Abstract
The recent several audit failures has raised the question of whether external auditors have effective role in governance of firms. This study examines how the value relevance of annual accounting earnings, varies according to the quality of the external auditor's (that is competence and independence) of publicly listed companies within the Tunisian capital market. Data analysis over a period of six years (2000-2005) revealed that the value relevance of accounting information is positively related to the reputation, the specialization and the experience of the external auditor's. However no relation was able to be identified between the auditor's tenure and the relevance of the accounting. We also find that the value relevance of earning accounting is increasing in tenure for clients of Big and specialist auditor's.
Keywords: External auditors, competence, independence, value relevance, Tunisian listed company
1. Introduction
Audit quality plays an important role in the corporate governance of organizations. External auditors serve a valuable function in capital markets by lending credibility to financial statements issued by public companies, thereby reducing information risk (Lu and Sivaramakrishnan, 2009). More specifically, quality of the external auditor guarantee that the actions of the senior managers are aligned with the interests of shareholders, creditors, investors and any other party that is related to the company.
The relation between external audit and the quality and the informativeness of annual earnings has been an issue of international research in recent years [1] . Many researchers found a positive relation between Big auditor's, auditor's experience and earnings quality (Teoh and Wong, 1993; Johnstone and al., 2002; Gul and al., 2006; Habib and Azim, 2008). Moreover, existing literature on the link between auditor's specialization and accounting quality has argued that the incident of financial statement fraud is weaker for firms audited by a specialist (Carcello and Nagy, 2002; Balsam and al., 2003; Krishnan, 2003; Velury, 2003; Jenkins and al., 2006; Gul and al., 2009).
Additionally, another stream of research examines the link between audit tenure and the quality of published financial statements. However this research reported mixed results. As pointed out by Brody and Moscove (1998), a positive relation between auditor rotation and audit quality. They find that auditor rotation enhances greater independence and reduce clients' inadequate influence on auditors. Myers and al. (2003) provides evidence to support that auditor tenure is positively related to earnings quality (proxies by absolute value of discretionary accruals).
We extend the line of research pertaining to earnings quality by examining whether earnings quality is related to auditor quality. More explicitly, we examine the effect of auditor competence (Big auditors, specialization and experience) and independence (audit tenure) on the stock return-earnings relation using a sample of 20 financial and 20 no financial listed Tunisian firms.
Our study contributes to the auditing and accounting literatures as follows. First, we provide evidence on the differences in earnings quality (proxies by return-earnings relation) between financial and no financial listed Tunisian firms. Second, our study examines the effect of auditor's quality on value relevance of the earnings and whether auditor's Big and industry specialization of auditors affect the association between auditor tenure and value relevance of accounting earnings.
The paper proceeds as follows. Section 2 provides literature review and hypothesis. Section 3 provides detail on the research design. Section 4 presents the empirical results and Section 5 concludes.
2. Literature review and hypothesis development
It is difficult to assess audit quality ex ante because the only observable outcome of the audit is the audit report which is a generic template and the overwhelming majority of reports are standard clean opinions (Francis, 2004). DeAngelo (1981) defines audit quality as the market assessed joint probability that an auditor will both discover and report an error in a client's accounting system. The likelihood that the auditor will discover the error is a function of auditor competence and the likelihood that the auditor will report the error is a function of auditor independence. It is generally believed that a high audit quality translates to high earnings quality (Jenkins and Velury, 2008). Because it is not possible to evaluate and measure the independence and the competence, researchers have typically examined the effect of auditor Big, industry specialization, auditor experience and tenure on earnings quality.
