Quarterly Audit Reviews and Earnings Quality: International Evidence

(1)

0

Quarterly Audit Reviews and Earnings Quality:

International Evidence

Abstract

In this paper I assess the relation between the voluntary purchase of quarterly audit reviews (reviews) and earnings quality, as measured by discretionary accruals. Moreover, I test the moderating effect of national culture on this relation. Drawing on a sample of 6,586 firm-year observations from five countries from 2005 to 2015, I find a positive correlation between the voluntary purchase of reviews and earnings quality. Furthermore, I find that power distance negatively moderates this association, which is consistent with high power distance countries being characterized by high amounts of secrecy. Uncertainty avoidance positively moderates the correlation between quarterly reviews and earnings quality. This is consistent with the conservative nature of cultures characterized by high uncertainty avoidance. This study contributes to the current debate on the desirability of reviews by documenting benefits of their voluntary purchase for earnings quality.

Keywords: quarterly reports, earnings quality, quarterly review, national culture.

Dennis TIJINK

Student number: 2152983 Master Thesis Accountancy Supervisor: Dr. V.A. PORUMB

(2)

1

1. Introduction

The audit review of quarterly reports (review) has been a requirement for all listed organizations in the United States (US) starting 2002 (Bédard and Courteau, 2015).1 Concurrently, in other countries it remains a voluntary practice.2 In these settings, some organizations choose to have their quarterly statements reviewed while other organization choose to not do this. The question which therefore arises is why do firm voluntarily choose to review their quarterly reports? Several researchers have looked into the costs and benefits for firms that voluntarily choose to review their quarterly reports3 (e.g. Bandyopadyay et al., 2007; Bédard and Courteau, 2015; Boritz and Liu, 2006). However, the literature provides weak and inconclusive results and this paper aims to get a better understanding of why firms review their quarterly reports. Specifically, I to add to this debate by studying how the assurance provided by the audit of quarterly reports impact the quality of earnings in an international setting.

On one hand, information asymmetry between the agents and principals of the organization would incentivize managers to disclose more information. This would translate into firms with poor earnings quality providing more voluntary disclosures, since the information asymmetry between the agent and principal is higher (Francis et al., 2008). On the other hand, when a firm has a higher amount of earnings quality, agents may be inclined to disclose more voluntary information as well, since the information is more relevant to the principals (Francis et al., 2008). By assessing the effect of voluntary purchasing a review on earnings quality, this paper advances knowledge on the usefulness of reviews in decreasing information asymmetry.

By using a sample of 6,586 observations from 5 countries in the 2005-2015 period, I find that the voluntary purchase of reviews is positively associated with the quality of quarterly reports. Furthermore, I test the effect of cultural differences on the association between reviews and earnings quality, since according to Gray (1988) culture influences the accounting systems in a country. While culture is therefore likely to have an impact in the usefulness of the review, this effect has not been studied within this specific research setting. I find that power distance negatively moderates the association between the voluntary purchase reviews and earnings quality. Additionally, I find that uncertainty avoidance positively moderates the association between reviews and earnings quality. The observations regarding the voluntary purchase of

1_{The Securities and Exchange Committee (SEC) requires listed US companies to have their quarterly reports} audited (Bédard & Courteau, 2015).

2_{Examples of countries that do not mandate quarterly reviews are Canada, the United Kingdom, the Netherlands,} and Spain.

(4)

3 reviews are hand-collected , whereas for the earnings quality and control variables I use the database Datastream. For the cultural variables I use the measures developed by Hofstede (1980). The hypotheses have been tested by using an OLS regression and additional analyses have been carried out to ensure the robustness of the results. These analyses include the use of an alternative measure for earnings quality and the use of a fixed-effects GLS panel regression, which I use since the data contains multiple firm-year observations.

This paper makes several contributions to the current literature. First, it adds to the debate of whether assurance improves the quality of voluntary information. Bandyopadhyay et al. (2007) examined the correlation of voluntary assurance on quarterly reports of organizations in Canada, comparing them to organizations in the US that have mandatory assurance on their quarterly reports. They found that voluntary assurance has value as a signalling mechanism by looking at the volatility of the accruals of the firms. This paper contributes to the literature of voluntary assurance by further looking at this correlation by examining other traits of earnings quality as described by Dechow et al. (2010).

Second, this paper contributes to the literature of quarterly reports. A requirement of the Securities and Exchange Committee (SEC) is that al organizations listed in the US have their quarterly reports reviewed (Bandyopadhyay et al., 2007). However, according to Bédard and Courteau (2015) there is still a debate going on whether mandatory assurance on the quarterly report adds value to the report or not. This paper adds to this debate and may provide results that could help regulators with future legislation regarding quarterly reports.

The results of this study could be relevant to regulatory bodies as well as firm’s managers. According to Boritz and Liu (2006) quarterly reports are not seen as reliable due to a variety of reasons. E.g. a significant part of the results is not known until the end of the year, seasonality factors, costs and earnings are mismatched between various quarters. By examining the potential link between a review and earnings quality, this paper adds to the current discussion of requiring reviews of organizations and could influence the choice of regulatory bodies to mandate firms to have their quarterly reports reviewed by an independent party or not.

Furthermore, the results of this study could be interesting from a managerial perspective. If the reliability of a quarterly report improves from having a quarterly review, this would mean that a manager could use the review to reduce the information asymmetry between the firm and its stakeholders and to increase the decision usefulness of the information in the quarterly report.

(5)

4 The remainder of the paper is organized as follows. Section two discusses the theoretical framework of the paper. Section three describes the methodology of the paper. Section four shows the results and the interpretation. Section five consists of the discussion and conclusion of the paper.

2. Theoretical framework

In this section the theoretical framework on which the paper builds is set out. First, the agency theory is discussed, which is followed by a review of earnings quality literature. After that, the background and history of reviews are discussed. Finally, the role of national culture within accounting literature is explained.

2.1 Quarterly reviews

Reviews differ significantly in scope from traditional audits. Generally, reviews have a more narrow scope and have limited inquiries and consist of restricted analytical procedures from which significant matters will be found (Bédard and Courteau, 2015; Ettredge et al., 1994). Furthermore, a review often lacks the evaluation of the internal control system, collection of confirmative evidence, physical inspection, confirmation from external parties, or tests of account balances and transactions (Bédard and Courteau, 2015; Ettredge et al., 1994). Since there is a distinction between the definition of a review and an audit, it is important to note that when a quarterly report states that it has not been audited, it could still have been reviewed.

According to Ettredge et al. (2000) there are two types of reviews: timely reviews and retrospective reviews. Timely reviews (or quarter-end reviews) are reviews that are conducted at the end of each quarter. Retrospective reviews (or year-end reviews) are reviews that are conducted in conjunction with the year-end audit. While there are researchers that focus on the difference between timely reviews and retrospective reviews (e.g. Ettredge et al., 1994; Ettredge et al., 2000; Manry et al., 2003), this is not in the scope of this paper. Therefore, no distinction is made between timely reviews and retrospective reviews in this paper.

