Amsterdam Business School
The effect of social capital on audit fee
Focusing on ethnicity and politicsName: Christiaan Koopmanschap
Student number: 10853944
Thesis supervisor: Dr. G. Georgakopoulos
Date: 17th June 2016
Word count: 17801, 0
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by student Christiaan Koopmanschap who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the content.
Abstract
This study explores whether non-financials have an effect on audit fee. More specifically, this study examines the association between ethnics, politics and audit fee. This study combines sociology literature with accounting literature for an empirical research that uses proxies concerning ethnicities and the red/blue social divide in the United States of America and finds evidence that both ethnic and political preferences are associated with audit fees. The data used in this regards are the financial years 2009-2014. This thesis however finds no evidence that companies residing in “African-American states” pay more audit fee, in the contrary this paper finds the opposite, however this does not necessarily mean that African-Americans do not get price discriminated concern audit fees.
Keywords: Ethnics, Politics, Blue states, Red states, Social cleavage, Risk, Trust, Audit fee.
1 INTRODUCTION --- 5 1.1 GENERAL NOTIONS --- 6 1.2 RESEARCH QUESTION --- 8 1.3 FINDINGS --- 8 1.4 CONTRIBUTION --- 8 1.5 STRUCTURE --- 9 2 LITERATURE REVIEW --- 11
2.1 THE DETERMINANTS OF AUDIT FEE --- 11
2.2 SOCIAL CAPITAL --- 16
2.3 RISK, TRUST AND ITS EFFECT ON AUDIT FEE --- 17
2.4 MINORITY GROUPS AND THEIR EFFECT ON AUDIT FEE --- 18
2.5 POLITICAL PREFERENCES AND SOCIAL CLEAVAGE WITHIN THE USA--- 20
2.6 HYPOTHESIS BUILDING --- 21 3 RESEARCH METHODOLOGY --- 23 3.1 RESEARCH METHOD --- 23 3.2 RESEARCH DESIGN --- 24 3.2.1Variables of interest --- 25 3.2.2 Control variables --- 28 4 RESULTS --- 32
4.1 DATA PROCESSING AND DESCRIPTIVE STATISTICS --- 32
4.2 DATASET TESTS --- 36
4.2 MAIN RESULTS --- 37
5 CONCLUSION & DISCUSSION --- 42
REFERENCES --- 45
APPENDIX I - MAIN VARIABLE LIST --- 49
APPENDIX II - USA ELECTION RESULTS 2000-2012 --- 51
APPENDIX III - CONSOLIDATED ELECTION RESULTS 2000-2012 --- 52
APPENDIX IV - SPECIFICATION OF DATA CREATION --- 53
APPENDIX V - INITIAL DATABASE TESTS --- 56
1 Introduction
Prior research found that social capital is associated with various accounting and control topics, like corporate misbehavior, financial development, stock markets, corporate decision making and audit fee (Grullon, Kanatas, and Weston, 2010; Guiso, Sapienza, and Zingales 2004, 2005, 2008; Hilary and Hui, 2008; Jha & Chen, 2014). A new addition to this stream of literature is that social capital is directly associated with audit fee (Jha & Chen, 2014). This new finding opens the way to research concerning many more non-financial determinants that might influence audit fee. This thesis is a direct follow-up of the Jha & Chen (2014) paper, by being an empirical study on how certain non-financial attributes affect the fee that companies have to pay for the audit of their financial statements. The main topics of this thesis are ethnicity and politics.
The past recent years, the American government has gained scrutiny over the fact that there is still a lot of racism in the United States of America (USA). The USA “has a historical legacy of racial sensitivity that has consistently favored racial elites at the expense of racial non-elites.” (Coates, 2008). Also authors are divided upon the existence of a cleavage in the USA population concerning voter’s political preferences. Where Monson and Mertens (2011) claim that there are social differences between the two groups of voters, while Doyle (2006) claims that this difference does not exist.
Since 2008 the president of the United states of America (USA) is Barack Obama. Barack Obama is the first non-Caucasian president of the USA and his election has placed the issue of racism in the center of political debate (Bobo & Charles, 2009). Therefore research within this issue might provide a greater insight within racial indifferences in the USA and therefore seems well timed.
This thesis uses data concerning American companies, therefore excluding Canadian companies, or companies which report in Canadian Dollars. Furthermore the data used concerns only fiscal years 2009-2014. Fiscal year 2015 has been excluded since companies currently have their 2015-2016 fiscal years, there also are companies that are having their 2015 audits and therefore this data is not available yet. This timespan has been chosen due to multiple reasons. The first is availability of the U.S. Census Bureau data, which has an official measurement of the USA demographics once every ten years. The last official measurements was in 2010. The second reason for this timespan is the “blueness” index, by Shin & Webber (2014), which is index derived from multiple elections in the USA. The index date however is 2014. Deviating too much from this date might influence the research too much in a negative way, by making wrong assumptions. A final reason for using this time span (2009-2014) is that this timespan, aligns with the election of Barack
Obama (2008-2016). As mentioned by Bobo & Charles (2009), his election has put racism is the center of political debate. Obama is officially president since January 2009. Using a longer time period with other presidents might influence the results. Based on above limitations of using earlier time-periods, the time span of this thesis is 2009-2014.
These facts and findings contribute to the basis for my thesis, which focusses on the effect of non-financial determinants of audit fee, and especially the effects of ethnic differences and political groups.
1.1 General notions
Audit fees are fees that a company pays an external auditor in exchange for performing a financial audit. These fees can be affected by multiple elements as described in a string of research done by various researchers. (Cobbin, 2002; De Simone, Ege and Stomberg, 2015; Gonthier-Besacier & Schatt, 2007; Hay, 2013; Hay, Knechel and Wong, 2006; Simunic, 1980; Stenley, Doucouliagos and Jarrell., 2008). These researchers commonly base their assumptions on Simunic on the assumption that audit fee is based on price*quantity and that certain determinants influence these two components (Simunic, 1980). Often used determinants which are recognized by most researchers are (1) audit/auditee size, (2) auditee complexity, (3) perceived (litigation) risk and (4) resources used by the auditee for internal control systems. (Cobbin, 2002; Gonthier-Besacier & Schatt, 2007; Hay, 2013; Hay et al., 2006; Simunic, 1980). Except for these determinants there has been more research claiming that there are more determinants concerning economic attributes. (Cobbin, 2002; Gonthier-Besacier & Schatt, 2007; Hay et al. 2006; Hay, 2013). Lately there have also been claims that non-financial determinants can have effect on audit fee (Hay, 2013; Stenley et al., 2008; Jha & Chen, 2014). This paper mostly focusses on the findings of Jah & Chen (2014), which show that other types of determinants (in this case social capital) also have an effect on audit fee. Social capital can be defined as “the norms and the networks that facilitate collective action. A high social-capital region has individuals with a greater propensity to honor an obligation and a greater mutual trust within a much denser network, all of which facilitate collective action.” (Jha & Chen, 2014). Guiso, Sapienza & Zingales (2004) define social capital as the mutual level of trust in society. Above definitions in general claim that social capital, has much to do with trust and collectives. According to Giddens (1990) trust and risk are intertwined and therefore trust can be associated with risk. Also “game-theoretic rationality1 prescribes that trust decisions should depend on the potential risk (egocentric costs and benefits) and the probability of reciprocity
(derived from the trustee's temptation to defect)” (Evans & Krueger, 2011, p. 171). Trust is “a psychological state comprising the intention to accept vulnerability based upon the positive expectations of the intentions or behavior of another” (Rousseau, Sitkin, Burt, & Camerer, 1998, p. 395). Therefore, Giddens notion of trust can be associated with risk. This paper tries to connect the mentioned literature by linking risk, trust and audit fee with ideology (politics) and ethnicity. Since (due to perceived risk) religion seems to have an effect on audit fee (Jha & Chen, 2014), it is possible that these factors also have effect on audit fee, strengthening that theory.
