (David, 2001; Conaway & Wardrope, 2010). Therefore, annual reports that cross cultural boundaries might increase the difficulty readers encounter in understanding annual report language. Foreign investors might find annual reports more difficult to analyze than would same-country nationals.

Rather than measuring the length of the audit report or the number of CAMs disclosed, we employ Gunning’s Fog Index to examine the clarity and honesty of annual reports. The index, developed in 1944 by Robert Gunning, uses the quantifiable aspects of a text to determine its grade level. According to Herman (2000), most people are comfortable reading two grades below their highest attained academic grade. Therefore, with the average accountant having an undergraduate degree, financial reports should aim for a Fog Index of 14 (our sample of audit reports has a mean of 21.77).

CAMs provide an opportunity for auditors to inform investors and other financial statement users of matters arising during the audit that required especially challenging, subjective, or complex auditor judgment. A PCAOB investor survey report on the initial impact of CAMs (conducted in April and May of 2022), showed that 44% of the respondents thought that the CAMs were easy to understand and provided sufficient detail about an audit. Therefore, the readability of the CAMs in an auditor’s report is important because it provides incremental information about the risks associated with a given company, which, in turn, can influence the auditor’s litigation risk. Prior research has found that in the US, statewide “cultural tightness and looseness” affects annual report readability (Noh, 2021). However, little research has been performed to explore the determinants of the readability of CAMs and, especially, how culture affects CAMs disclosures.

Highly individualistic societies are less inclined toward corruption and accounting errors. The results obtained by Caban-Garcia et al. (2017) suggest that managers from countries with high individualism are more likely to implement controls that prevent corruption and tax evasion and are, therefore, less likely to report internal control weaknesses. We expect, therefore, that companies in highly individualistic societies will want CAMs that are easy to decipher, i.e., short sentences and minimal auditing jargon, so as to help investors understand the matters that require special attention and to reduce litigation risk. Increased readability leads to an increase in transparency, i.e., more people understand what you are trying to communicate. Less jargon coupled with a belief that earnings are more important than interesting, work in a culture with a high individualism index to promote greater social mobility across occupations (Cowperthwaite, 2010). This leads to our fifth hypothesis (in alternative form):

H5: 

The strength of individualism is positively associated with the readability of CAMs,1 i.e., negatively associated with the Fog Index.

High uncertainty avoidance societies tend to have higher levels of anxiety. They endeavor to avoid risk and unpredictability by adhering to precise rules and instructions. Auditors in a culture with a high uncertainty avoidance index resist innovation and are more attracted to rules-based systems to reduce ambiguity (Cowperthwaite, 2010). Given that investors will attempt to search for any material information the auditor might have overlooked during the auditing process, companies in high UA societies will require auditors to reduce litigation risk by providing lengthy and complex CAMs disclosures. This leads to our sixth hypothesis (in alternative form):

H6: 

The strength of uncertainty avoidance is negatively associated with the readability of CAMs (i.e., positively associated with the Fog Index).

Consistent with the view that high power distance societies rely on formal codes of conduct, we would expect companies in high power distance cultures to issue reports that are easy to understand in order to reduce scrutiny from the PCAOB. “Audit teams are likely to have a centralized decision-making structure in which the engagement partner is an autocratic team leader. Subordinates expect to be told what to do” (Cowperthwaite, 2010). Mangers provide authoritative leadership that relies on formal rules. There is a concentration of authority. It is readily accepted that power is distributed unequally (Hofstede, 2024). Zarzeski (1996) found that high power distance results in more firm-level disclosures. This leads to our seventh hypothesis (in alternative form):

H7: 

The strength of power distance is positively associated with the readability of CAMs (i.e., negatively associated with the Fog Index).

Prior research has examined how the readability levels of Management Discussion and Analysis (MD&A) and earnings press releases differ across countries due to linguistic distance, accounting distance, and investor protection distance (Lundholm et al., 2014). Lundholm et al. (2014) found that MD&A and press release disclosure readability increase when the similarity between the foreign country’s legal system and the US legal system are closer. We focus on the effect of investor protection on the textual disclosure of CAMs, which inform investors and other financial statement users of matters from the audit that required especially challenging, subjective, or complex auditor judgment in a clear, precise, business-like language. This leads to the following hypothesis (in alternative form):

H8: 

The strength of investor protection is positively associated with the readability of text in CAMs reports, i.e., negatively associated with the Fog Index.

3. Sample, Descriptive Statistics, and Models

3.1. Sample and Descriptive Statistics

We examine how language (number of CAMs and readability) interacts with culture (cultural indices and the legal environment) to affect the audit report, in particular, CAMs. Our primary analysis was conducted for the years 2019–2021 for foreign firms that cross-listed on the New York or NYSE, American Stock Exchange, or on Nasdaq. We included all firms that cross-listed during a fiscal year included in the time period of voluntary CAMs disclosure. We examine the period from the PCAOB’s adoption of AS 3101 through the time that CAMs disclosure became mandatory for all firms.2 Our data thus include critical audit matters that were issued voluntarily prior to the effective dates. By studying the influence of culture on CAMs using different individual country attributes, we are able to observe the confluence of country-level culture on individual decisions and the market forces within a country that are themselves influenced by culture. We focus on cross-listed foreign firms in US market to study the how the unique culture of the country in which the issuer is headquartered adapts to a US exchange that requires the issuer and auditor to conduct financial reporting and the audit process in a manner that more closely reflects Western culture. We posit that the variations in CAMs are due to both a country’s culture and the interaction of individuals within that culture attempting to apply PCAOB rules that were formed in a different culture.3 By studying difference aspects of culture, we avoid problems that arise from studying single institutional aspects of culture, where co-dependencies between individual country attributes might erroneously yield significant results (Federsel & Hörner, 2023).