Reviews add two kinds of value to the reports (Kajüter et al., 2016). First, there is assurance value, for which the review helps to contain measurement errors and curbs earnings management. Second, the review adds signalling value, which is that investors expect the quality to be higher by the sole fact that the quarterly report is reviewed even though the earnings quality has not been improved. The assurance value can be measured by the use of

(6)

5 category 1 proxies of earnings quality (Dechow et al., 2010)4_{, which is defined as the properties}

of earnings. While the signalling value of the review could be measured by looking at category 2 proxies of earnings quality (Dechow et al., 2010), which consists of the investor responsiveness of the earnings. This paper intends to look at the assurance value of the quarterly review.

Bédard and Courteau (2015) tried to examine the cost and benefits of having a voluntary review on quarterly reports in Canada. However, they did not find a significant correlation between a quarterly report being reviewed and the earnings quality of that report. Bandyopadhyay et al. (2007) conducted similar research on Canadian organizations. They found that a voluntary review signals quality of their quarterly reports to outsiders of the organization. Both Bandyopadyay et al. (2007) and Bédard and Courteau (2015) find that the reviews do not add assurance value, while Banyopadhyay et al. (2007) finds that the quarterly reviews do add signalling value to the report.

An example of a country in which reviews are mandatory is the United States. In 1999, the Blue Ribbon Committee issued ten recommendations to enhance the relevance and reliability of reported earnings to the SEC5, of which one recommendation was to timely review quarterly reports (Manry et al., 2003). In March 2000, the SEC adopted this recommendation and required SEC-listed organizations to have their quarterly reports reviewed on a timely basis (Manry et al., 2003). However, while it is mandatory for SEC-listed firm to have their quarterly reports reviewed, this is not the case in a significant amount of countries6_{(Bédard and Courteau}

2015). The fact that organizations in many countries can choose to have their quarterly reports reviewed provides the basis for this research.

2.2 Agency theory

The main theory on which this paper builds is the agency theory. The agency theory sees the organization as a nexus of contracts between various resource holders inside and outside the firm (Hill and Jones, 1992). This theory defines a relationship between a principal and an agent, in which the agent performs activities on behalf of the principal. This is done by the delegation of decision-making authority from the principal to the agent (Hill and Jones, 1992; Jensen and Meckling, 1976). However, it is possible that the agent will not act in the

4_{See section 2.3 for more information on earnings quality proxies.} 5_{The Securities and Exchange Commission.}

6_{Examples of countries that do not mandate quarterly reviews are Canada, the United Kingdom, the} Netherlands, and Spain.

(7)

6 interest of the principal since their interests could deviate due to the separation of ownership and control and due to information asymmetry between both parties (Fama and Jensen, 1983). Hence, Agency costs have to be incurred to align the interests of both parties involved (Hill and Jones, 1992; Jensen and Meckling, 1976). Agency costs consists of bonding costs, monitoring costs and the residual loss. Bonding costs are incurred by the agent to ensure the principal that no harm will be done to the principal or his interests. Monitoring costs are made by the principal to limit the opportunistic behaviour by the agent. The residual loss consists of the divergence between the interests of the principal and the actions of the agent after the monitoring and bonding costs that have been incurred by both parties (Hill and Jones, 1992; Jensen and Meckling, 1976).

In accounting literature one argument is to reduce the amount of information asymmetry between the principal and agent by having regulation in place regarding the disclosure of information in the form of accounting rules and standards (Healy and Palepu, 2001). When the information asymmetry between both parties is reduced, it would result in lesser agency costs by possible reduction of the monitoring cost, bonding costs, or the residual loss.

2.3 Earnings quality

For the definition of earnings quality this paper looks to the comprehensive research of Dechow et al. (2010),which defines earnings quality as follows: higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.7 There are three things about this definition that should be noted. Firstly, earnings quality is dependent on the decision-relevance of the information. This means that earnings quality has to be defined in relation to a specific decision model. Secondly, earnings quality should be informative about the performance of an organization. However, not all parts of a firm’s performance can be observed. Lastly, earnings quality is determined by both the relevance of the financial performance to the decision-maker and the competence of the accounting system that measures this performance (Dechow et al., 2010).

7_{For this definition Dechow et al. (2010) used language as stated in the Statement of Financial Accounting} Concepts No. 1 (SFAC No. 1).

(8)

7 Subsequently, Dechow et al. (2010) classified proxies to measure earnings quality into three categories: the properties of earnings (category 1), investor responsiveness to earnings (category 2), and external indicators of earnings misstatements (category 3).

Properties of earnings (category 1) includes earnings persistence, abnormal accruals, residuals from accrual models, earnings smoothness, asymmetric timeliness and timely loss recognition, and target beating (Dechow et al., 2010). An example of research regarding quarterly reviews that uses a properties of earnings proxy is the research of Bandyopadhyay et al. (2007) in which the researchers specifically focussed on the volatility of accruals within Canadian listed firms. Ettredge et al. (2000) have used the timing of recognition of adjustment of quarterly earnings as proxy in examining quarterly reviews. Boritz and Liu (2006) have focussed on the timeliness of the report when examining quarterly reports. Bédard and Courtreau (2015) have measured the unexpected accruals of Canadian firms when trying to capture earnings quality. The residual from accruals model has become one of the most prevalent methods used within accounting literature (Dechow et al., 2010). In this method, the residual estimates management discretion or estimation errors. This paper uses a residual from accruals model, which falls into category 1 proxy to measure earnings quality.

In the literature stream of investor responsiveness (category 2) the market response regarding the share prices is usually measured (Dechow et al., 2010). In the quarterly report literature examples of researchers using investor responsiveness as a proxy for earnings quality are Balsam et al. (2000) who use the proxy to measure the cumulative abnormal returns around the filing date of Form 10-Q, Griffin (2003) who measures the excess returns of share prices after Form 10-Q filings, and Brown and Caylor (2005) who use investor reaction to examine quarterly earnings thresholds. Furthermore, Manry et al. (2003) measure the difference in the market reaction to quarterly reports reviewed on a timely versus a retrospective basis.

The last group that consists of external indicators of earnings misstatements (category 3) contains the following subgroups: accounting and auditing enforcement releases, earnings restatements, and internal control deficiencies (Dechow et al., 2010). An example of authors using external indicators of earnings misstatements in quarterly reporting are Palmrose et al. (2004) who measure the market reaction to earnings restatements of quarterly and annual reports.

2.4 National culture

One of the most used frameworks for the measurement of national culture within the accounting literature is the framework of Hofstede (Sivakumar and Nakata, 2001; Doupnik and

(9)

8 Tsakumis, 2004) The definition of culture that this paper uses comes from the research of Hofstede (1980) which defines culture as “the collective programming of the mind which distinguishes the members of one human group from another”. Hofstede divided national culture into four dimensions, namely power distance, uncertainty avoidance, individualism versus collectivism, and masculinity versus femininity8 (Hofstede, 1980; Hofstede, 1983).