This paper focuses on the USA. Multiple reasons have led to this decision. First of all, because of data-availability. Much of the data available at the University of Amsterdam databases concern firms registered at the USA stock exchanges. Furthermore, detailed information concerning the different states of the USA are available either in the different databases or are available on the world wide web. The second reason as described by Dutter (2013);
“Traditionally, a 'plural' society has been defined as one that is ethnically diverse and whose ethnic divisions are politically relevant. Arguably, the US falls within the general purview of this definition and its implications, especially with regard to African Americans, Asian Americans, Hispanic Americans and non-Hispanic, European Americans.” (Dutter, 2013 pp. 50).
The ethnic diversity in the USA has increased rapidly, as the shares of Caucasians fell from 83% in 1970 to 67% in 2005. (Martin & Midgley, 2006). The biggest minority in the USA are the African Americans consisting of almost 13% of the total population. This is followed by Asian-American’s consisting of 4.43% (Central Intelligence Agency, 2015). This makes the USA an interesting context for research on ethnic groups. This also reduces noise given that the different ethnic groups can be differentiated more easily. A third and final reason to conduct this research focused on the USA is the split in Republicans and Democrats. This might be explained by the differences in social capital. “Red states” are states that favor Republicans and “blue states” favor Democrats. In terms of social capital, the two groups can be divided into different social groups in terms of their capital and risk that might be associated with those groups.
Figure 1 shows a recap of the theoretical background.. The most important determinant in this paper is risk. Increasing or decreasing the factors in this figure should have a significant influence on audit fee.
Figure 1: Recap of theoretical background
1.2 Research question
The aim of this paper is to extend on prior research (Jha & Chen, 2014; Stenley et al., 2008), researching different non-financial variables and mostly focusing on ethnicity and politic backgrounds in different regions of the USA. The main goal is to lay a basis for future research concerning the interactions between sociology and audit fee and thereby answering to the call of Jha & Chen (2014), for more research in non-financials. Therefore, the research question is; To what extend do cultural and political background influence the audit fee in the USA?
1.3 Findings
The findings suggest that cultural and political background affect audit fee. The regressions show that minority regions pay more audit fee than Caucasian regions. The regressions also show that “blue” states pay less audit fee than other states within the USA. The findings however also find that companies residing in African-American regions pay less audit fee than other states in the USA. This final outcome however, might be biased due to a large amount of observations being centered in California.
1.4 Contribution
This research provides a theoretical contribution by further elaborating the effects that social capital/non-financial attributes have on audit fee (Stenley et al., 2008; Jha & Chen, 2014). This is done in multiple ways. Firstly, by examining the effect of state differences concerning ethnic groups and its perceived risk, which might cause audit fees to increase. Within the analysis of ethnic differences, it also examines the effect of price discrimination of minority groups. Minority groups is in the case a broad definition of different kind of ways a minority can exist as being the lesser (in size) of multiple groups in a given space/context, however in this case being minority ethnic groups. It explains how being a minority might have a negative association with audit fee. This has as far as I currently know not been researched before. It further increases our knowledge in what
Price (Hourly Fee)
Quantity
(Hours) Audit Fee
(PxQ) 1. Auditee size
2. Auditee complexity
3. Resources used by auditee for control systems
4. Perceived risk Social capital: 1. Ethnicity 2. Religion 3. politics 4. majority “racism”
This thesis also has an empirical contribution by answering to the call of Jha & Chen (2014), who state that further research can be done on the effects of non-financials on audit fee. Is also answers to the call of Stenley et al. (2008) who state that further research has to be done regarding to audit fee (preferably by meta-regressions). This paper aims to contribute by adding to existing literature of audit fee, but also contributes to research concerning Socio-economics, American studies and Human behavior. This is achieved by doing database research concerning the effect of social capital on audit fee in the American society. This thesis is one of the first papers to actively make connections between sociology and audit fee. This is done by looking at research by some of the most important sociologists like Bourdieu (1987) and Giddens (1990), but also uses research done by other respected authors like Baxter (2009), Beck (1992), Evans & Krueger (2011), Guiso et al. (2004), Rousseau, Sitkin, Burt & Camerer (1998), Schmid, Al Ramiah & Hewstone (2014) and Yeung & Rauscher (2014). This paper therefore contributes by connections multiple streams of research to each other and might be the beginning of more research towards sociology in accounting, as which is already done in sustainability and accountability research.
As mentioned before Barack Obamas’ election has placed the issue of racism in the center of political debate. (Bobo & Charles, 2009) Therefore research within this issue might provide a greater insight within racial indifferences and therefore also contributes in the ongoing debate of discrimination within the USA.
1.5 Structure
The remainder of this thesis will consist of a literature review, which goes in further detail about the determinants of audit fee and why certain groups within society might pay a higher audit fee than others. The final part of the literature review will consist of the building of hypotheses, giving clear fundamentals for the expected associations.
After the literature review there is a methodology section explaining what kind of research this is, how data has been collected, what variables have been used and why, whereas this section will also go in further detail about the regression model.
After the methodology section there is a findings section. This section starts with how the dataset has been manipulated within Stata and all relevant sources and explanations for the choices made while manipulating and collecting the data. This section will then elaborate on the findings and how these finding have been found (univariate, multivariate, ANOVA, etc.). This section will give a brief explanation of all results that have been found. This section will thereby also give a
clear view of the results and how they connect with the hypotheses as stated in the hypotheses development section.
The final part of this thesis consists of a conclusion and discussion. Here there will be a short recap, elaboration of the results, strengths and weaknesses of this thesis, if this thesis has indeed contributed to the causes mentioned before and finally a section about further research that can be done.