A non-US company that seeks to list its securities on the New York or NYSE, American Stock Exchange, or on Nasdaq must register its securities with the SEC by filing an Exchange Act registration statement and must subsequently file annual reports that have been audited in accordance with the standards of the PCAOB. The PCAOB has inspected non-US registered firms since 2005 and has often entered into “formal cooperative arrangements with foreign audit regulators in order to minimize administrative burdens and potential legal or other conflicts that non-US firms may face in the foreign jurisdiction in question” (PCAOB, 2023). Despite this, the board was not able to secure access to inspect auditors in the People’s Republic of China until 15 December 2022 (PCAOB, 2022). One significant recent development in the US that has impacted foreign issuers is the adoption by the PCAOB, in 2017, of AS 3101: The Auditor’s Report on an Audit of Financial Statements When the Auditor Expresses an Unqualified Opinion. This standard expands the existing auditor’s report by including the communication of CAMs and the clarification of the auditor’s role and responsibilities (Release 2017-001). Early studies on the determinants of CAMs disclosures have suggested that the characteristics of both audit clients and audit firms play an important role in determining the number, sentiment, and readability of CAMs disclosures (Burke et al., 2023).

Cowperthwaite (2010) offered insight into how national culture influences the implementation of global standards. He focused on three of Hofstede’s cultural dimensions that he felt were very relevant to the auditing profession: individualism, uncertainty avoidance, and power distance. Kitiwong and Srijunpetch (2019) examined the association of Hofstede’s measures of masculinity and uncertainty avoidance with the number of KAMs disclosed, and found no association between the number of KAMs and Hofstede’s characteristics. Federsel and Hörner (2023) conducted a similar analysis (with 2017–2021 data) using individualism, uncertainty avoidance, and power distance in their models, and found that the resulting factor was negatively associated with the number of KAMs.

There is, however, no comprehensive review of the impact of Hofstede’s framework on whether or how it may influence the determinants of CAMs disclosures when using cross-listed firms on US stock exchanges. Studying how firms implement accounting standards, and the influence of national cultures on this implementation, has long been a topic in the auditing literature. As Cowperthwaite (2010) posited, “Culture is to society as personality is to the individual… These (cultural) values are hard-wired into the individual’s personality for life.” (p. 176). Following Federsel and Hörner (2023), we use Hofstede’s indices of individualism, uncertainty avoidance, and power distance to study how the culture of the home country of a company influences the number of CAMs.

We rely on Audit Analytics for companies’ CAMs disclosures (19,544). Next, we merge the sample with firms provided by Compustat and eliminate those that are missing financial data or other needed variables. This leaves us with 9816 CAMs. The financial data are from Compustat, and the cultural variables are from Hofstede (2001). The final sample for the main international analysis includes 1387 non-US observations.

Table 1. (A): Sample selection. (B): Sample distribution by year. (C): Sample distribution by country.

(A)
			Obs
CAMs Reported by Audit Analytics			19,544
(−) Merge with Financial Data			(9342)
(−) Missing Financial Data or Variables			(386)
Total			9816
International			1387
USA			8429
Total			9816
(B)
Year	Number of CAMs		%
2019	343		24.7%
2020	604		43.5%
2021	440		31.7%
Total	1387		100.0%
(C)
Country		Freq	%
ARGENTINA		41	2.96%
AUSTRALIA		9	0.65%
BELGIUM		11	0.79%
BERMUDA		14	1.01%
BRAZIL		68	4.90%
CANADA		419	30.21%
CHILE		21	1.51%
CHINA		177	12.76%
COLOMBIA		10	0.72%
DENMARK		12	0.87%
FINLAND		3	0.22%
FRANCE		29	2.09%
GERMANY		25	1.80%
GREECE		28	2.02%
INDIA		24	1.73%
INDONESIA		2	0.14%
ISRAEL		140	10.09%
JAPAN		20	1.44%
KOREA (SOUTH)		21	1.51%
LUXEMBOURG		12	0.87%
MALAYSIA		1	0.07%
MEXICO		28	2.02%
NETHERLANDS		29	2.09%
NORWAY		14	1.01%
PANAMA		5	0.36%
PERU		7	0.50%
PHILIPPINES		3	0.22%
RUSSIA		8	0.58%
SINGAPORE		15	1.08%
SOUTH AFRICA		17	1.23%
SPAIN		14	1.01%
SWEDEN		12	0.87%
SWITZERLAND		26	1.87%
TAIWAN (CHINA)		17	1.23%
THAILAND		2	0.14%
TURKEY		2	0.14%
UNITED KINGDOM		101	7.28%
Total		1387	100%

Note: Total Audit Analytics CAMs between 30 June 2019 and 30 April 2022.

, Panel A, shows the sample selection process.

Table 1, Panel B shows the yearly distribution of the sample. The lowest percentage of observations occurred in 2019 (24.7%), and the highest in 2020 (43.5%). Table 1, Panel C shows the distribution of the sample observations for the 37 non-US countries in our sample. It is not surprising that Canada has the highest percentage (30.2%), since there are more Canadian firms listed on US stock exchanges than firms from other countries. Canada is followed by China (12.7%) and Israel (10.0%).

Table 2. Sample distribution by type of CAMs.