Power distance refers to the degree of centralized decision-making and the extent to which society accepts the hierarchical order within society (Gleason et al., 2000; Hofstede, 1983; Kanagaretnam et al., 2011). Uncertainty avoidance refers to the degree of uncertainty to which a society is feeling comfortable (Hofstede, 1980). A culture with high uncertainty avoidance relates to a lower preference for ambiguity (Nabar and Boonlert-U-Thai, 2007). In a culture with a high amount of individualism society emphasizes the importance of personal time, freedom and challenge. Furthermore, in societies with a high amount of individualism managers have a tendency to look after one’s own interest (Gleason et al., 2000). A culture with a lower individualism focusses more on training, use of skills, physical conditions, and benefits (Hofstede, 1983). A highly masculine culture is an assertive culture that entails decisiveness and competitiveness, and that focusses on performance and achievement (Hofstede and Bond, 1988; Kanagaretnam et al., 2011; Nabar and Boonlert-U-Thai, 2007). The fundamental issue in this dimension is the allocation of social roles to the sexes (Gray, 1988), while a culture with a low amount of masculinity values modesty and nurturing (Hofstede and Bond, 1988). Within accounting research, the dimensions of uncertainty avoidance and individualism are used the most (Hooghiemstra et al., 2015; Hope, 2003).

Based on Hofstede’s culture dimensions Gray (1988) proposed a theory trying to link national culture with the development of accounting systems. This theory consisted of four hypotheses which later have been assessed by other researchers who found significant results. These results are summarized in the paper of Doupnik and Tsakumis (2004). In the theory of Gray (1988) accounting practices subsist of professionalism, uniformity, conservatism, and secrecy. Professionalism consists of professional judgement and is closely related to individualism and uncertainty avoidance. Uniformity of accounting practices is linked with uncertainty avoidance and individualism. Conservatism is the reserved attitude of accounting in reporting profits and is associated with uncertainty-avoidance. Lastly, secrecy can be defined as the amount of information that is available to parties outside the organization and relates to

8_{Hereafter, the categories will be referred to as: power distance, uncertainty avoidance, individualism, and} masculinity. This means that a low degree of individualism (masculinity) refers to collectivism (femininity).

(10)

9 uncertainty avoidance, individualism, and power distance. Gray’s framework can be found below in figure 1. Note that this framework has no direct link between Hofstede’s dimensions and the accounting systems in a society, but rather an indirect link through a countries accounting values.

Figure 1. Gray’s framework of culture’s influence on accounting (Gray, 1988).

2.5 Hypothesis building

In this subsection I am building the hypotheses of this paper. In section 2.5.1, the hypothesis regarding quarterly reviews and earnings quality is formulated. In section 2.5.2, the hypotheses for culture as a moderator are set out.

2.5.1 Quarterly review influence on earnings quality.

Brown and Pinello (2007) find that, relative to annual reports, interim reports are more subject to income-increasing earnings management due to the fact that annual reports are being audited. Hence, it could be said that the quarterly report is not seen as having the same level of reliability as the annual report9. While the scope of an audit differs significantly from the scope of a review, it would be a logical argument that a review (to a lesser degree than the audit of the annual report) does limit a manager’s options to engage in earnings management by means of accruals management. This would result that reviews do have an assurance value.

According to the agency theory, it is possible for agents to not act in the best interest of the principals due to the separation of ownership and control, since there is information

(11)

10 asymmetry between both parties (Fama and Jensen, 1983). To align the interests of the agent and principal the information asymmetry could be reduced by incurring agency costs. One way to reduce the information asymmetry between both parties is for the principal (shareholder) to pressure the agent (manager) to voluntarily choose to have the quarterly report reviewed by an auditor. When the report is subject to a review the agent should have less possibilities for earnings management.

While previous literature brings mixed evidence on the assurance value of the review, I expect that when drawing on data from listed companies with a voluntary choice to purchase the reviews, there exists a positive relation between a voluntary review and earnings quality. Hence, the following hypothesis is stated:

H1: Firms which voluntarily purchase reviews have earnings of higher quality relative

to no-review firms.

2.5.2 Culture as a moderator

For national culture the models of Hofstede (1983) and Gray (1988) are used as explained in section 2.4. I hypothesize that national culture influences the association between reviews and earnings quality as described in H1. Following the theory of Gray (1988) it seems likely that national culture does not directly influence the accounting systems directly, but indirectly through accounting values embedded in society. Since national culture consists of four dimensions each with different characteristics, the moderating influence of each dimension on the relation between reviews and earnings quality will be separately hypothesized.

High power distance societies focus on centralized decision-making and have a distinct hierarchical order, hence managers have more influence over financial reporting choices (Kanagaretnam et al., (2011). Furthermore, in a culture with high power distance, equality and interpersonal trust have less value than in a lower power distance culture (Gleason et al., 2000). According to Hope (2003) organizations in higher power distance societies are more likely to discourage extensive information sharing. According to Gray’s theory (1988) power distance is positively related to secrecy, which in turn has influence on the disclosure behaviour of organization (Zarzeski, 1996).

I expect, consistent with current literature, that when managers are less willing to voluntarily disclose information, they should also be less willing to have their quarterly report reviewed and hence, have more room for earnings management.

(12)

11

H2: Power distance negatively influences the association between quarterly reviews

and earnings quality.

Societies with a high amount of uncertainty avoidance prefer a lower degree of ambiguity (Hofstede, 1983). In the research of Doupnik (2008) uncertainty avoidance is related to earnings management, as measured by earnings smoothing and managers’ earnings discretion.

According to Hooghiemstra et al. (2015) it is more likely that shareholders pressure managers to have increased bonding mechanisms in societies with high uncertainty avoidance. This could have multiple effects. On one hand, it could result in increased pressure from shareholders to include having a quarterly review as a bonding mechanism. On the other hand, it could result in not needing the quarterly review as a consequence of the increase of other bonding mechanisms. Gray (1988) suggests that uncertainty avoidance is positively related to conservatism, which is the reserved attitude to report profits. There is mixed evidence on whether the relation of uncertainty avoidance and conservatism is significant. However, the evidence predominantly points to the existence of this relation10 (Doupnik and Tsakumis, 2004).

While the evidence in current literature is partially mixed regarding the effect of uncertainty avoidance on the association between reviews and earnings quality, I predict a positive relation, consistent with the research of Dopunik (2008).

H3: Uncertainty avoidance positively influences the association between quarterly

reviews and earnings quality.

In societies with a high amount of individualism the importance of personal time, freedom and challenge is emphasized (Hofstede, 1983). These societies tend to be more competitive and less secretive (Hope, 2003), which would suggest a positive relation with the amount of voluntary disclosure.

In individualistic cultures, managers have the tendency to look after their own interests and personal achievement (Gleason et al., 2000; Hooghiemstra et al., 2015). This could imply that managers are more focussed on career and reputation building (Hooghiemstra et al., 2015).

10_{While most researches as summarized by Doupnik and Taskumis (2004) do support the relation, some do not} find significant results.

(13)

12 To build a good reputation, managers have to disclose credible information (Healy and Palupu, 2001). Zarzeski (1996) finds a similar positive link between individualism and the disclosure amount.

Nabar and Boonlert-U-Thai (2007) state that earnings management is more likely to happen in societies with a high amount of individualism due to the fact that managers have less concern for other stakeholders’ welfare. Similarly, managers have more incentives for taking risks (Kanagaretnam and Lobo, 2011). This would imply a positive relation between individualism and earnings management and, thus, a negative relation between individualism with the amount of voluntary disclosure.

Doupnik (2008) finds a negative correlation between individualism and earnings management, measured as earnings smoothing and earnings discretion by managers. This can be explained by the fact that members in a society with a low amount of individualism have an urge to protect their in-group members.