2 Literature review
This section of the paper contains a review of the literature and how it contributes to the hypotheses that are developed also in this section. It will first further elaborate on the traditional determinants of audit fee by Simunic (1980) and other authors that further elaborate on his thinking. After the traditional determinants there will be more focus on how other factors might influence audit fee. There is a section about social capital, which mostly focusses on the findings of Jha & Chen (2014). After this there is a section about how risk and trust influence audit fee, by explaining the concepts of two important sociologists, Beck (1992) and Giddens (1990). These concepts are then transferred into accounting literature using how risk influences audit fee (Bell, Landsman & Schackelford, 2001). When above concepts are clear there is more extensive research towards minority research, explaining how the risk concept might transfer into minority groups. This will be followed by literature about the USA, its political parties and how social cleavage within the political landscape might affect audit fee. The final part of the literature review will contain a section about how the hypotheses have been build, using all the literature described in the mentioned sections.
2.1 The determinants of audit fee
“Audit fee is the product of unit price and the quantity of audit services demanded by the management of the audited company …, cross-sectional differences in fees represent either the effect of quantity differences of price differences” (Simunic, 1980, p. 161-162). In prior literature by Simunic (1980) it has been stated that four important determinants have an effect on audit fee. The first one being the size of the auditee. If the auditee is bigger in size, more evidence has to be collected. This also causes an increase in audit complexity. Secondly the resources used by the auditors to complete the audit, being the hours needed to complete the audit. This has an effect since audit prices are being calculated by a general PxQ formula. The quantity used is man-hours since, audit companies can be labeled as service firms that provides in knowledge, which is done with human capital. Thirdly the resources used by the auditee, to maintain their control systems. This will reduce the chances of errors in the accounting systems thus reducing the risk that is perceived by the auditor. Simunic (1980) also claims that an increase in resources used by the auditee, will mean a decrease in resources used by the auditor, thus leading to a lower audit fee. The fourth determinant is the fee per hour, including provisions for profit, being the price in the general PxQ formula (Simunic, 1980). Research by Cobbin (2002) states that audit fee is determined by risk, complexity of the firm and firm size. The audit risk of the company is
associated with the nature of the business and its control environment. The complexity of the company is based on the number of unusual transactions of the auditee and the internal control quality. The size can be measured by using total assets, turnover or total employment costs. These three determinants together are the most significant when the audit firm is pricing the audit. Also in research by Gonthier-Besacier & Schatt (2007) is stated that the determinants of audit fee are the size of the auditee, the perceived risk and the complexity of the auditee. Figure 2 contains a brief overview of the main determinants, described in this section.
Figure 2: Determinants of audit fee
New streams of research like Jha & Chen (2014) however, seem to focus on other determinants that affect audit fee, either by itself or by affecting the four determinants in Figure 2. Jha & Chen (2014) claim that social capital has an effect on audit fee. The primary cause is that the different variables for social capital used by Jha & Chen (2014) would increase/decrease the perceived risk, resulting in a risk premium. This association is shown in Figure 3.
Figure 3: Social capital and audit fee
Other research claims that there are many variables that can be associated with audit fee. A prominent paper in audit fee research is the Hay et al. (2006) paper. This paper summarizes often used determinants used in 147 scientific papers concerning audit fees and also elaborates on which variables commonly represents those determinants. They claim that when looking at independent variables often used variables are; size, complexity, inherent risk, profitability, leverage, form of ownership, internal control, governance, industry, auditor quality, auditor tenure, auditor location, report lag, busy season, audit problem, non-audit services and reporting. In figure 4 which can be seen below, the most commonly used / important determinants per section are shown.
Price (Hourly Fee) Quantity (Hours) Audit Fee (PxQ) 1) Auditee size 2) Auditee complexity 3) Perceived (litigation) risk
4) Resources used by auditee for control systems
Perceived risk
Social Capital Price
(Hourly Fee)
Audit Fee (PxQ)
Section Determinant Amount of variables used
Client attributes Complexity
Size Inherent-risk Profitability Leverage 240 147 129 106 90
Auditor attributes Auditor quality
Auditor tenure
157 49
Engagement attributes Audit problems Busy season Non-audit services Report lag 79 35 26 12 Figure 4: Most used control variables (based on Hay et al. (2006))
As already stated by Simunic (1980), more complex clients have harder and more time consuming audits. Therefore, it is generally argued that more complex firms pay higher audit fees. Common used variables for auditee complexity are the number of segments, foreign subsidiaries (County presence). Hay et al. (2006), claim in their meta-analysis that the number of subsidiaries is the most significant proxy for auditee complexity. However, the number of subsidiaries is not publicly available through databases, at least not the ones provided by the Universiteitsbibliotheek van Amsterdam (UBA). There is the possibility to hand collect this data. However due to the large size of observations that is expected, this is not an option and therefore will be excluded from this thesis. Due to supply and demand busy seasons tend to increase audit pricing. In busy times auditors are more scarce, which means a shift in the supply and demand curve, which suggests that the pricing of the audit should increase. Hay et al. (2006) also state that the Busy season variable is a commonly used control variable. However, since like many recent papers like Gonthier-Besacier & Schatt (2007) do not use this variable at all or like Cobbin (2002) and Jha & Chen (2014) do not clearly define how they measure and therefore there is no recent way to measure this, which is why this variable has been excluded from the regressions in this thesis. Another important determinant is profitability. Auditees that are making a loss, expose the auditor to increased litigation risk, since the auditor might be held responsible for losses occurred by stakeholders of the company, which is also mentioned in the deep pockets theory, which will be mentioned in greater detail later on (Sun & Liu, 2011). Auditors also face the risk their client does not pay for the provided services
(Simunic, 1980). This is also supported by more recent research (Hay, 2013; Hay et al., 2014; Jha & Chen, 2014. Jha & Chen (2014) use the variable Loss, as is also supported by Hay et al. (2006) in their meta-analysis, who claim that loss in another proxy related to risks. Another often used variable for profitability is the ROA ratio. Also bad leverage generally increases the loss exposure of the auditor, thereby being very likely to profitability and increasing audit fees. Common measures of leverage are debt divided by total assets and the current/quick-ratio. Hay et al. (2006) find that debt divided by total assets is the most significant variable, concerning leverage. The debt-equity ratio will therefore be used as a control variable in this research. There however also are a lot of insignificant results concerning leverage (Hay et al. 2006). So although it is an often used variable, a direct relation might not exist. Another general used control variable is the Big4 auditor dummy which are significantly positively related to audit fee. This price increase is also called the Big 4 premium. Big4 companies in general have higher prices than other accountants due to superior quality. This superior quality can also be measured with other variables, but big 4 auditor seems to be the most commonly used proxy for auditor quality (Hay et al., 2014).