Type of CAMs	Freq	%
Financial statements and disclosures	1	0.07%
Fresh start accounting	1	0.07%
Related party transactions	1	0.07%
Short-term investments	1	0.07%
Discontinued operations	2	0.14%
Interest revenue	2	0.14%
Depreciation and amortization	3	0.22%
Other expenses	3	0.22%
Other liabilities and provisions	3	0.22%
Policy changes	3	0.22%
Deferred and capitalized costs	4	0.29%
Vendor/supplier rebates	4	0.29%
Disposals and divestitures	5	0.36%
Real estate investments	5	0.36%
Other revenue	6	0.43%
Deferred and stock-based compensation	8	0.58%
Derivatives and hedging	8	0.58%
Leases	8	0.58%
Regulatory assets and liabilities	8	0.58%
Other assets	9	0.65%
Other income taxes	13	0.94%
Warranty liabilities	13	0.94%
Asset retirement and environmental obligations	14	1.01%
Pension and other post-employment benefits	15	1.08%
Other debt	17	1.23%
Accounts/loans receivable	18	1.30%
Sales return and allowances	21	1.51%
Insurance contract liabilities	23	1.66%
Going concern	24	1.73%
Research and development expenses	26	1.87%
Allowance for credit losses	28	2.02%
Equity investments and joint ventures	37	2.67%
Goodwill and intangible assets	41	2.96%
Other intangible assets	45	3.24%
Proven and unproven reserves	47	3.39%
Inventory	49	3.53%
Deferred income taxes	58	4.18%
Other contingent liabilities	60	4.33%
Uncertain tax positions	62	4.47%
Other investments	67	4.83%
Business combinations	81	5.84%
Long-lived assets	87	6.27%
Property, plant, and equipment	127	9.16%
Goodwill	135	9.73%
Revenue from customer contracts	194	14.0%
Total	1387	100%

provides information on the type of CAMs disclosed. The most common types of CAMs relate to property, plant, and equipment (9.16%); Goodwill (9.73%); and revenue from customer contracts (14.0%).

Table 3. Sample descriptive statistics.

Variable	Mean	S.D.	0.25	Mdn	0.75
NUMCAM	2.22	1.61	1.00	2.00	3.00
IND	48.33	20.96	38.00	39.00	68.00
PD	57.67	24.80	38.00	67.00	80.00
UA	56.79	21.52	44.00	48.00	81.00
FOG	21.77	3.53	19.33	21.33	23.68
COMMON	0.39	0.49	0.00	0.00	1.00
RULE OF LAW	0.98	0.83	−0.03	1.43	1.65
CONTROL OF CORRUPTION	0.97	0.90	−0.05	1.49	1.66
ICMW	0.10	0.30	0.00	0.00	0.00
RESTATE	0.11	0.32	0.00	0.00	0.00
AUCHANGE	0.04	0.19	0.00	0.00	0.00
AUTENURE	1.73	1.25	0.69	1.79	2.77
BIG4	0.80	0.40	1.00	1.00	1.00
SIZE	8.37	2.58	6.82	8.48	10.20
AGE	2.77	0.65	2.30	3.00	3.30
BM	0.73	0.78	0.24	0.51	1.01
SALESGROWTH	0.04	0.20	−0.02	0.01	0.08
SEGMENTS	1.23	0.58	0.69	1.10	1.79
FOREIGNSALES	0.48	0.44	0.00	0.53	0.97
LEV	0.22	0.21	0.03	0.19	0.37
ROA	−0.04	0.24	−0.04	0.01	0.05
LOSS	0.37	0.48	0.00	0.00	1.00
ZSCORE	3.15	7.56	0.55	1.68	3.66
LITRISK	0.30	0.46	0.00	0.00	1.00
MA	0.27	0.44	0.00	0.00	1.00
RESTRUCTURE	0.25	0.43	0.00	0.00	1.00
MKTCAP	1.08	0.55	0.63	0.97	1.60
DISCLOSURE	66.45	8.77	60.93	65.00	74.00

presents descriptive statistics for the dependent and independent variables. Auditors report, on average, 2.22 CAMs per firm year. The mean values for individualism (IND), uncertainty avoidance (UA), and power distance (PD) scores are 48.33, 57.67, and 56.79, respectively. The average number of internal control material weaknesses (ICMWs) is 0.10. In all, 11% of the samples report a financial restatement, and 4% an auditor change. On average, the auditor tenure for our sample firms is 1.73 years. On average, the firms appear to be in good financial health with an average Zscore of 3.15. An average firm has 1.23 segments and the average rate of sales growth in the sample period is 4%.

3.2. Models

Our test models for these hypotheses are as follows:

$$\begin{array}{c}CAMDISCLOSUREQUALITY\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}={\beta }_0+{\beta }_1{CULTURE}_{i,t}+{\beta }_2{ICMW}_{i,t}+{\beta }_3{RESTATE}_{i,t}+{\beta }_4{AUCHANGE}_{i,t}+{\beta }_5{AUTENURE}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_6{BIG4}_{i,t}+{\beta }_7{SIZE}_{i,t}+{\beta }_8{AGE}_{i,t}+{\beta }_9{BM}_{i,t}+{\beta }_{10}{SALESGROWTH}_{i,t}+{\beta }_{11}{SEGMENTS}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_{12}{FOREIGNSALES}_{i,t}+{\beta }_{13}{LEV}_{i,t}+{\beta }_{14}{ROA}_{i,t}+{\beta }_{15}{LOSS}_{i,t}+{\beta }_{16}{ZSCORE}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_{17}{LITRISK}_{i,t}+{\beta }_{18}{MA}_{i,t}+{\beta }_{19}{RESTRUCTURE}_{i,t}+{\beta }_{20}{MKTCAP}_{i,t}+{\beta }_{21}{DISCLOSURE}_{i,t}\hfill \\ +INDUSTRYandYEAR+{\epsilon }_{i,t}\end{array}$$

(1)

where CAM DISCLOSURE QUALITY is NUMCAMS or FOG and CULTURE is IND, UA, or PD.