While an argument could be made that the amount of individualism could both positively, as well as negatively influences manager’s disclosure amount, I find the evidence suggesting a positive relation is more conclusive. Therefore, the following hypothesis is made:

H4: Individualism positively influences the association between quarterly reviews and

earnings quality.

The Hofstede dimension masculinity and its link with accounting systems is one of the least researched dimension of the Hofstede model and focusses on performance and achievement (Hooghiemstra et al., 2015; Hope, 2003). In societies with a high masculinity it is more likely encouraged to report strong profits that beat target benchmarks (Nabar and Boonlert-U-Thai, 2007). This could mean that managers have more incentives to engage in earnings management. Johnson and Droege (2004) found that a more masculine society relates more to the principles of the agency theory regarding conflicts of interest and goal congruency based on outcome-based pay.

According to Zarseski (1996) countries with a high masculinity are related to increased information disclosure. However, in a the study of Gray and Vint (1995), no such relation is found. Furthermore, in the study of Doupnik (2008) no significant relation was found between masculinity and earnings management.

(14)

13 While there are theories that do suggest an indirect relation between masculinity and earnings management, many researchers do not find such a link. Therefore, the following hypothesis is made:

H5: Masculinity does not influence the association between quarterly reviews and

earnings quality.

3. Methodology

In this section the methodology of this paper is described. First, the sampling method is explained. Following this, the variables, including regression models, which have been used in this research are described. Finally, validity and reliability issues are dealt with in the last subsection.

3.1 Sample

The sample starts with 10.356 firm-year observations from 2005 to 2015 for which review data was collected by hand. Information regarding the other financial variables was obtained from Datastream, with the exception of the cultural variables, for which I use the index of Hofstede (1980). The observations are collected from firms in Canada, Netherlands, Hong Kong, Sweden, and Germany. First, from the initial sample of 10.356 observations, no information was found on earnings quality for 173 observations, which were then omitted from the sample. Second, no information was found on the inventory ratio of 3.327 observations. Third, 42 observations had no data regarding firms’ leverage, which were then omitted. Fourth, 84 observations missing data on the Return on Assets were omitted. Finally, 144 observations were omitted from the sample, since information was missing on the Market-to-Book ratio. This resulted in a final sample size of 6.586 firm-year observations, as depicted in Table 1.

[Insert Table 1 here]

3.2 Variables

A summary of the variables used in this paper is given in Table 2 below. The variables are explained in further details in section 3.2.1 to 3.2.4.

(15)

14 With the variables from Table 2 Models (1) and (2) are estimated to test the hypotheses. Model (1) is used to test Hypothesis 1 regarding the association between earnings quality and reviews.

(1) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽₁𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽₂𝑆𝐼𝑍𝐸 + 𝛽₃𝐼𝑁𝑉 + 𝛽₄𝐿𝐸𝑉 + 𝛽₅𝑅𝑂𝐴 + 𝛽₆𝑀𝑇𝐵 + 𝛽₇𝐼𝑁𝐷 + 𝜀

Model (2) is estimated to test the moderating effect of culture, in which I look at the effect of the interaction between reviews and the cultural dimension. Then, in Model (2) I substitute CULTURE by the cultural dimension as specified by Hypothesis 2 to 5. Resulting in Models (3) to (6) which I use to test the Hypotheses regarding the moderating effect of culture per dimension. (2) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽₁𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽₂𝐶𝑈𝐿𝑇𝑈𝑅𝐸 + 𝛽₃𝑅𝐸𝑉𝐼𝐸𝑊 ∗ 𝐶𝑈𝐿𝑇𝑈𝑅𝐸 + 𝛽4𝑆𝐼𝑍𝐸 + 𝛽5𝐼𝑁𝑉 + 𝛽6𝐿𝐸𝑉 + 𝛽7𝑅𝑂𝐴 + 𝛽8𝑀𝑇𝐵 + 𝛽9𝐼𝑁𝐷 + 𝜀 (3) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽₁𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽₂𝑃𝐷𝐼 + 𝛽₃𝑅𝐸𝑉𝐼𝐸𝑊 ∗ 𝑃𝐷𝐼 + 𝛽₄𝑆𝐼𝑍𝐸 + 𝛽5𝐼𝑁𝑉 + 𝛽6𝐿𝐸𝑉 + 𝛽7𝑅𝑂𝐴 + 𝛽8𝑀𝑇𝐵 + 𝛽9𝐼𝑁𝐷 + 𝜀 (4) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽₁𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽₂𝑈𝐴𝐼 + 𝛽₃𝑅𝐸𝑉𝐼𝐸𝑊 ∗ 𝑈𝐴𝐼 + 𝛽₄𝑆𝐼𝑍𝐸 + 𝛽₅𝐼𝑁𝑉 + 𝛽₆𝐿𝐸𝑉 + 𝛽₇𝑅𝑂𝐴 + 𝛽₈𝑀𝑇𝐵 + 𝛽₉𝐼𝑁𝐷 + 𝜀 (5) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽1𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽2𝐼𝐷𝑉 + 𝛽3𝑅𝐸𝑉𝐼𝐸𝑊 ∗ 𝐼𝐷𝑉 + 𝛽4𝑆𝐼𝑍𝐸 + 𝛽5𝐼𝑁𝑉 + 𝛽6𝐿𝐸𝑉 + 𝛽7𝑅𝑂𝐴 + 𝛽8𝑀𝑇𝐵 + 𝛽9𝐼𝑁𝐷 + 𝜀 (6) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 = 𝛼 + 𝛽₁𝑅𝐸𝑉𝐼𝐸𝑊 + 𝛽₂𝑀𝐴𝑆 + 𝛽₃𝑅𝐸𝑉𝐼𝐸𝑊 ∗ 𝑀𝐴𝑆 + 𝛽4𝑆𝐼𝑍𝐸 + 𝛽5𝐼𝑁𝑉 + 𝛽6𝐿𝐸𝑉 + 𝛽7𝑅𝑂𝐴 + 𝛽8𝑀𝑇𝐵 + 𝛽9𝐼𝑁𝐷 + 𝜀 3.2.1 Dependent variable

The variable REVIEW depicts whether quarterly reports in the observation have been reviewed and is hand-collected from quarterly reports and annual reports. The variable is

(16)

15 dichotomous which receives a 1 for the observation having a quarterly review and a 0 for the observation not having a quarterly review. This method is consistent with current literature (e.g. Bandyopadyay et al., 2007; Bédard and Courteau, 2015)

For an observation to receive a 1, all quarterly reports should have a review. However, in some cases not all review letters are made public (Ettredge et al., 1994), which could have the following results for reviewed reports; no review letters are attached to the quarterly report and the report is indistinguishable from a report without review. Alternatively, a review letter is attached to some, but not all, quarterly reports.

In the first option, no distinction can be made between the report having a review or not. In this case, the report should not contain a statement specifying that the report is not audited, resulting in the observation not being filled in. If the quarterly report does have a statement that it has been reviewed, but no review letter is attached, the observation still receives a 1.

The second option could have two explanations as well. Firstly, only some review letters are made public, while others are not. In this case, the other reports have no statement that no review has been performed, it is assumed that the other letters are not made public and the observation receives a 1. Secondly, only a part, but not all reports, have been reviewed. In this case, the reports have not been reviewed should include a statement stating the absence of a review and the observation receives a 0.