Report lag is a variable that is added in order to examine potential audit problems during the audit. A longer period between balance date and the auditors’ opinion might indicate that problems have arisen during the audit (Hay et al., 2006). Hay et al. (2006) also found that this variable is significantly (positively) related to audit fee. Hay et al. (2006) also state that many audit fee papers use industry as a control variable. This is due to the fact that different industries have a different complexity, which requires more work. This is recognized by a wide range of audit fee researchers, which mostly use utilities or financial institutions as a variable. In this case companies can be labeled as a financial institution by using a dummy 0,1 variable which state that 0 is not a financial institution and 1 that it is a financial institution. However, it is also not unlikely to exclude the financial sector altogether. Audit problems might increase audit fee’s due to increased risk. As stated before audit fee consists of a risk premium. This means that if there are more audit problems, it is likely to increase audit fee. The audit opinion is a proxy for audit problems. And it is expected that audit fees rise if the opinion is qualified or modified. (Hay et al. 2006)
Other authors
City-level industry specialization of the auditors and their economies of scale have an impact on audit fees. (Fung, Gul & Krishnan, 2012). Also audit competition has an effect on audit fee (Newton, Wang, and Wilkins, 2013), due to the fact that when there is more competition the prices are likely to drop. Especially in the audit market, where most of the assurance engagements are provided by the big 4, which this means that in the current market there is an oligopoly. A
Characteristic of an oligopoly is that there can be high competition on prices. (Liu & Simunic, 2005). The number of clients’ geographic segments might affect audit fee due to the fact that an environment might affect the financial reporting of a company. Key managers who influence financial reporting are located in the firm’s headquarters, but also other employees in other geographic locations also have an effect on firms financial reporting (Jha & Chen, 2014). Jha & Chen (2014) have found interactions between audit fee and social capital. Although this paper focusses on some elements which overlap the general notion of social capital, there still might be need to be in control over this general notion. Therefore, social capital could be added to the regression model using the Rupasingha and Goetz (2008) approach. Of which voter turnout and Census response data could be used as an operationalization of social capital. (Jha & Chen, 2014). However, Jha & Chen (2014) used this variable, by adding advanced mathematics and their data is not available, nor is the exact way they calculated their variable. Therefore, it is not possible to control for social capital in this thesis. It should however be noted the many aspects of social capital are also measured part of the variables of interest, which means that these also have connections with social capital. When the auditor also provides non-accounting services, the auditor will be signaled earlier in case of possible material misstatements. This will increase the internal control of the given company. An increase in internal control, leads to less audit fee as stated by Simunic (1980). Therefore an increase in NAS should lead to a decrease in audit fee. De Simone et al. (2015) support this by stating;
Prior research reports a positive association between tax NAS and both financial reporting quality and audit quality. We extend this literature by providing evidence that tax NAS improve internal control quality, which is an important component of financial reporting quality. We propose that tax NAS improve internal control quality by facilitating earlier audit firm awareness of material transactions. (De Simone et al., 2015, pp. 24)
According to Hayes, Wallage & Gortemaker (2014) going concern paragraphs indicate financial distress. The future of the client is uncertain and therefore the risk that the auditor cannot collect the audit fee too. The common reaction for the auditor is increase the fee due to the higher risk (loss exposure). Also Jha & Chen (2014), indicate that the going concern paragraph in general has an effect on audit fee and therefore needs to be controlled for. It has been proven that many auditors use low-balling when attracting new customers (Simon & Francis, 1988). Low-balling means “Setting audit fees below total current costs on initial audit engagements” (DeAngelo, 1981). Therefore, if there has been an audit switch it is likely that the audit fee has dropped. As
stated before risk is a determinant of audit fee. One of the most common forms of this is litigation risk (Simunic, 1980). Firms with their main operations in biotechnology, computers, electronics, and retail industries are more likely to overstate their net assets (Watts, 2003). It is also claimed by Reichelt & Wang (2010) that certain firms have high litigation risks. Sun & Liu (2011) state that there also is a theory called the “deep pockets” theory. Which means that when companies and/or auditors are in stricter financial regimes or when there is a high litigation risk, the auditor will provide better quality work, thus doing more work and therefore increasing audit fees.
2.2 Social capital
Social capital was first mentioned by Bourdieu, who claims that there are multiple kinds of capital. Bourdieu (Bourdieu (1987, pp. 3–4)) mentions that;
…the structure of a [social] space is given by the distribution of the various forms of capital, that is, by the distribution of the properties which are active within the universe under study – those properties capable of conferring strength, power and consequently profit on their holder… These fundamental social powers are, according to my empirical investigations, firstly economic capital, in its various kinds; secondly, cultural capital or better, informational capital, again in its different kinds; and thirdly two forms of capital that are very strongly correlated, social capital, which consists of resources based on the connections and group membership, and symbolic capital, which is the form the different types of capital take once they are perceived and recognized as legitimate. (Bourdieu, 1987, p. 3–4))
Bourdieu (1987) in his early research mentions economic-, cultural-, social- and symbolic capital. Bourdieu also mentions habitus, which is the personal dimension, which concerns personal behavior and experiences. In this thesis there will only be notion for social capital. This is for various reasons, among which the scope of this thesis, but also since this thesis mainly tries to contribute to Jha & Chen (2014), which is about social capital. Therefore, this does not mean that other forms of capital automatically do not influence audit fee. This can be researched in another thesis or research paper. As mentioned before social capital can be defined as “the norms and the networks that facilitate collective action. A high social-capital region has individuals with a greater propensity to honor an obligation and a greater mutual trust within a much denser network, all of which facilitate collective action.” (Jha & Chen, 2014). Guiso et al. (2004) define social capital as the mutual level of trust in society.
Social capital influences audit fee for a multiple reasons. First there are social norms which induces managers to behave more honestly (Jha & Chen, 2014). Yeung & Rauscher (2014) study youth employment and behavior. They find that African-American youths have more behavioral problems than Caucasian youths and are mediated by peer influence and socialization. It seems that social capital has an effect on the behavior of youths. Based on psychology literature it is known that companies are more likely to hire personnel that share the same set of values as the company. And also the fact that companies often hire personnel that reside close to the company. (Jha & Chen, 2014). However as mentioned above social capital has a mediating effect on possible behavioral problems. So it might also be that there is not an increased risk, due to better behavior within a company. It does however indicate that this group of people does have behavioral problems, thus meaning that region with many African-American youths, might lead to a riskier culture within a company. It also seems to be the case that religion has an effect on audit fee (Jha & Chen, 2014) and it might be possible that other non-financial factors have an influence on audit fee. It is most likely that Social capital is associated with audit fee, because of the perceived risk associated with different settings. (Jha & Chen, 2014).