We slightly modified the control variables for the test model for INVPRO, our 4th measure of CULTURE.

$$\begin{array}{c}CAMDISCLOSUREQUALITY\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}={\beta }_0+{\beta }_1{INVPRO}_{i,t}+{\beta }_2{ICMW}_{i,t}+{\beta }_3{RESTATE}_{i,t}+{\beta }_4{AUCHANGE}_{i,t}+{\beta }_5{AUTENURE}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_6{BIG4}_{i,t}+{\beta }_7{SIZE}_{i,t}+{\beta }_8{AGE}_{i,t}+{\beta }_9{BM}_{i,t}+{\beta }_{10}{SALESGROWTH}_{i,t}+{\beta }_{11}{SEGMENTS}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_{12}{FOREIGNSALES}_{i,t}+{\beta }_{13}{LEV}_{i,t}+{\beta }_{14}{ROA}_{i,t}+{\beta }_{15}{LOSS}_{i,t}+{\beta }_{16}{ZSCORE}_{i,t}\hfill \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{\beta }_{17}{LITRISK}_{i,t}+{\beta }_{18}{MA}_{i,t}+{\beta }_{19}{RESTRUCTURE}_{i,t}+COUNTRYandYEAR+{\epsilon }_{i,t}\hfill \end{array}$$

(2)

where INVPRO measures are defined as COMMON = 1 if the country of the issuer is a common law country, and 0 otherwise.

We calculate FOG as

FOG = 0.4 × words/sentences + 100 × number of words with at least 3 syllables/number of words)

(3)

Following the CAMs literature, we control for firm and audit firm characteristics that affect corporate disclosure practices (Burke et al., 2023; Li et al., 2020). We accomplish this by including variables that capture the client’s history and characteristics of the client, such as size, accounting and operating complexity, and performance. We also include variables that control for significant and complicated economic events, such as mergers (MAs), restatements (RESTATEs), and restructurings (RESTRUCTUREs).

We predict that these, along with internal control weaknesses (ICMWs), firm size (SIZE), age of the firm (AGE), percentage of foreign sales (FOREIGNSALES), rate of sales growth (SALESGROWTH), increase in profitability (ROA), the incidence of losses (LOSS), and company complexity (SEGMENTS) will be positively associated with NUMCAMS, consistent with the notion that past reporting issues and greater complexity increase the number of CAMs reported. Based on prior research, we include litigation risk (LITRISK), since firms in high-risk industries will have a positive association with CAMs (Sulcaj, 2020). Furthermore, we include two country-level variables to control for country wealth and transparency that may affect cross-country variations in CAMs disclosures (MKTCAP and DISCLOSURE).

Next, we predict that several situations at the firm level, regarding uncertainty and predictability, may cause auditors to provide fewer CAMs disclosures. CAMs help to produce quality financial reports by highlighting matters that are material and involve “especially challenging, subjective, or complex auditor judgment” (AS 3101). Subsequently, CAMs increase transparency and disclosure. Abraham et al. (2024) measured the effect of audit oversight on reducing insolvency risk. They linked effective oversight with selecting auditors who provide quality financial reports through highlighting CAMs. They modeled insolvency risk with the Altman Z (ZSCORE), a common measure of bankruptcy. Since ZSCORE comprises working capital and operating income, lower levels of working capital lead to a lower ZSCORE, which is associated with an increase in the risk of insolvency. Similarly, high level of debt (LEV) may lead to a greater risk of insolvency. A high book-to-market (BM) value means that the securities markets are valuing equity cheaply compared to book value, and thus the markets are undervaluing high BM firms. We expect that as the difference between the book value and market value increases, the number of CAMs will be lower since fewer CAMs would be associated with uncertainty.

We include auditor tenure (AUTENURE) because auditors with long tenure are less inclined to provide transparent CAMs disclosures (Li et al., 2020). Similarly, auditors with new clients have less information to impart than auditors who have spent many years with the same client. Ideagen Audit Analytics reported that in 2019, the BIG4 averaged 1.6 CAMs per report, less than the average number of CAMs in our study, which is 2.2 (Coleman & McKeon, 2020). This may be due to the type of client attracted to BIG4 audit firms: large, relatively stable publicly traded companies. We predict that employing a BIG4 auditor (BIG4) will result in a lower number of CAMs when compared to other audit firms.

Dependent Variables

NUMCAMS = number of CAMs disclosed in the audit opinion;

FOG = IND0.4 × (words/sentences+100 × number of words with at least 3 syllables/number of words).

Variables of interest

Culture Variables

IND: scale of 1–100; individualism is the extent to which people feel independent;

UA: scale of 1–100; uncertainty avoidance deals with a society’s tolerance for uncertainty and ambiguity;

PD: scale of 1–100; power distance is the extent to which the less powerful members of organizations and institutions (like the family) accept and expect that power is distributed unequally.

Investor Protection Variables

COMMON = 1 if the country of the issuer is a common law country, and 0 otherwise;

RULE OF LAW INDEX: a measure of investor protections that computes agents’ perceptions of how much they trust and follow the laws of society, specifically the effectiveness of the enforcement of contracts, property rights, the police, and the courts, as well as the likelihood of crime and violence (Kaufmann et al., 2011);

CONTROL OF CORRUPTION: a proxy that captures the perceptions of the extent to which public power is used for personal gain, including both small-scale and large-scale types of corruption (Kaufmann et al., 2011).