3.2.2 Independent variables

The variable QUALITY is measured by measuring the discretionary accruals (DA) of the organization. This is a widely used method that is used in a significant amount of researches (Dechow et al., 2010). I use the Modified Jones model (Dechow et al, 1995), which is an adapted version of the Jones model (Jones, 1991), to estimate the level of nondiscretionary accruals. Then I use the residual of the model to estimate the discretionary accruals (DA) as depicted in Model (7).

(7) 𝐷𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝑎𝑐𝑐𝑟𝑢𝑎𝑙𝑠 (𝐷𝐴) = 𝑇𝑜𝑡𝑎𝑙 𝑎𝑐𝑐𝑟𝑢𝑎𝑙𝑠 (𝑇𝐴) − 𝑛𝑜𝑛𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑖𝑜𝑛𝑎𝑟𝑦 𝑎𝑐𝑐𝑟𝑢𝑎𝑙𝑠 (𝑁𝐷𝐴)

First, consistent with Dechow et al. (1995), I use Model (8) to estimate the level of nondiscretionary accruals scaled by the total level of assets at t-1.

(17)

16 (8) 𝑇𝐴𝑖,𝑡 𝐴𝑖,𝑡−1 = 𝛼1( 1 𝐴𝑖,𝑡−1) + 𝛼2( ∆𝑅𝐸𝑉𝑖,𝑡−∆𝑅𝐸𝐶𝑖,𝑡 𝐴𝑖,𝑡−1 ) + 𝛼3( 𝑃𝑃𝐸𝑖,𝑡 𝐴𝑖,𝑡−1) + 𝜀𝑖,𝑡

Where ∆𝑅𝑒𝑣𝑖,𝑡 is calculated as revenues of firm i in year t minus revenues in year t-1

scaled by total assets at t-1; ∆𝑅𝑒𝑐𝑖,𝑡 denotes the net receivables of firm i in year t minus net

receivables in year t-1 scaled by total assets at t-1; 𝑃𝑃𝐸𝑖,𝑡 is calculated as gross property plant

and equipment of firm i in year t scaled by total assets at t-1.

Then I estimate the total accruals by the following model as used by Dechow et al. (1995):

(9) 𝑇𝐴_𝑖,𝑡 =∆𝐶𝐴𝑖,𝑡−∆𝐶𝐿𝑖,𝑡−∆𝐶𝑎𝑠ℎ𝑖,𝑡+∆𝑃𝐿𝑇𝐷𝑖,𝑡−𝐷𝑒𝑝𝑖,𝑡

𝐴𝑖,𝑡−1

Where ∆𝐶𝐴_𝑖,𝑡 denotes the change in current assets of firm i in year t; ∆𝐶𝐿_𝑖,𝑡 represents the change in current liabilities of firm i in year t; ∆𝐶𝑎𝑠ℎ_𝑖,𝑡 depicts the cash and cash equivalents of firm i in year t; ∆𝑃𝐿𝑇𝐷_𝑖,𝑡 represents the change in debt that is included in the current liabilities of firm i in year t, and 𝑑𝑒𝑝_𝑡 denotes the depreciation and amortization in year t. All components are scaled by the total level of assets at t-1.

Then, by substituting Model (8) and (9) into Model (7) the final equation to calculate the variable QUALITY is made, in which 𝛼₁, 𝛼₂, and 𝛼₃ are estimated regression coefficients. In the regression models the absolute of Model (10) is used.

(10) 𝑄𝑈𝐴𝐿𝐼𝑇𝑌 =𝑇𝐴𝑖,𝑡−(𝛼1+𝛼2(𝑅𝐸𝑉𝑖,𝑡−𝑅𝐸𝐶𝑖,𝑡)+𝛼3𝑃𝑃𝐸𝑖,𝑡)

𝐴𝑖,𝑡−1 + 𝜀𝑖,𝑡

3.2.3 Moderating variable

The variable CULTURE consists of the Hofstede national culture variables as used in Hofstede (1980) and is split into four dimensions. Respectively, power distance (PDI), uncertainty avoidance (UAI), individualism (IDV), and masculinity (MAS). The use of Hofstede’s culture dimensions is consistent with existing literature11_{. Within accounting}

(18)

17 research, the dimensions of uncertainty avoidance and individualism are the most used ones (Hooghiemstra et al., 2015; Hope, 2003).

3.2.4 Control variables

The first control variable I use is firm size (SIZE), calculated by the natural logarithm of the total assets, as firm size has been shown to be positively correlated to the having a review (Boritz and Liu, 2006). Similarly, other papers use firm size as a control variable in quarterly review literature (Bandyopadhyay et al., 2007; Ettredge et al., 2000; Manry et al., 2003). The second control variable I use is the inventory to total assets ratio (INV), as this can be seen as a proxy for the incremental costs associated with providing a review (Ettredge et al., 1994). The third control variable I use, consistent with other researchers (Bandyopadhyay et al., 2007; Manry et al., 2003), is the leverage ratio as measured by long-term debt as a percentage of the total assets (LEV), since leverage is associated with the choice of purchasing a review (Ettredge et al., 1994). Bédard and Courteau (2015) find that return on assets (ROA), a measure of profitability, is positively related to earnings quality. Therefore, the fourth control variable I use is return on assets (ROA). The fifth control variable I use is the market-to-book ratio (MTB), on the grounds that Boritz and Liu (2006) found that there is a negative association between growth opportunities and the choice to purchase a review. Lastly, I control for cross-sectional differences between industries (IND). For this I use industry dummy variables based on the single digit Standard Industrial Classification (SIC) codes. This results in the sample being classified into ten industry groups, each having a dichotomous variable of 1 if the industry is included in the group and a 0 otherwise.

3.3 Validity and reliability

To ensure the reliability of the data multiple data sources a used. I use primary, hand-collected, data for the measurement of REVIEW. Then, for the measurement of the cultural variables I use secondary data, which was collected by other researchers. Finally, I use tertiary data from the Datastream database for the control variables. By using multiple data sources I achieve to be not too dependent on one data source.

Furthermore, to make sure the data was not influenced by outliers, I winsorized the variables at the 1% and 99% levels, a procedure which is consistent with other researches that work with quantitative data (e.g. Bédard and Courteau, 2015; Dechow et al., 2010; Ernstberger et al., 2015).

(19)

18

4. Results

In this section the results of this paper are described starting with descriptive statistics. Next, the regression results are explained in detail, which are followed by robustness checks and additional analyses.

4.1 Descriptive statistics

In Table 3 the mean and standard deviation are shown for each variable. The table shows that 47% of the organizations from the sample were reviewed. The discretionary part of the total accruals is 43% in the sample. Furthermore, for the detection of potential multicollinearity problems I calculated the correlations between the variables as depicted in the pairwise correlation matrix in Table 3. In the matrix the significance levels are represented with one star (*) for significance at the 10% level; two stars (**) for significance at the 5% levels; and three stars (***) for significance at the 1% level.

From the correlation matrix can be concluded that most variables associate at the 1% level. However, while most correlations seem to be relatively small, there are four correlations that stand out. The correlation between IDV and PDI at -0.697 (p<0.01); MAS and UAI at 0.7225 (p<0.01); LEV and SIZE at 0.5011 (p<0.01); and ROA and SIZE at 0.5308 (p<0.01), respectively.