2.3 Risk, trust and its effect on audit fee
In today’s “modernity” we are exposed to a lot of different risks. This risk can be from multiple sources like environmental, but also by different social hierarchies and class distinctions, where risks are divided unequally between those groups (Beck, 1992). Giddens (1990) claims that trust is also important since not all activities of individuals are visible and transparent and the working of systems is not always known or understood. Giddens states that trust is connected to contingency. Trust implies looking at the reliability of contingent outcomes and derives from faith in the reliability of a person or system (Giddens, 1990). So both respected authors state that trust and (perceived) risk can be dissimilar in different societies or systems. Trust is “a psychological state comprising the intention to accept vulnerability based upon the positive expectations of the intentions or behavior of another” (Rousseau, Sitkin, Burt, & Camerer, 1998, p. 395). Therefore, Giddens notion of trust can be associated with risk.
When looking towards accounting literature Bell et al. (2001) found that when there is a higher business risk, auditors tend to increase their audit prices. They do this however by collecting more audit evidence and not by increasing their hourly fee. This means that when in a riskier environment the auditor will be likely to collect more audit evidence, which in turn takes more time and which will result in a higher audit fee. As stated before (perceived) risk increases audit fee (Gonthier-Besacier & Schatt, 2007; Jha & Chen, 2014; Simunic, 1980). Another form of risk is
inherent risk. Inherent risk is a term within accounting literature which is part of the audit risk model (Houston, Peters & Pratt, 1999; Ruhnke & Schmidt, 2014). The audit risk model is a model which is used to determine the amount of audit evidence that is needed. The audit risk model is audit risk (AR) = the risk of material misstatements (RMM) * detection risk (DR). The RMM can be split into inherent risk * control risk. This mean that the formula can be rewritten as AR = IR * CR * DR. The general idea behind the ARM is that the AR is a fixed percentage. This means that if one of the determinants increases, the others have to decrease. In general, this means that the DR has to decrease (Houston et al., 1999; Ruhnke & Schmidt, 2014). This means that due to higher IR, more work has to be done to adjust the AR to the permitted (fixed) percentage.
As can be seen in the sections before the insider-outsider relationship consists of perceived risks. This relationship can also be examined in different communities. A community can be described as “virtual communities of people with like interests or ways of life” (Baxter, 2009). Meaning that there can be different perceived risks towards different societies. This in turn means that different societies might demand higher audit fees to other societies. In a more specific paper Schmid et al. (2014) have indicated that there is less trust when there is more ethnic diversity. Since less trust can be associated with more risk (Evans & Krueger, 2011), it is likely that ethnic diversity will influence audit fee. They describe in their paper that “A theory seeking to explain decisions in the trust game must recognize that the payoff structure comprises three distinct elements. The first two elements of trust are cost and benefit, which jointly correspond to the concept of personal vulnerability or risk” (Evans & Krueger, 2011)
Based on Giddens (1990), Beck (1992), Baxter (2009), Evans & Krueger (2011) and Schmid et al. (2014) it can be concluded that different social constructions can have a different perceived risk towards other social constructions. This is mainly because of information asymmetry and decreased mutual trust. Bell et al (2001) found that risk is a determinant of audit fee and therefore different social structures should also influence audit fee.
2.4 Minority groups and their effect on audit fee
As stated before risk is one of the most important determinants of audit fee. However multiple elements might increase this risk and therefore also increase the audit fee. African-American youths seem to have more problematic behavior than those of Caucasian heritage (Yeung & Rauscher, 2014). However, since companies tend to employ personnel with the same set of values and beliefs, this might not be enough for audit fees to rise. Other research by Tajfel and Turner (2001) states that;
There have been a number of studies (Tajfel etal., 1971; Billig & Tajfel, 1973; Tajfel & Billig, 1974; Doise, Csepeli, Dann, Gouge, Larsen, & Ostell, 1972; Turner, 1975), all showing that the mere perception of belonging to two distinct groups—that is, social categorization per se—is sufficient to trigger intergroup discrimination favoring the in-group. In other words, the mere awareness of the presence of an out-group is sufficient to provoke intergroup competitive or discriminatory responses on the part of the in-group. (Tajfel and Turner, 2001, p. 38)
Taifel and Turner (2001) continue by adding that; intergroup categorization leads to in-group favoritism and discrimination against the out-group. Tsutsui & Zizzo (2014) did an computerized experiment, where individuals were allowed to distribute (“give”) experimental points to other individuals. All individuals were part of a majority of minority group. They found that individuals belonging to the minority groups seem to be less likely to discriminate in giving experience points, which suggests that majority groups are more likely to discriminate in “giving”. In other words, majority groups are more likely to discriminate minority groups, than minority groups are likely to discriminate majority groups. (Tsutsui & Zizzo, 2014). This might mean that minority and majority groups, might pay different for audit fees, due to experiential preferences (a low-status groups does not like to be low-status and therefore are less likely to treat others as low-status) or intergroup discrimination. (Tsutsui & Zizzo, 2014) This notion of discrimination seems relevant since Bobo & Charles (2009) claim that due to the election of Obama in 2008 the issue of racism is in the middle of political debate. In their research they find that “White” people were nearly twice as likely as African Americans to state that racial relation will worsen due to Obamas election. Indicating that interracial relations might have worsened since 2008. Prior research also suggests that there is a “massive incarceration of young black men, a problem undermining the collective efficacy of minority communities.” (Unnever, Gabbidon & Higgins, 2011, p. 36). This research like Tajfel and Turner (2001), indicates that young black men have social/behavioral problems due to the fact that they state that there is a “mass incarceration” of black men. This together with the fact that racial relations might worsen might mean that a) non-caucasians have lower social capital and b) the racial problem within America persists in today’s society and is very relevant in the social context of the USA. Since there is discrimination towards minorities and social capital has an effect on audit fee, it is likely that minorities will pay significantly higher audit fee as well.
As described in chapter 1.2.2, research by Jha & Chen (2014) show that religion influences audit fee. They claim that this influences audit fee due to the perceived risk experienced by the auditor. They claim that people living in religious regions tend to act more ethical and use less
earnings management. This statement can be supported by the fact that perceived risk influences audit fee (Cobbin, 2002; Gonthier-Besacier & Schatt, 2007). This would mean that Religion > decreased risk > decreased audit fee. This shows that there has been research claiming some societies pay more audit fee than other. Which means that this also might be the case in other contexts, like ethnics.