Auditor's competence and value relevance of annual accounting earnings
Prior empirical research of audit quality suggest that large audit firms product higher-quality services, and thus provide better reputations than smaller audit firms (Balvers, McDonald, and Miller, 1988; Beatty, 1989; Clarkson and Simunic, 1994; Datar, Feltham, and Hughes, 1991; Teoh andWong, 1993). Francis and Simon (1987) and Palmrose (1988, 1989) find that auditors Big Six demand higher audit fees, spend more time on audits mission, and have lower lawsuits than non-Big Six auditors, and thus enjoy higher quality audits than non-Big Six auditors. Furthermore, Becker and al. (1998) and Francis and al. (1999) document that Big Six auditor's restrict clients' earnings management activities and suggest that clients of Big Six auditors report weak discretionary accruals compared to clients of non-Big Six auditors. Teoh and Wong (1993) reported a positive association between audit firm size and the earnings ratio coefficient (ERC). As noted by Gul and al. (2006) Big Six auditors have better methods for detecting agency problem areas and to provide higher quality audits than non-Big Six auditors. The author's document a significant inverse relationship between non-audit services (NAS) and the value relevance of earnings, and this inverse relationship is lower for Big Six auditors.
Habib and Azim (2008) examine the relationship between corporate governance and the value-relevance of accounting information in Australia. They results show that firms audited by a Big auditor's exhibit higher value-relevance of accounting information (measured by the adjusted R2 derived from a regression of stock price on earnings and equity book values following Ohlosn's accounting-based valuation framework). Based on this argument, we propose the following hypotheses:
H.1: Ceteris paribus, there is a positive relation between value relevance of accounting earnings and Big auditor's.
Other than the Big auditor's, the competence of auditor's depends in part on whether the auditor is an industry specialist or not. Prior literature suggests that client-firms of industry specialists provide higher quality of audits and report superior quality of earnings (Gramling and al., 1999; Balsam and al., 2003; Carcello and Nagy, 2002). Industry specialist auditors are generally considered to have both the competence and incentive to perform superior audits relative to non-specialist auditors (Jenkin and al., 2006).Such knowledge and experience presumably allow industry specialist auditors identify industry-specific issues and problems. As noted by Beasley and Petroni (2001) high industry knowledge and better audit technologies help industry specialist's product better audits. This result was verified by Hammersley (2006), Chen et al. (2006), Rusmin and al. (2006) and Lowensohn and al. (2007)
Researchers in accounting and auditing literature have suggested a positive impact of industry specific knowledge on the quality of earnings. Using different measures of earnings quality, studies have reported that client firms of industry specialists perform higher earnings response coefficients (Balsam and al., 2003), provide higher earnings persistence (Gramling and al., 1999), lower discretionary accruals (Balsam and al., 2003; Krishnan, 2003) and lower fraudulent financial statements (Carcello and Nagy, 2002) compared to client firms of non-specialist auditors.
Balsam and al. (2003) found evidence that clients of industry specialist auditors have higher association between returns-earnings than clients of non-specialist auditors. One recent study by Jenkins and al. (2006) argues that industry specialist auditors perform earnings quality, as measured by discretionary accruals and the returns-earnings relation. The authors make their study during the late 1990s, a period marked by extraordinary stock market activity. Jenkins and al. (2006) compared two periods, the 1997-1999 periods relative to those of the 1990-1996 period, and find a significant increase in the magnitude of discretionary accruals and a significant decrease in the relation between returns and earnings (ERC). Both the increase in discretionary accruals and the decrease in ERC are shown to be significantly smaller for clients of industry specialist auditors.
Gul and al. (2009) examine whether industry specialization of auditors affect the association between auditor tenure and earnings quality and find that the association between shorter auditor tenure and lower earnings quality is weaker for firms audited by industry specialists compared to non-specialists. Considering the above arguments and findings we predict a positive relation between industry specialist auditor and earnings quality, leading us to our second assumption.
H.2: Ceteris paribus, there is a positive relation between value relevance of accounting earnings and industry specialist auditor's.
Other than the Big and industry specialist auditor's, the competence of auditor's depends in part on the experience auditor. Prior research has documented that experienced auditors are better at detecting audit problems (Bedard and Biggs, 1991), can perform audit mission by explaining audit procedure (Bedard and Biggs, 1991) and use a more cost effective approach to resolve audit problems (Biggs, Mock and Watkins, 1988). Abdolmohammadi and Wright (1987) and Bonner (1990) report that auditors with greater experience were more likely to detect the error and the fraud present in a complex analytical procedure task. Similarly, Choo and Trotman (1991) provide evidence that auditors with greater task experience exhibit an enhanced knowledge structure and ability to produce a better quality of audit. Bedard, Biggs, Bedard and Wright (1994) document that experience of auditor enhance the planning of appropriate audit tests. This result converges with the conclusion advanced by Libby and Frederick (1990). Indeed, the authors show that the experience improves performance of the auditors. In this respect, the experience to repeat on tasks very specific increase the level of knowledge of the auditor's.