The negative correlation IDV and PDI is also noted by Hofstede (1983), who finds a correlation of -0.67. The correlation can be explained by the fact that individualism denotes the independence on the organization, while power distance indicates the dependence on a superior (Hofstede, 1983). The difference between the correlation that I found and the one that Hofstede found can be explained by the fact that my sample uses scores from five countries, whereas Hofstede calculates the correlation with data from 50 countries. Since the difference of the correlation is quite small, I conclude that the correlation is consistent with Hofstede’s work.

The variables MAS and UAI have the highest correlation coefficient. I expect this correlation to be this high due to the uneven distribution of culture scores by Sweden, with a score 5 on the masculinity index and a 29 on the uncertainty avoidance index12. When I omit Sweden from the sample, the correlation between MAS and UAI decreases to 0.418 (p<0.01).

(20)

19 This score is similar to the work of Hope (2003). However, since both variables are not simultaneously used in the regression models. I expect no problems to arise and keep Sweden included in the final sample.

Both correlations between SIZE and LEV and between SIZE and ROA are consistent with the research of Boritz and Liu (2006), who find similar results. Hope (2003) also finds a correlation between a firm’s leverage, and the size of the firm. Although she finds a correlation of 0.30, which is significantly lower. Since both correlations do not exceed the 0.6 level, no issues of multicollinearity should arise.

Furthermore, I excluded industry dummy variables 6 and 9 from the analysis since these industry dummies showed signs of multicollinearity in further analyses.

In Table 4, the final sample with the observations per country is depicted. In the final sample, most of the observations are from Canada, in which 61% of the observations had a review. This is consistent with the research of Bandyopadhyay et al. (2007). The observations from Germany had 17% of the reports reviewed. This number is close to the research of Kajüter et al. (2016) who found that 13% of the German reports in their sample were reviewed. The Netherlands included the least amount of observations with a total of 247. Furthermore, the fact that 2% of the observations from the Netherlands included a review stands out. For this reason, additional analyses will be done with the Netherlands excluded from the sample.

The national culture scores of Hofstede (1983) that were used can be found in Table 5 below. Four out of five countries have a score between 31 and 39 on the power distance, while Hong Kong scores a 68 on the index. Both Sweden and Hong Kong score the lowest on the uncertainty avoidance index with a 29, while Germany has the highest uncertainty avoidance with a score of 65. Most countries score quite high on the individualism index with the Netherlands and Canada having the highest index score of 80. To the contrary, Hong Kong has a low amount of individualism and only scores a 25. The amount of masculinity is quite low in Sweden as well as in the Netherlands with a score of 5 and 14 respectively. Germany has a higher amount of masculinity with an index of 66. Followed by Hong Kong and Canada with a score of 57 and 52 respectively.

(21)

20

4.2 Regression results

For the testing of the hypotheses I use an OLS regression of which the results can be found in Table 6 below.

For the first hypothesis I look at the interaction between QUALITY and REVIEW in column 1 of Table 6, which represents Model 1. This correlation depicts 0.186 (p<0.01), which shows support for hypothesis 1.

In the second to fifth regressions Model 3 to 6 are depicted, which represent the hypotheses of the moderating effect of culture. In these columns the main variables to look at are PDI*REVIEW, UAI*REVIEW, IDV*REVIEW, and MAS*REVIEW. Column 2 and 3 show evidence that supports hypothesis 2 and 3, with PDI*REVIEW having a coefficient of -0.008 (p<0.01) and UAI*REVIEW having a coefficient of 0.017 (p<0.01). In column 4 IDV*REVIEW depicts the coefficient -0.003 (p<0.097). While the coefficient is small and is approaching the line of non-significance, it does contradict hypothesis 4. The last column shows that the coefficient of MAS*REVIEW has a significant effect of 0.007 (p<0.01) on the relation between REVIEW and QUALITY, which seems to contradict hypothesis 5.

An interesting thing to note is that together with a moderating effect, all cultural dimensions themselves seem to be significantly related to earning quality. These effects seem to have an opposite effect on the moderating effect of all the cultural dimensions. Meaning that the cultural effects on earnings quality are diminished when a review is purchased by an organization.

Looking at the correlations between the control variables and earnings quality, the control variables behave consistent with current literature. Firm size, as measured by the natural logarithm of total assets, is highly significantly related to earnings quality. This is consistent with the research of Bédard and Courteau (2015), Boritz and Liu (2006), Boritz and Liu (2015), and Kajüter et al. (2015). Furthermore, the inventory ratio and leverage seems to be correlated with earnings quality at the 10% and 1% levels respectively, which was found by Kajüter et al. (2015) as well. Return on Assets seems to be highly correlated to earnings quality, which was found likewise by Bédard and Courteau (2015). Lastly, the market-to-book ratio correlates at 5% in my model, which is consistent with the work of Boritz and Liu (2006).

(22)

21

4.3 Robustness checks and additional analyses

In this section additional analyses are performed to ensure that the results are robust. First, I substitute the Modified Jones earnings quality variable with a variable used by Dechow and Dichev (2002). Second, I use a panel regression, since I use data with firm-year observations from 2005 to 2015. Finally, I exclude the Netherlands from the sample to make sure the sample is not biased due to the fact that the Netherlands only had 2% of their reports reviewed.

4.3.1 Earnings quality alternative measure

To make sure the results are robust, I substituted the Modified Jones (Dechow, 1995) estimator to measure earnings quality with the model used by Dechow and Dichev (2002) and performed a second OLS regression. Whereas the Modified Jones model measures discretionary accruals as a function of revenue, receivables, and plant property and equipment, the Dechow and Dichev model measures discretionary accruals by cash flows from the past, present, and future (Dechow and Dichev, 2002). The results of this regression can be found in Table 7.

It is interesting to see that R-Squared is significantly higher in this model, which means that the model has increased explanatory power. On the contrary, the relevant coefficients decrease drastically. Most coefficients seem to be significant, consistent with the first model, with a few exceptions.

In regression 6 REVIEW correlates significantly (p<0.01) with QUALITY, providing further proof of hypothesis 1. In regression 7 and 8, both moderating effects of PDI*REVIEW and UAI*REVIEW are significant at p<0.01, consistent with the regression model from Table 6. While the coefficients are significantly smaller than in the first regression model, both are highly significant and thus provide further proof for hypothesis 2 and 3.

When looking at regression models 9 and 10, the variables IDV*REVIEW and MAS*REVIEW have lost their significance (p<0.495, p<0.189). This contradicts hypothesis 4, but provides evidence for hypothesis 5. Furthermore, in regression model 10 REVIEW does not correlate significantly to QUALITY, which provides further evidence for hypothesis 5.

(23)

22 The control variables SIZE, ROA, and LEV are again significant in all models (with the exception of ROA not being significant in regression 10 (p<0.756)). Both INV and MTB have lost their significance in this model.

4.3.2 Panel regression

In addition to the substitution of QUALITY, a fixed-effects GLS panel regression has been performed using the original variables, which groups the various firm-year observations to the specific organization. The results from this regression can be found Table 8 below. In this regressions, R-squared is similar to the regressions in Table 6.