Control Variables

ICMW = 1 if the issuer or its auditor discloses a material weakness under SOX 404 or SOX 302 in the current or prior two years, and 0 otherwise;

RESTATE = 1 if the issuer discloses a restatement in the current or prior two years, and 0 otherwise;

AUCHANGE = 1 if the audit firm changed, and 0 otherwise;

AUTENURE = natural log of the number of years of audit firm–client relationship;

BIG4 = 1 if the auditor is a Big4 auditor, and 0 otherwise;

SIZE = natural log of total assets;

AGE = natural log of the number of years for which total assets are reported in Compustat;

BM = book value of equity divided by market value of equity;

SALESGROWTH = current sales minus lagged sales divided by lagged sales;

SEGMENTS = natural log of 1 plus the total number of segments from the Compustat WRDS_SEGMERGED dataset;

FOREIGNSALES = Total non-domestic sales divided by total sales. Non-domestic sales are taken from the Compustat WRDS_SEGMERGED dataset for segments with GEOTP = 3;

LEV = total debt divided by total assets;

ROA = income before extraordinary items divided by average total assets;

LOSS = 1 if income before extraordinary items is negative, and 0 otherwise;

ZSCORE: Altman’s Z-score calculated following DeFond and Hung (2003) and Altman (1968); equal to 1.2 × [net working capital/total assets] + 1.4 × [retained earnings/total assets] + 3.3 × [earnings before interest and taxes/total assets] + 0.6 × [market value of equity/book value of liabilities] + 1.0 × [sales/total assets];

LITRISK = 1 if the issuer belongs to an industry with high litigation risk, 0 otherwise. High-litigation-risk industries are as defined by J. Francis et al. (1994): biotech (2833–36, 8731–34), computer (3570–77, 7370–74), electronics (3670–74), and retail (5200–5961);

MA = 1 if the issuer reports a non-zero pre-tax earnings impact from mergers or acquisitions, and 0 otherwise;

RESTRUCTURE = 1 if the issuer reports non-zero pre-tax restructuring costs, and 0 otherwise;

MKTCAP: market capitalization of listed domestic companies (% of GDP) is a measure of the total value of a country’s publicly traded companies as a percentage of its gross domestic product (GDP) as reported by the World Bank.

DISCLOSURE: LaPorta et al. (1998) established a proxy for transparency. This index measures the inclusion or omission of 90 accounting items in firms’ annual reports, and hence captures firms’ disclosure policies at the country level.

4. Results

Table 4. Regression Results.

Regression of Number of CAMs and Country-Level National Culture
Variable		(1) IND		(2) UA		(3) PD
CULTURE		0.013	***	0.005	**	−0.004	*
		(4.72)		(2.02)		(−1.71)
ICMW		0.237	*	0.213		0.188
		(1.76)		(1.53)		(1.37)
RESTATE		−0.014		−0.014		−0.020
		(−0.11)		(−0.11)		(−0.15)
AUCHANGE		−0.879	***	−0.876	***	−0.862	***
		(−4.76)		(−4.57)		(−4.52)
AUTENURE		−0.298	***	−0.309	***	−0.301	***
		(−7.67)		(−7.64)		(−7.45)
BIG4		−0.621	***	−0.624	***	−0.631	***
		(−4.72)		(−4.74)		(−4.79)
SIZE		0.234	***	0.234	***	0.234	***
		(8.36)		(8.30)		(8.39)
AGE		0.300	***	0.327	***	0.335	***
		(4.79)		(5.17)		(5.30)
BM		0.003		−0.031		−0.031
		(0.04)		(−0.48)		(−0.48)
SALESGROWTH		0.047		−0.013		−0.009
		(0.23)		(−0.06)		(−0.04)
SEGMENTS		0.026		−0.006		0.002
		(0.31)		(−0.07)		(0.02)
FOREIGNSALES		0.089		0.073		0.058
		(0.79)		(0.64)		(0.50)
LEV		0.059		0.156		0.204
		(0.22)		(0.60)		(0.78)
ROA		−0.502	*	−0.506	*	−0.522	*
		(−1.83)		(−1.83)		(−1.89)
LOSS		0.084		0.101		0.098
		(0.79)		(0.94)		(0.91)
ZSCORE		−0.024	***	−0.022	***	−0.022	***
		(−2.89)		(−2.70)		(−2.65)
LITRISK		0.399	**	0.400	**	0.410	**
		(2.34)		(2.37)		(2.40)
MA		0.192	*	0.277	***	0.282	***
		(1.92)		(2.87)		(2.78)
RESTRUCTURE		0.100		0.170		0.177	*
		(0.94)		(1.62)		(1.68)
MKTCAP		−0.016	*	−0.011		−0.013
		(−1.67)		(−1.14)		(−1.29)
DISCLOSURE		−0.037		−0.225	***	−0.148	**
		(−0.50)		(−3.33)		(−2.32)
Intercept		−1.935	***	−3.548	***	−2.701	***
		(−2.88)		(−4.59)		(−3.65)
Industry and year fixed effects		Yes		Yes		Yes
Adj. R-Square		0.288		0.278		0.276
N		1358		1358		1358

*, **, and *** indicate 0.10, 0.05, and 0.01 significance levels, respectively, for a two-tailed test (one-tailed for predicted directions).

presents the regression modeling the association between the number of CAMs and the dimensions of national culture: IND, UA, and PD. We use a fixed effects OLS regression. We absorb the effects for a particular industry and time by using dummies for each two-digit sic code and year. The dependent variable is the number of CAMs reported, and the main variables of interest are the three dimensions of culture. Column 1 shows that the regression coefficient for IND is positive and significant (0.013, p-value ≤ 0.01). Consistent with our expectations, this suggests that high IND societies are associated with a higher number of CAMs. This result is consistent with H1, that auditors within a highly individualistic society that is less tolerant of corruption and accounting errors will report a higher number of CAMs (Davis & Ruhe, 2003; Caban-Garcia et al., 2017).