2.5 Political preferences and social cleavage within the USA
The USA is a democracy, which means that a president (or other head of state) is elected. In the USA there are two main parties, which both have their own chosen representative who tries to become president. These parties as mentioned before are the democrats and the republicans. These parties have their own sets of views concerning different issues concerning domestic and international issues. In the final voting round the citizens vote for a new president. When looking at the past view years it can be noted that some specific states are loyal to their favorite parties. This brings up the question; Why do these states vote for a representative of that party. Monson and Mertens (2011) suggest that there are demographic and socials differences between the two groups, which explains why they vote republican or democrat. This suggests that there clearly are differences between the two groups, which might explain that if those groups live in those regions, those specific regions might vote for that party. Shin & Webber (2014) state that “The red/blue state label used widely in the media to characterize the 50 states is not without explanatory meaning. Of 79 indicators, there are 40 where red and blue states differ significantly.” (Shin & Webber, 2014). Shin & Webber (2014) also claim that blue states receive higher social capital scores, but that this is not significant. They further claim that “citizen ideology” is the largest factor within the political differences of the two groups. Furthermore, Brooks & Manza (1997), claim that there is a “social cleavage” within the USA. This means that there indeed might be some kind of a social distinction between the two groups of voters. They also found that this cleavage has not decreased. It is however important to note that the paper might be outdated in this case, but the concept of social cleavage exists nonetheless. Baker & Faulkner (2009), claim that there is a double embeddedness within the USA. Which can be described as a “class 4” double embeddedness. This means that the USA has both “Fragmented Networks” and “Divergent Values”, which is shown by the red & blue split of the country. Which is also supported by Brooks (2004). They do however claim that this is a contested view, since other researchers also claim that this distinction does not exist in the USA. Research done by Doyle (2006) however suggests that the division between blue and red states in just a myth and unfounded. This suggests that there should be no difference in audit fee between these regions.
As can be seen many authors claim that there is a social cleavage within the USA, which can be translated into red and blue states. Double embeddedness also has a role in this distinction. Other research however claim that this split does not really exist (Doyle, 2006). There is however limited research into the differences of the population itself.
2.6 Hypothesis building Ethnic regions
In today’s society we live in a risk society, which means that there is more focus on risks. These can be environmental, but also social aspects like class distinctions play a role in risk (Beck, 1992). As mentioned before there is a perceived risk towards people from other “societies”, due to information asymmetry, which causes mutual distrust. (Giddens, 1990). Therefore, Ethnic differences can be labeled as possible threats and can be perceived as more risky, than from within their own “class” ethnicities. Since risk and litigation risk are a determinants of audit fee (Bell et al, 2001; Gonthier-Besacier & Schatt, 2007; Hay et al., 2006; Simunic 1980), this would indicate that regions that are dominated by specific ethnicities would pay more audit fee than when it would be dominated by the “own” class. A second reason why other ethnic groups might pay more audit fee is that African-American youths can be linked with low social capital and behavioral problems. This would indicate that African-Americans, are riskier and therefore would pay more fee than Caucasians. Therefore, it can be expected that companies in regions with a higher percentage of other ethnicities and in the case of the USA that the African-American society will have to pay more audit fee that those in Caucasian society. When looking at research concerning minority groups it seems that majorities are likely to price discriminate minorities as stated by Tsutsui & Zizzo (2014). First of all, if majorities are likely to increase the price and minorities are less likely to discriminate it is likely to see at least minor differences since the minorities do get discriminated by some but are not likely to get benefits from even their own ranks. Second, the accounting profession seems to be dominated by Caucasians, as 70 – 90 % of the American accounting profession seems to be consisting of Caucasians (AICPA, 2013). This further supports that regions consisting of minorities are likely to pay more for audits. The first hypotheses can be stated as; H1: Companies residing in areas not dominated by Caucasians pay more audit fee than those residing in areas dominated by Caucasians.
H2: Companies residing in areas with a “high” percentage of African-American’s pay more audit fee than those residing in areas dominated by Caucasians.
As mentioned before this paper is a pioneer in this kind of research. The most related paper by Jha & Chen (2014) does not concern ethnicities and also examines county data. Therefore there is a lack of prior research to safely determine a “high” percentage. Therefore, “high” is defined as the double of the national standard (12,6%) (U.S. Census Bureau, 2016), being 25%.
Political regions
As seen in chapter 2.5 the USA can be divided into two groups, occupying multiple regions throughout the US. Prior research by Monson and Mertens (2011) suggests that there are demographic and socials differences between the two groups. Also Shin & Webber (2014) claim that there is clear evidence that there are differences between these two groups of voters. These demographic differences might affect audit fee. Also Brooks & Manza (1997) claim that there is a social cleavage within the USA. Baker & Faulkner (2009), claim that there is a “class 4” double embeddedness within the USA, which means that the USA has both “Fragmented Networks” and “Divergent Values”, which is shown by the red & blue split of the country. This is also supported by Brooks (2004). They do however claim that this is a contested view, since other researchers also claim that this distinction does not exist in the USA. Research done by Doyle (2006) suggests that the division between blue and red states is just a myth and unfounded. This suggests that there should be no difference in audit fee between these regions. Since there is little research that supports the claim that there are no demographic differences between the voters.
Many researchers claim that there is a cleavage within the USA, with limited opposition who claims that this political/social cleavage does not exist. This means that there probably is a difference between those two groups. Shin & Webber (2014) also claim that blue states receive higher social capital scores, but that this is not significant. Since there is limited research within this string of research it is likely that political states should affect audit fee. The direction of the hypothesis however, is hard to define. Since there seems to be a social difference and although it is not significant it seems that blue states have more social capital therefore the third hypothesis is;
H3: Blue states pay less audit fee than red and mixed states.
Testing this hypothesis will however not immediately mean that there are (or are no) social differences between the two different groups. It does however give more insight and credibility to the claims that there are. So even if there is an association, there will be need for more follow up research in order to provide more strength to this relation. In case the hypothesis finds no support, it might still be that there are differences, but that these do not reflect upon audit fee.
3 Research methodology
This sections contains the methodology which will be used to test the hypotheses as stated in chapter 2. First there is a section explaining the research method and the main sources for data. This section is followed by the research design. In the research design, there is a section explaining the general design, followed by a more in depth section of the variables of interest, where the main variables will be explained. This is followed by a control variable section. This section contains research about control variables, which ones have been used, why they are being used and how this influences the research design. This will then in the end be summarized by a regression model.