Libby and Tan (1994), Libby (1995), Tan and Kao (1999), and Thibodeau (2003) finds that experience auditor is associated with superior auditor performance gained by repeated experience and feedback in a particular domain.
In a more recent study, Johnstone and al. (2002) show that the most experimented auditors have more capacity to resist to their client during a conflict on the establishment and the presentation of fraudulent financial statements. We therefore propose our third hypothesis:
H.3: Ceteris paribus, there is a positive relation between value relevance of accounting earnings and experience auditor's.
Auditor's independence and value relevance of annual accounting earnings
Regulators and lawmakers have historically argued that auditor independence is affected by audit tenure (Code of the commercial companies of Tunisia, SEC,…) and that lengthy tenure reduces an auditor's independence. In auditing literature, research on the relation between audit tenure and audit quality report a mixed results. Shockley (1981), experimentally tested the effect of tenure on auditor independence, find that tenure more than five years were greater at risk of decreasing independence than auditors with tenure of less than five years. Also, he document no effect of the length of auditor-client relationship on the perception of auditors' reporting of errors, however, he argue that have a difference in the two group (longer auditor-client relationship and shorter auditor-client relationship) in auditors' ability to discover errors. Consistent with this view, Brody and Moscove (1998) evaluated auditor rotation affects auditor independence. They suggest that rotation enhances independence and improves the overall audit quality through a decreasing the ability of clients' inadequate influence on auditors. Imhoff, (2003) and Dopuch and al. (2001) assert that auditor rotation provide higher quality of audits and report superior quality of earnings enhance greater quality of the financial reporting process.
In recent research, Myers and al. (2003) find that the length of auditor-client relationship is positively related to earnings quality (proxied by discretionary and current accruals).Similarly, Chung and Kallapur (2003) found an inverse association between the length of the auditor-client relationship and the abnormal accruals. Carcello and Nagy (2004) found that the likelihood of fraudulent financial reporting is greater in the in the early years of auditor-client relationships. Alternatively, they do not support the assumption that the length of auditor-client relationship increased likelihood of fraud. Consistent with the theory of Carcelle and Nagy (2004), empirical work presents evidence that audit tenure enhances greater independence and improves audit quality. Stanley and DeZoort (2007) document that the quality of financial reports does not deteriorate when the length of auditor-client relationship. They report a negative relation between audit tenure and the likelihood of financial restatement.
Other recent studies extend the literature by considering that audit tenure enhances greater earnings quality. Ghosh and Moon (2005) studied the relationship of the length of auditor-client relationship and investor perceptions of earnings quality. They document a positive association between earnings response coefficient and auditor tenure. Carey and Simnett (2006) find an inverse association between the length of auditor-client relationship and the propensity to issue a going-concern opinion. Alternatively, they document that longer audit partner tenure is associated with a higher propensity for clients to just beat earnings benchmarks. He results consistent with long audit partner tenure deteriorate audit quality.
More recently, Davis and al. (2009) find that the length of auditor-client relationship (13-15 years or more) are greater likely to notice levels of discretionary accruals that allow them to meet or beat earnings forecasts, they also find the same result for companies with short audit tenure (two to three years).
We extend the literature involving the audit tenure to examine whether previous findings are robust in the Tunisia context of financial and no financial firms. Specifically, we examine the impact of the length of the auditor-client relationship on the returns-earnings relation. We base on the study of Ghosh and Moon (2005) and we predict a positive relation between audit tenure and value relevance of accounting. Stated formally:
H.4: Ceteris paribus, there is a positive relation between value relevance of accounting earnings and the length of the auditor-client relationship.