Examining our control variables, we find that the variables that control for significant and complicated economic events (RESTATE and RESTRUCTURE) are statistically insignificant, implying there is not enough evidence to suggest that the information about them has an effect on the number of CAMs. We also find a positive association between the NUMCAMS and mergers and acquisitions (MAs). The variables that control for the rapid growth and complexity of a firm’s operations, SALESGROWTH and FOREIGNSALES, are not significant. Other characteristics associated with the firms (SIZE and AGE) are statistically significant and positively associated with the NUMCAMS. This implies that older and larger firms are more likely to experience and report CAMs than smaller and younger firms. Furthermore, we find negative associations between the NUMCAMS and ROA and ZSCORE. The effects associated with auditor characteristics (AUCHANGE, BIG4, and AUTENURE) are significant and are negatively associated with the NUMCAMS. The number of segments (SEGMENTS) and LEV also do not appear to effect the NUMCAMS.

The coefficient for UA in column (2) is also positive and significant (p-value ≤ 0.05). This evidence suggests that auditors from countries with high uncertainty avoidance want to avoid risk and unpredictability and will, therefore, provide more CAMs (Pinto et al., 2020). This supports H2. AUCHANGE has a negative association with the NUMCAMS. The BM does not have a statistically significant effect on the NUMCAMS, while MA has increased statistical significance. All the other control variables show little change.

Again, consistent with our expectations, the coefficient for PD in column (3) is negative and weakly significant (p-value ≤ 0.10). This result indicates that higher levels of power distance are associated with fewer CAMs being reported. This supports H3 that auditors in high power distance countries are more likely to be secretive and less transparent, which results in fewer CAMs. In column (3), MA has increased in statistical significance and magnitude (it is still positively associated with the NUMCAMS as predicted). FOREIGNSALES and LEV have no statistical significance. A firm’s market capitalization of listed domestic companies (% of GDP) (MKTCAP and country-level transparency (DISCLOSURE)) is negatively associated with the NUMCAMS when it is statistically significant.

The associations among our three measures of investor protection and the NUMCAMS are reported in

Table 5. Regression Results.

Regression of Number of CAMs and Country-Level Investor Protection Measures
Variable	(1) Common Law		(2) Rule of Law		(3) Control of Corruption
INVPRO	0.504	***	0.015	**	0.108	*
	(4.60)		(2.27)		(1.78)
ICMW	0.190		0.190		0.199
	(1.37)		(1.40)		(1.43)
RESTATE	0.011		−0.018		0.006
	(0.09)		(−0.14)		(0.05)
AUCHANGE	−0.789	***	−0.896	***	−0.856	***
	(−3.68)		(−4.68)		(−3.97)
AUTENURE	−0.292	***	−0.305	***	−0.300	***
	(−8.80)		(−7.72)		(−8.96)
BIG4	−0.586	***	−0.639	***	−0.599	***
	(−5.04)		(−4.85)		(−5.12)
SIZE	0.244	***	0.232	***	0.240	***
	(9.57)		(8.27)		(9.36)
AGE	0.284	***	0.336	***	0.316	***
	(3.76)		(5.39)		(4.19)
BM	−0.015		−0.026		−0.028
	(−0.25)		(−0.41)		(−0.46)
SALESGROWTH	−0.087		0.016		−0.135
	(−0.40)		(0.08)		(−0.63)
SEGMENTS	0.060		0.017		0.021
	(0.78)		(0.21)		(0.28)
FOREIGNSALES	0.120		0.057		0.076
	(1.08)		(0.50)		(0.67)
LEV	0.170		0.189		0.203
	(0.69)		(0.74)		(0.82)
ROA	−0.547	**	−0.542	**	−0.601	**
	(−2.34)		(−1.99)		(−2.56)
LOSS	0.068		0.082		0.061
	(0.65)		(0.78)		(0.58)
ZSCORE	−0.023	***	−0.022	***	−0.021	***
	(−3.58)		(−2.62)		(−3.30)
LITRISK	0.462	**	0.416	**	0.392	**
	(2.39)		(2.49)		(2.02)
MA	0.215	**	0.277	***	0.268	***
	(2.22)		(2.88)		(2.75)
RESTRUCTURE	0.174	*	0.173	*	0.176	*
	(1.78)		(1.68)		(1.79)
MKTCAP	−0.021	**	−0.008		−0.011
	(−2.21)		(−0.86)		(−1.06)
DISCLOSURE	−0.619	*	−0.479	**	−0.500
	(−1.94)		(−2.19)		(−1.55)
Intercept	−0.852		−0.268		−0.611
	(−0.59)		(−0.63)		(−0.42)
Industry and year fixed effects	Yes		Yes		Yes
Adj. R-Square	0.273		0.273		0.263
N	1358		1358		1358

*, **, and *** indicate 0.10, 0.05, and 0.01 significance levels, respectively, for a two-tailed test (one-tailed for predicted directions).

. All of the measures are positively and significantly associated with the number of CAMs. The coefficient of INVPRO is positive and significant at the p ≤ 0.01 level (column 1) for common law, p ≤ 0.05 (column 2) for rule of law, and p ≤ 0.10 (column 3) for control of corruption. Overall, the evidence is consistent and shows that stricter investor protection regimens lead to higher numbers of CAMs being reported.

Table 6. Regression Results.