3.1 Research method
The mentioned hypotheses are being tested using multiple databases which are available at the WRDS website, which is accessible via University of Amsterdam library, further referenced as “database research”. The databases that have been used are Audit analytics, which contains information about audits, like which auditor has conducted an audit, which statements has been given and how high the audit fee and/or audit related fee is. The second database that has been used is the Compustat North-American (annual) database, which contains general information concerning financial statements, company information and other company related data. The database of Audit analytics is split into multiple sections. In order to get all relevant information that is needed for this thesis, two of these have been used, being the audit fees and audit opinions databases. This means that the data from this dataset used in this thesis comes from three datasets in total
The queries from these databases have been merged and processed using Stata. There however also will be other sources of data, being Cencus (American governmental agency) and Shin & Webber (2014), which both will be used to make variables, in combination with the earlier mentioned databases. The main source of data, outside of databases is the Census website, which is an American government agency tasked with a ten yearly observation of the American demographics. This data has also been used by Dutter (2013), Jha & Chen (2014) & Shin & Webber (2014). So the census data although not provided by scientific research is a reliable source of information, which is used in other scientific research and therefore can and will also be used in this research as a reliable source of data.
3.2 Research design
To test for the mentioned hypotheses, the research will be done on the same manner as by Jha & Chen (2014). This means that to prove the stated hypothesis data will be collected and matched using geography. An important difference is however that this thesis uses data concerning states, while Jha & Chen (2014) used data concerning US counties. To be more specific this means that we assume that companies get influenced by the place where they are positioned. Researching this can be done in two ways. Either the headquarter of a given company or second the regional headquarters of the audit firm (Jha & Chen, 2014). For data availability reasons and the database research method. The company headquarters will be used and not the regional headquarters. In order to control for changes in laws and regulation there will also be year indicators. In this specific case the sample will be taken from the years 2009-2014, which has been chosen due to several complexity issues and data limitations, concerning the Census data but also to control for different US presidents as has been explained in chapter 1. The hypotheses are tested using univariate T-tests, multivariate T-tests and the one-way ANOVA. The one-way ANOVA together with yearly T-tests and a robustness check for standard errors will be used as robustness checks. The multiple T-test in general should have the same outcome as the ANOVA. The yearly analysis will be used to determine if some years have differences, due to financial crisis, elections or any other event that might cause differences in audit fees, or the predictor variables. The tests will be run using a regression model. This model in general is stated as;
Audit fee = α + β „Variables of Interest‟ + β „Control Group‟ + Year indicators + ε
When taking the variables that have been used in this thesis into account, the above model can be translated into;
Ln Audit fee = α + β1 USA_African_American + β2 USA_Non_Caucasian + β3 Politics + β4 Auditee size + β5 Profitability + β6 Leverage + β7 Big4 + β8 Report lag + β09 Inherent risk + β10 Auditor change + β11 Audit opinion + β12 Litigation + β13 Going concern + Year indicators + ε
The “Variables of Interest” in above regression are “USA_African_American”, “USA_Non_Caucasian” and politics. Followed by the “Control Group” which are β4 up to and including β13.
Using the mentioned databases, a lot of data can be obtained. However not all the data from these databases is needed. This means that a selection of the data has to be made, in accordance with the stated hypotheses. In Appendix I a list of the variables can be seen. In the section below these variables will be explained.
3.2.1 Variables of interest
The dependent variable is audit fee, measured as the total fee paid from the company to the auditor in year x, which will be obtained using the audit analytics database. The audit fee is operationalized by using the natural logarithm of audit fee; (LN) AUDIT FEE. This has been done for two reasons. First because this is in accordance with prior literature (Hay, 2008; Hay et al., 2006; Jha & Chen, 2014) and second is to control for non-normal distributions. The independent variables which are needed to test the hypotheses are ethnicity and politics. In which ethnicity is divided in non-Caucasian and African-American regions. Politics has been operationalized as a dummy variable being 0 for “red” and “mixed” states and 1 for “blue” states. States have been linked to a specific group based on same way as Shin and Webber (2014) has measured the variable, who used the election results from 1964 till 2012 to make a proxy for “blueness”. Political regions in the USA can be identified by looking at the history of votes in those specific regions. It seems that in the last four elections, some states can be clearly labeled as “red” or “blue”. It seems to be that the west-coast is “blue” and the North-east of the USA is “blue”. The middle and south-east seem to be more “red” orientated. Remaining stated have no clear preference. “Red” states are stated to favor republican and “blue” states favor republicans. These statements are supported by maps provided by the American government, the maps of the elections between 2000 and 2012 can be seen in appendix II and a consolidation of these can be seen in appendix III. The government website contains maps, going back to 1964. In figure 5 the figure from Shin & Webber (2014) has been added. This will be used to code the observations. The states in the observation will be coded 0,1 according to the codes mentioned in this figure.
Ethnicity can be measured on multiple ways, ethnicity can be examined in board structures, but it is also a possibility is to check in what region a company is active. The first one would probably best be used when doing a study on board structures, but also reduces the “noise” caused by other factors. The second one can be used to see how different regions and context influence audit fee. This paper is about how non-financial attributes (and in some extent social capital) can influence audit fee. As can be read in chapter 2.2, social capital can be defined as “the norms and the networks that facilitate collective action. A high social-capital region has individuals with a greater propensity to honor an obligation and a greater mutual trust within a much denser network, all of which facilitate collective action.” (Jha & Chen, 2014, p. 612). This would indicate that a variable based on regions and broader network and societies would be more suited for studies on social capital. However, there are several flaws when examining regions. First of all, there are many unknown variables that might influence audit fee in a region. Second, in the case of ethnic differences there might be no association because the minorities resemble a small part of society. So the association on audit fee might be there, but it might be that this research will not find this association. In order to get the needed data, regions will be coded using the Census data2 and maps. In figure 6 and 7 the most recent Census maps (2010) can be seen.
Figure 6: Census map, percentage of African Americans, per county. Source: http://datamapper.geo.census.gov/map.html (visited 12-05-2016)
Figure 7: Census map, percentage of (“white”) Caucasians, per county. Source: http://datamapper.geo.census.gov/map.html (visited 12-05-2016)
As can be seen there are no states with a majority of African-Americans. There however also are other regions that are not clearly dominated by Caucasians, but by other groups together. The maps above together with tables from the census bureau, show us that there are state differences concerning the spread of certain ethnicities. Figure 8 and 9 lists the states that have been added as a dummy, together with their respective percentages, concerning the dummy that they are in. States have been listed non-Caucasian, if less than 50% of the population is reported as being Caucasian. One exception has been made for Maryland, which in total has a major Caucasian population, but the major cities, where the companies are located have a Caucasian minority. 12,6% of the American population is African American. Therefore, state with >25% African American population will be given a 1 in the dummy. Since 25% is double the amount of the USA average, stated with a percentage above 25% have been labeled as “high percentage” regions. The district of Columbia has been listed by the census data in their research. This region however is not one of the 50 states of the USA. However, it is an important region within Maryland. The district of Columbia is amongst others known for Washington D.C, which is the capital of the USA. This region has not been added as a dummy. But they do contribute to the research, concerning Maryland.