In regression 11 the correlation between REVIEW and QUALITY is significant once more. However, the correlation is now significant at the 5% level (p<0.011). This provides more evidence that hypothesis 1 is correct. Furthermore, in regressions 12 and 13 both the moderating effects of power distance and uncertainty avoidance are again significant at -0.007, p<0.01 and 0.013, p<0.01, consistent with the original model and the substitution model.

The moderating effect of individualism is not significant at -0.002, p<0.488, which is consistent with the substitution model from Table 7 and contradicts hypothesis 4. Finally, in regression 15 the moderating effect of masculinity is significant at 0.005 (p<0.01). This is consistent with the original model, but contradicts the substitution model.

The control variables SIZE, LEV, ROA, and MTB are significantly related to earnings quality in this model, which is consistent with the original model and the substitution model (with the exception of MTB).

4.3.3 Exclusion of the Netherlands

As noted in section 4.1 additional analyses have been done with the exclusion of the Netherlands from the sample to make sure the dataset is correct. However, no significant changes have been found when interpreting the results with the exception of IDV*REVIEW, which increased in significance from p<0.097 to p<0.042 in the OLS regression. However, when substituting the earnings quality variable with the Dechow and Dichev (2002) calculated variable, all variables behave similarly to the analyses done with the full sample. As a result I interpret that no significant changes occur when excluding the Netherlands from the sample size.

(24)

23

5. Conclusion

5.1 Interpretation of the results

In this paper I look at the association between the decision to purchase a voluntary review and its effect on earnings quality. Furthermore, I look at the moderating effects of culture on this association. To do this I draw from a sample of 6586 firm-year observations from five countries in the period of 2005 to 2015. I find that the decision to purchase a voluntary review is positively correlated to earnings quality. This suggests that voluntary reviews do influence the quality of quarterly reports.

I conclude from multiple analyses that the cultural dimension power distance of Hofstede (1980) negatively moderates this correlation and that uncertainty avoidance positively moderates this correlation. The negative moderating effect of power distance could be explained by the fact that societies with a high amount of power distance are characterized by having a high amount of secrecy (Gray, 1988), which could make it more difficult to detect earnings management. The positive moderating effect of uncertainty avoidance could be explained by the fact that uncertainty avoidance is correlated to conservatism (Doupnik and Tsakumis, 2004). This could mean that the auditor could be more conservative in carrying out the review, resulting in less allowance for earnings management.

The analyses predominantly show that individualism does not have a significant effect on the correlation between a review and earnings quality with the coefficient in the original analysis being on the line of non-significance. However, from the literature I infer that both a positive and negative effect could exist. It is possible that the model that I use is not correct to capture such an effect or that no effect exists at all. More research could be done to look into this specifically.

I find conflicting evidence regarding the moderating effect of masculinity. From these results I conclude that more research with distinctive models should be done before any assumptions can be made about this effect.

In all regression models that I build I found that firm size, leverage, and return on assets significantly influence the quality of earnings, which confirms the work of Bédard and Courteau (2015), Boritz and Liu (2006), Boritz and Liu (2015), and Kajüter et al. (2015).

(25)

24 The results of this paper do have several theoretical implications. First of all, by finding a positive link between the voluntary purchase of a review and earnings quality, this paper adds to the debate of whether a review adds value to quarterly reports by showing that the review does have benefits. Secondly, this paper shows support for the agency theory. In section 2.1 I talked about the fact that an option to reduce information asymmetry is to have accounting rules and standards in place (Healy and Palepu, 2001). This paper shows that the enforcement of accounting standards by a reviewer results in better earnings quality, and thus, a reduction of information asymmetry. Thirdly, this paper shows further support of the theory of Gray (1988) that national culture does effect accounting systems by proving that the cultural dimensions power distance and uncertainty avoidance do moderate the effect between reviews and earnings quality. Finally, this paper adds to the literature by confirming several firm characteristics that influence the quality of earnings.

This paper also has implications for managers and regulators. In the introduction I talked about how quarterly reports are not always perceived to be reliable (Boritz and Liu, 2006). Hence, by purchasing a review, managers can signal to shareholders that their report is of higher quality and more reliable. Bandyopadhyay et al. (2007) have found a similar effect in their research. In some countries (e.g. United States) regulators mandate that organizations purchase a review. However, according to Bédard and Courteau (2015) there is still no consensus on whether reviews add value to the report or not. The results of this paper can help regulators in their argument to make reviews required.

5.3 Limitations and further research

There are several limitations to this research. First of all, no distinction has been made between timely reviews and retrospective reviews in this paper as these were indistinguishable with the data collection method that was used. Further research using a data collection method in the form of questionnaires could solve this problem. Second, the national culture scores of Hofstede (1980) are quite old and differ 25 to 35 years with observations that were used in this paper. This could mean that several culture scores could have changed in the meantime. However, Hope (2003) argues that culture changes slowly over time and that the differences would be minimal. Third, this paper uses two models to estimate earnings quality through discretionary accruals. In the paper of Dechow et al. (2010) several methods to measure discretionary accruals are summarized. Future research could use these models to measure earnings quality in other ways to further validate or reject my findings.

(26)

25 Aside from the recommendations for further research in last section I suggest the following; Since I found conflicting results regarding the moderating effect of masculinity, I recommend that follow up study that looks further into this effect. Furthermore, I use cultural data collected from five countries. Future research could include more cultural data from alternative countries to endorse the results I found.

(27)

26

References

Balsam, S., Bartov, E., & Marquardt, C. (2002). Accruals management, investor sophistication, and equity valuation: evidence from 10-Q filings. Journal of Accounting Research. 40(4), 987-1012.

Bandyopadhyay, S. P., Boritz, J. E., & Liu, G. (2007). Voluntary assurance on interim financial statements and earnings quality. Canadian Academic Accounting Association. 2008 Annual conference paper.

Bédard, J., & Courteau, L. (2015). Benefits and costs of auditor’s assurance: evidence from the review of quarterly financial statements. Contemporary Accounting Research. 32(1), 308-335.

Boritz, J. E., & Liu, G. (2006). Why do firms voluntarily have interim financial statements reviewed by auditors? Unpublished manuscript, University of Waterloo and Ryerson University.

Brown, L. D., & Caylor, M. L. (2005). A temporal analysis of quarterly earnings thresholds: propensities and valuation consequences. The Accounting Review. 80(2), 423-440. Brown, L. D., & Pinello, A. S. (2007) To what extend does the financial reporting process curb

earnings surprise games? Journal of Accounting Research, 45(5), 947-981.

Dechow, P., & Dichev, I. (2002). The quality of accruals and earnings: the role of accrual estimation errors. The Accounting Review. 77(s-1), 35-59.

Dechow, P., Ge, W., & Schrand, C. (2010). Understanding earnings quality: a review of their determinants and their consequences. Journal of Accounting and Economics. 50(2), 344-401.

Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995). Detecting earnings management. The Accounting Review. 70(2), 193-225.

Doupnik, T. S., & Tsakumis, G. T. (2004). A critical review of tests of Gray’s theory of cultural relevance and suggestions for future research. Journal of Accounting Literature. 23, 1-48.

Doupnik, T. S. (2008). Influence of culture on earnings management: a note. Abacus. 44(3), 317-340.

Ernstberger, J., Link, B., Stich, M., & Yolger, O. (2015). The real effects of mandatory quarterly reporting. The Accounting Review. 92(5), 33-60.