Regression of FOG and Country-Level National Culture
Variable		(1) IND		(2) UA		(3) PD
CULTURE		0.005		0.018	**	−0.026	***
		(0.64)		(2.52)		(−3.82)
ICMW		1.227	***	1.299	***	1.173	***
		(3.07)		(3.26)		(2.96)
RESTATE		0.443		0.470		0.469
		(1.20)		(1.28)		(1.28)
AUCHANGE		0.894		0.818		0.778
		(1.45)		(1.33)		(1.27)
AUTENURE		0.151		0.108		0.108
		(1.59)		(1.11)		(1.13)
BIG4		0.351		0.356		0.267
		(1.04)		(1.06)		(0.80)
SIZE		0.056		0.069		0.106
		(0.76)		(0.93)		(1.42)
AGE		−0.078		−0.130		−0.195
		(−0.36)		(−0.59)		(−0.89)
BM		−0.129		−0.140		−0.127
		(−0.74)		(−0.81)		(−0.74)
SALESGROWTH		0.439		0.389		0.371
		(0.70)		(0.62)		(0.59)
SEGMENTS		0.569	***	0.560	**	0.672	***
		(2.58)		(2.55)		(3.05)
FOREIGNSALES		0.152		0.206		0.172
		(0.47)		(0.64)		(0.54)
LEV		0.511		0.296		0.216
		(0.72)		(0.42)		(0.31)
ROA		−0.217		−0.151		−0.173
		(−0.32)		(−0.22)		(−0.26)
LOSS		0.244		0.257		0.233
		(0.81)		(0.85)		(0.77)
ZSCORE		0.014		0.012		0.010
		(0.76)		(0.65)		(0.58)
LITRISK		1.224	**	1.141	**	1.040	*
		(2.21)		(2.06)		(1.88)
MA		−0.420		−0.464	*	−0.618	**
		(−1.46)		(−1.65)		(−2.16)
RESTRUCTURE		0.348		0.334		0.303
		(1.21)		(1.18)		(1.07)
MKTCAP		−0.009		0.001		0.003
		(−0.29)		(0.04)		(0.10)
DISCLOSURE		−0.399	*	−0.089		−0.567	***
		(−1.83)		(−0.44)		(−2.99)
Intercept		30.970	***	28.160	***	33.760	***
		(6.02)		(5.43)		(6.56)
Industry and year fixed effects		Yes		Yes		Yes
Adj. R-Square		0.049		0.054		0.060
N		1358		1358		1358

*, **, and *** indicate 0.10, 0.05, and 0.01 significance levels, respectively, for a two-tailed test (one-tailed for predicted directions).

presents further results examining the relation between the readability of CAMs disclosures and culture. We find that the coefficients of IND in column (1) are not significant. Thus, H4 is not supported.

Auditors from countries with high uncertainty avoidance tend to provide lengthy and complex CAMs disclosures. The results of the association between UA and readability indicate that, as predicted, high UA cultures are associated with more complex (less readable) reports. The coefficient of UA in column (2) is positive and significant (p-value ≤ 0.05). Auditors in high UA societies are likely attempting to avoid omitting material information that might be overlooked during the audit process. They would rather communicate more detail in an attempt not to leave something of importance out. This provides evidence that supports H5.

The coefficient of PD in column (3) is negative and significant (p-value ≤ 0.01), as predicted. The evidence also indicates that auditors in high power distance cultures issue reports that are easy to understand in order to reduce scrutiny from the PCAOB and because they rely on formal codes of conduct. In high power distance cultures, auditors write more concise and clear CAMs disclosures. This provides evidence that supports H6.

Examining our control variables, we find that the variables that control for significant and complicated economic events (MA and LITRISK) are statistically significant. Thus, there is evidence to suggest that information about them influences the number of CAMs. In Column 1, we find MA is not significant, meaning that mergers and acquisitions are not associated with readability. This contrasts with the positive (less readability) association with SEGMENTS in columns 4 and 5. These results highlight the difference between these two complex situations. The evidence also suggests that in the presence of a disclosed material weakness under SOX 404 or SOX 302 in the current or prior two years (ICMW), CAMs become, as expected, more complex and less readable.

Once again, there is not enough evidence to suggest an effect from SALESGROWTH, ROA, AGE, or SIZE. We do find, however, a positive and statistically significant association between LITRISK and FOG, which also supports the desire to avoid litigation.

Surprisingly, there is not enough evidence to suggest that LEV, LOSS, or ZSCORE are associated with readability. This may be because managers have sufficiently communicated information concerning the risk of failure to investors through other means besides the audit report. The effects associated with auditor characteristics (AUCHANGE, BIG4, and AUTENURE) are statistically insignificant or only marginally, positively associated with FOG.

Table 7. Regression Results.