State African American District of Columbia (Maryland) 50,7% Mississippi 37% Georgia 30,5% Louisiana 30% Maryland 29,4% South Carolina 27,5% Alabama 26%
Figure 8: Regions for African-American dummy. Figure 9: Regions for non-Caucasian dummy.
3.2.2 Control variables
In order to make sure that the association is not caused by other interactions, control variables need to be added to the regression. As stated before the major determinants of audit fee are; company size, complexity, risk and resources used. However, these are not the only control variables that need to be added, as there is a wide horizon of variables that might cause an association. Hay et al. (2006) argue that not only size, complexity and risk are important, but that also; profitability, leverage, auditor type, auditor report lag, and the busy season for audits. Fung, Gul and Krishnan (2012) find that the city-level industry specialization of the auditors and their economies of scale have an impact on audit fees. Jha & Chen (2014) have shown that social capital also has an effects on audit fee. Since this paper is derived from the work of Jha & Chen (2014), it is very likely that the same control variables have to be used. All variables listen in the regression model are also mostly general variables that have been associated with audit fee. The regression model, which is used in this paper is;
Ln Audit fee = α + β1 USA_African_American + β2 USA_Non_Caucasian + β3 Politics + β4 Auditee size + β5 Profitability + β6 Leverage + β7 Big4 + β8 Report lag + β09 Inherent risk + β10 Auditor change + β11 Audit opinion + β12 Litigation + β13 Going concern + Year indicators + ε
The used (underlined) control variables will be explained in the following sections .
Auditee size
Auditee size can affect audit fee due to the fact that when a company is bigger, it is likely that more
State Non-white Hawaii 77,3% District of Columbia (Maryland) 65,2% California 59,9% New Mexico 59,5% Texas 54,7%
and therefore the audit fee increases. Hay et al. (2006), claim that most studies use the natural logarithm of total assets. Furthermore, they claim that auditee size explains a fast majority of audit fee increases.
Profitability
Auditees that are making a loss, expose the auditor to increased risk, since there is a chance that the auditee cannot pay the auditor for the services it provides (Simunic, 1980). This is also supported by more recent research by Jha & Chen (2014), Hay et al. (2006) and Hay (2013). Jha & Chen use the variables Loss as is also supported by Hay et al. (2006) in their meta-analysis.
Leverage
Since bad leverage generally increases the loss exposure of the auditor and thereby increasing audit fees, this variable also needs to be added to the regression model. Common measures of leverage are debt divided by total assets and the current/quick-ratio. Hay et al. (2006) find that debt divided by total assets is the most significant variable.
Big 4 companies
Hay et al. (2006) found that Big4 auditors are significantly positively related to audit fee. Big4 companies in general have higher prices than other accountants due to their brand recognition. Since Big4 companies generally demand higher audit fees, the dummy “big 4” has been added to the regression.
Report lag
Report lag is a proxy for potential problems during the audit. A longer period between balance date and the auditor’s opinion might indicate that problems have arisen during the audit (Hay et al., 2006). Hay et al. (2006) also found that this variable is significantly (positively) related to audit fee. Therefore, also report lag has been added to the regression.
Inherent risk
As stated before perceived risk and inherent risk increase audit fee (Bell et al, 2001; Gonthier-Besacier & Schatt, 2007; Hay et al., 2006; Houston et al. 1999; Simunic 1980). Inherent risk increases audit pricing, because it is part of the audit risk model. The audit risk model is a model which is used to determine the amount of audit evidence that is needed to successfully complete the audit. Since the general notion is that if inherent risk increases, you need to collect more audit
evidence and more works needs to be done, it means that if IR increases, DR has to decrease. If DR decreases, audit evidence increases and with that the audit fee also increases (Houston et al., 1999; Ruhnke & Schmidt, 2014).
Change of auditor
Due to the competitive market, many companies use “low-balling” when attracting new customers. Low-balling means “Setting audit fees below total current costs on initial audit engagements” (DeAngelo, 1981). Therefore, if there has been an audit switch it is likely that the audit fee has dropped. To measure this a dummy variable has been created, labeled 1 if there has been an auditor switch and 0 if the auditor has stayed the same.
Audit opinion
Audit problems might increase audit fees due to an increased risk. This means that if there are more audit problems, there is a higher chance for misstatements and therefore it is likely to increase audit fee. Therefore, a dummy has been made 1 if the auditor issues an unqualified opinion without any additional language and 0 if otherwise. Which is consistent with research done within prior research (Hay et al., 2006; Jha & Chen, 2014).
Litigation risk
One of the most common forms of risk is litigation risk (Simunic, 1980). Firms with their main operations in biotechnology, computers, electronics, and retail industries are more likely to overstate their net assets (Watts, 2003). It is also claimed by Reichelt & Wang (2010) that certain firms have high litigation risks. These firms in general have the following sic codes 2833-2836, 3570 – 3577, 3600-3674, 5200 – 5961, and 7370-7370. These codes are also used by Jha & Chen (2014). And have been made into a dummy variable being 1 if the company is in one of those sections.
Going concern
According to Hayes et al. (2014) going concern paragraphs indicate financial distress. The future of the client is uncertain and therefore the risk that the auditor cannot collect the audit fee too. The common reaction for the auditor is to increase the fee due to the higher risk (loss exposure). Also Jha & Chen (2014), indicate that the going concern paragraph in general has an effect on audit fee and therefore needs to be controlled for.
Industry
Different industries have different complexity, which requires more work to be done. This is recognized by a wide range of audit fee researchers among which Hay et al. (2006). Those researchers use utilities or financial institutions as a proxy for complexity. In this case companies can be labeled as a financial institution by using a dummy 0,1 variable which state that 0 is not a financial institution and 1 that the company is a financial institution. However, since it is not unlikely to remove the sector altogether, this industry has been excluded from the dataset, which means that during the data collection phase the financial sector has been excluded from the proxies altogether.
Figure 10 shows a summary of the variables and their measurement. It also shows the expected relations of all the variables added in the regression. A more detailed list of all variables that are part of the regression can be seen in Appendix I.
Variable Measurement Expected
relation
Ln audit fee Natural logarithm of audit fee
Non-Caucasian state (dummy) Dummy labeled 1 if a Caucasian minority +
African-American state (dummy) Dummy labeled 1 if African American
majority +
Blue state (dummy) Dummy labeled 1 if the state has is blue in
accordance with Shin & Webber (2014) -
Auditee size Natural logarithm of total assets +
Auditee complexity Exclusion of FS
Profitability Dummy Loss +
Leverage Debt(-equity) ratio +
Big 4 company Dummy big 4 +
Report lag Days between financial statement and
financial year end
+
Litigation risk Industry dummy +
Litigation Dummy variable based on SIC +
Going concern Going concern +
Other Inherent risk
Auditor change (dummy) Audit opinion (dummy)
+ -/- +