Ettredge, M., Simon, D., Smith, D., & Stone, M. (1994). Why do companies purchase timely quarterly reviews? Journal of Accounting and Economics. 18(2), 131-155.

(28)

27 Ettredge, M. L., Simon, D. T., Smith, D. B., & Stone, M. S. (2000). The effect of the external accountant’s review on the timing of adjustments to quarterly earnings. Journal of Accounting Research. 38(1), 195-207.

Fama, E. F., & Jensen, M. C. (1983). Separation of ownership and control. Journal of Law & Economics. 26(2), 301-325.

Francis, J., Nanda, D., & Olsson, P. (2008). Voluntary disclosure, earnings quality, and cost of capital. Journal of Accounting research. 46(1), 53-99.

Gleason, K. C., Mathur, L. K., & Mathur, I. (2000). The interrelationship between culture, capital structure, and performance: evidence from European retailers. Journal of Business Research. 50(2), 185-191.

Gray, S. J. (1988). Towards a theory of cultural influence on the development of accounting systems internationally. Abacus. 24(1), 1-15.

Gray, S. J., & Vint, H. M. (1995). The impact of culture on accounting disclosures: some international evidence. Asia-Pacific Journal of Accounting, 2(1), 33-43.

Griffin, P. A. (2003). Got information? Investor response to Form 10-K and Form 10-Q EDGAR filings. Review of Accounting Studies. 8(4), 433-460.

Healy, P. M, Palepu, K. G. (2001). Information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature. Journal of Accounting and Economics. 31(1-3), 405-440.

Hill, C. W., & Jones, T. M. (1992). Stakeholder-agency theory. Journal of Management Studies. 29(2), 131-154.

Hofstede, G. (1980). Culture’s consequences. Sage Publications. 1980.

Hofstede, G. (1983). National cultures in four dimensions: A research-based theory of cultural differences among nations. International Studies of Management & Organization. 13(1-2), 46-75.

Hofstede, G., & Bond, M. H. (1988). The Confucius connection: from cultural roots to economic growth. Organizational dynamics, 16(4), 5-21.

Hooghiemstra, R., Hermes, N., & Emanuels, J. (2015). National culture and internal control disclosures: a cross-country analysis. Corporate Governance: An International Review. 23(4), 357-377.

Hope, O. K. (2003). Firm-level disclosures and the relative roles of culture and legal origin. Journal of International Financial Management and Accounting. 14(3), 218-248. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: managerial behaviour, agency

(29)

28 Johnson, N. B., & Droege, S. (2004). Reflections on the generalization of agency theory:

cross-cultural considerations. Human Resource Management Review. 14(3), 325-335. Jones, J. J. (1991). Earnings management during import relief investigations. Journal of

Accounting Research. 29, 193-228.

Kajüter, P., Klassmann, F., & Nienhaus, M. (2016). Do reviews by external auditors improve the information content of interim financial statements? The International Journal of Accounting, 51(1), 23-50.

Kanagaretnam, K., Lim, C. Y., & Lobo, G. J. (2011). Effects of national culture on earnings quality of banks. Journal of International Business Studies. 42(6), 853-874.

Kirkman, B. L., Lowe, K. B. & Gibson, C. (2006). A quarter century of culture’s consequences: a review of the empirical research incorporating Hofstede’s cultural value framework. Journal of International Business Studies. 36(3). 285-320.

Manry, D., Tiras, S. L., & Wheatley, C. M. (2003). The influence of interim auditor reviews on the association of returns with earnings. The Accounting Review. 78(10), 251-274. Nabar, S., & Boonlert-U-Thai, K. K. (2007). Earnings management, investor protection, and

national culture. Journal of International Accounting Research. 6(2), 35-54.

Palmrose, Z. V., Richardson, V. J., & Scholz, S. (2004). Determinants of market reactions to restatement announcements. Journal of Accounting and Economics. 37(1), 59-89. Sivakumar, K., & Nakata, C. (2001) The stampede toward Hofstede’s framework: avoiding the

sample design pit in cross-cultural research. Journal of International Business Studies. 32(3), 555-574.

Zarzeski, M. T. (1996). Spontaneous harmonization effects of culture and market forces on accounting disclosure practices. Accounting Horizons. 10(1), 8-37.

(30)

29

Table 1

Table 1: Sample size

Observations with review data 10356 No data of QUALITY 173 No data of SIZE 0 No data of INV 3327 No data of LEV 42 No data of ROA 84 No data of MTB 144 Total sample 6586

(31)

30

Table 2

Table 2: Variable definitions

Variable Description

REVIEW Dichotomous variable: 1 for a report being reviewed, 0 for not being reviewed.

QUALITY Earnings quality measured by discretionary accruals using the modified Jones model

PDI Hofstede’s cultural scores regarding Power distance

UAI Hofstede’s cultural scores regarding Uncertainty avoidance IDV Hofstede’s cultural scores regarding Individualism

MAS Hofstede’s cultural scores regarding Masculinity SIZE LN Total assets

INV Inventories as a percentage of total assets

LEV Long-term debt as a percentage of total assets

ROA Return on assets

MTB Market value of equity as a percentage of book value of equity

(32)

31

Table 3

Table 3: Correlation matrix

MEAN STD.

DEV. QUALITY REVIEW PDI UAI IDV MAS SIZE INV LEV ROA MTB

QUALITY 0.432 1.739 ₁ REVIEW 0.468 0.499 -0.003 1 PDI 38.745 8.203 0.01 0.102*** 1 UAI 48.864 12.117 -0.010 -0.249*** -0.293*** 1 IDV 71.846 13.597 0.113*** 0.090*** -0.680*** 0.214*** 1 MAS 48.317 19.651 0.043*** -0.069*** 0.304*** 0.723*** -0.089*** 1 SIZE 5.985 2.694 -0.250*** 0.196*** 0.082*** -0.134*** -0.333*** -0.177*** 1 INV 105.363 125.776 -0.075*** -0.148*** -0.059*** 0.086*** -0.180*** -0.034*** 0.155*** 1 LEV 1350174 4480240 -0.054*** 0.081*** 0.103*** -0.137*** -0.207*** -0.117*** 0.501*** -0.046*** 1 ROA -7.003 41.776 -0.284*** 0.060*** -0.003 0.031** -0.149*** -0.025** 0.531*** 0.119*** 0.090*** 1 MTB 2.162 4.797 -0.085*** 0.012 -0.067*** -0.053*** -0.008 -0.103*** 0.054*** -0.036*** -0.005 0.142*** 1

(33)

32

Table 4

Table 4: Observations per country

Country No review Review Total observations

Canada 1287 (39%) 2035 (61%) 3322 (100%) Germany 1424 (83%) 292 (17%) 1716 (100%) Hongkong 158 (37%) 270 (63%) 428 (100%) The Netherlands 241 (98%) 6 (2%) 247 (100%) Sweden 394 (45%) 479 (55%) 873 (100%) Total 3504 (53%) 3082 (47%) 6586 (100%)

(34)

33

Table 5

Table 5: Hofstede’s national culture scores

Country PDI UAI IDV MAS

Germany 35 65 67 66

Sweden 31 29 71 5

The Netherlands 38 53 80 14

Canada 39 48 80 52

Quarterly Audit Reviews and Earnings Quality: International Evidence