Regression of FOG and Country-Level Investor Protection Measures
Variable	(1) Common Law		(2) Rule of Law		(3) Control of Corruption
INVPRO	−0.988	**	−0.054	**	−0.508	**
	(−2.02)		(−2.08)		(−2.24)
ICMW	1.430	***	1.428	***	1.448	***
	(3.73)		(3.33)		(3.45)
RESTATE	0.499		0.619		0.480
	(1.13)		(1.62)		(1.27)
AUCHANGE	0.602		0.815		0.805
	(1.05)		(1.32)		(1.32)
AUTENURE	0.080		0.122		0.110
	(0.80)		(1.23)		(1.13)
BIG4	0.193		0.152		0.227
	(0.52)		(0.44)		(0.67)
SIZE	0.035		0.079		0.042
	(0.40)		(0.95)		(0.53)
AGE	0.082		0.053		0.027
	(0.33)		(0.22)		(0.12)
BM	−0.160		−0.110		−0.142
	(−0.98)		(−0.61)		(−0.81)
SALESGROWTH	0.215		0.270		0.296
	(0.28)		(0.41)		(0.46)
SEGMENTS	0.494	*	0.462	*	0.520	**
	(1.89)		(1.90)		(2.19)
FOREIGNSALES	−0.101		0.025		0.015
	(−0.28)		(0.07)		(0.04)
LEV	0.970		0.898		0.982
	(1.16)		(1.15)		(1.31)
ROA	−0.318		−0.673		−0.228
	(−0.36)		(−0.94)		(−0.33)
LOSS	0.302		0.148		0.387
	(0.92)		(0.47)		(1.26)
ZSCORE	0.025		0.020		0.023
	(1.07)		(1.08)		(1.27)
LITRISK	−1.429		−4.641		−1.392
	(−0.55)		(−1.17)		(−0.61)
MA	−0.117		−0.229		−0.111
	(−0.41)		(−0.79)		(−0.39)
RESTRUCTURE	0.268		0.360		0.324
	(0.88)		(1.20)		(1.12)
MKTCAP	−0.015		0.010		−0.009
	(−0.42)		(0.32)		(−0.28)
DISCLOSURE	0.040		−0.052		0.015
	(0.17)		(−0.21)		(0.06)
Intercept	25.030	***	25.160	***	24.830	***
	(15.83)		(6.00)		(5.95)
Industry and year fixed effects	Yes		Yes		Yes
Adj. R-Square	0.107		0.108		0.107
N	1358		1358		1358

*, **, and *** indicate 0.10, 0.05, and 0.01 significance levels, respectively, for a two-tailed test (one-tailed for predicted directions).

presents the results for the association between FOG and investor protection. The three columns present the three different investor protection measures. The coefficients of the investor protection variables are significantly negative (p ≤ 0.05) in each case. This suggests that in countries with greater investor protection, an auditor’s CAMs are concise and easier to read.

The PCAOB has employed joint inspections and other cooperative arrangements as its primary approach to monitor and ensure the quality and integrity of international audits. Since prior research has shown that joint inspections countries experience lower abnormal accruals (Krishnan et al., 2017), we examined whether the PCAOB joint arrangement inspections affects the number of CAMs. In an untabulated test, we confirmed that the PCAOBCOOP is negative and significant (p-value ≤ 0.05). This suggests that cooperation among international regulators strengthens global compliance with auditing standards. Furthermore, countries that have an investigation and discipline system according to the IFAC (International Federation of Accountants) reported lower numbers of CAMs (p-value ≤ 0.01); this is consistent with the notion that “Investigation and discipline mechanisms are equally vital for sustainable professional accountancy organizations and ensuring public trust in the accountancy profession” (IFAC, 2019).

5. Conclusions

In response to providing more information to investors, the PCAOB now requires auditor reports to include CAMs paragraphs that provide information concerning accounts or disclosures deemed to be material and that involve especially challenging, subjective, or complex auditor judgment. Several studies have examined the relationship between audit practice and Hofstede’s cultural indices. Doupnik and Riccio (2006) developed a hypothesis about how the interaction of culture and the accounting values of conservatism and secrecy influence accountants’ interpretations of those expressions. Chen et al. (2007), using a sample of firms from Taiwan and Singapore, used Hofstede’s cultural dimensions of power distance, uncertainty avoidance, and individualism to conclude that audit practice cannot be culture-free and varies the risk-assessment level. Li et al. (2020) found that auditor tenure is negatively associated with the number of CAMs disclosed. We are the first to study how national cultural values can affect accountants’ interpretations of CAMs using a large sample of cross-listed firms on US stock exchanges. CAMs are an indication that SEC/PCAOB audits allow auditors to adjust accounting principles to a company’s transactions, rather than adjusting a company’s transactions to accounting rules. CAMs allow for flexibility while maintaining credibility, trust, and confidence in the financial statements of emerging and international entities that are traded in US capital markets. These findings should be of significant interest to the PCAOB as it broadens its oversight across international borders.

We are also the first to examine the relationship between culture and the readability and number of CAMs. We found that the number of CAMs reported is significantly associated with Hofstede’s (2001) cultural dimensions and investor protection (as measured by the common law, rule of law, and control of corruption). We controlled for several measures in the CAMs literature related to complex economic situations (such as mergers, restatements, litigation risk, and restructurings). We also controlled for changes in auditors, auditor tenure, and the use of a Big4 auditor. In addition, we controlled for firm characteristics that are related to the risk of failure (leverage, bankruptcy, and whether the firm has experienced a recent loss), as well increases in complexity related to financial statements (segments, size, book-to-market value, sales growth, return on assets, and foreign sales). We predicted and found that national culture is associated with the number of critical audit matters disclosed in an audit report, and with the audit report’s readability.

As an overseer of audit quality, the extent to which cultural characteristics are associated with the number of CAMs should be of significant interest to the PCAOB. The extent to which readability is associated with culture should similarly be of concern. The extent to which investor protections are associated with culture and readability should be of interest to the users of financial statements as well as to regulators. Our results should inform policy, as additional guidance for auditors may be necessary in certain countries.

Limitations: Our research is subject to the limitations of language bias. English is very different from other non-Germanic languages. Translation, no matter how well performed, will always experience problems in trying to achieve equivalent wording and meaning. Nuances in language cannot be fully captured by the Fog Index. As we and other studies have demonstrated, culture affects risk. We expect that CAMs disclosures will change over time as auditors become more experienced with global accounting regulations. Currently, we have only two years of CAMs disclosures and it is difficult to assess the impact that the COVID-19 pandemic may have had on our results. Future research could extend our study and even explore related spillover effects to other areas of financial statements and audit reports.