3. Research Design
3.1. Sample selection
Our sample comprises data from 20 non-financial and 20 financial firms listed on the Tunisian Stock Exchange for the years 2000 through 2005. The firms placed on the stock market during their first year are not included in the sample. The Return and the stock price data were extracted from the Tunisian Stock Exchange, while data on the earnings, auditor Big, tenure and debt were hand collected from each firm's annual financial reports. The data on the Industry specialization of the auditors are collected by means of a questionnaire for research send with the auditors. The data on the auditor experience are collected by means of a questionnaire for the auditors Big and also by the sites of the auditors non Big.
3.2. Model
To test our Hypothesis, we estimate the following panel data regression model that relates return-earning relation level-based measure of value relevance of accounting, the attributes of the quality of the external auditor's and other control variables. The control variables include firm size and debt.
We measure value relevance of accounting (VRit) by return-earning relation. The return-earning relation is measured in the period following the announcement of annual reports. The value relevance of accounting is measured in the period following the announcement of annual reports. We prefer a window which includes the period of the date financial statement publication and the fiscal years end because we verify how the market assesses the quality of the audited financial statements.
The earnings value relevance is equal in adjusted R-square with the following regression:
Rit: are market-adjusted returns computed from the period of the date financial statement publication and the fiscal years end
EA and ∆EA represent levels of and changes in annual earnings, both deflated by the market value of equity at the end of fiscal year t -1.
BIG is a dummy variable coded 1 if the firm is audited by a "Big 4" and 0 if not. SPEC is a dummy variable coded 1 if the firm is audited by a specialist and 0 if not. We define industry specialization (SPEC) if the auditor detains at least a 10 percent market share for any industry. Auditor experience (EXPER) is measured by the number of consecutive years of auditor experience. Audit Tenure (TENURE) is the number of consecutive years that the auditor proceeded to audit for the companies. Size is the natural log of the market capitalization of a firm and debt is measured by the ratio of the total debts with regard to the total assets.
4. Empirical Results
4.1. Descriptive statistics
Table 1 contains the descriptive statistics of the non-financial variables and Table 2 contains the descriptive statistics of the financial variables from 2000 to 2005. Approximately, 56 percent of the sub-sample of non financial companies is audited by an auditor Big 4 and 38 percent is audited by an industry specialist. Table 2 reports that 45 percent of the sub-sample of financial companies is audited by an auditor Big 4 and 42 percent is audited by an industry specialist.
Univariate tests show that Big auditor, auditors' industry specialization and auditors' experience are positively associated with value relevance of accounting earnings for the two sub-sample. Also, result show that audit tenure have significant positively effect on value relevance in the sub-sample of non-financial companies and no significant effect on the sub-sample of financial companies. However, size and debt have no significantly impact on value relevance of accounting.
Table 1: Descriptive statistics and univariate tests on the sub-sample of Tunisian non-financial listed companies
valeur
Tous les groupes
Intra-groupe
Inter-groupe
VR
Freq
percent
Freq
Percent
Percent
Variable
SPEC
0
1
67
42
61.47
38.53
15
9
75.00
45.00
84.81
80.77
3.04***
(0.000)
BIG
0
1
47
62
43.12
56.88
13
14
65.00
70.00
68.12
78.48
4.09***
(0.000)
Mean
St. Deviation
Minimum
Maximum
VR
EXPER
Intra-groupe
Inter-groupe
17.87156
3.149712
1.951179
9
26
5.32***
(0.000)
TENURE
Intra-groupe
Inter-groupe
4.495413
1.132198
2.150222
1
9
2.04**
(0.041)
SIZE
Intra-groupe
Inter-groupe
17.89204
.9410656
.4527693
16.48924
21.90919
0.46
(0.646)
DEBT
Intra-groupe
Inter-groupe
7.279099
28.21619
65.75156
0
1.090521
-0.16
(0.871)
The statistics reproduced in the table are the average (Mean), the standard deviation (St Deviation), the minimum (Minimum) and the maximum (Maximum), frequency (Freq) and percentage (Percent).
BIG = 1 if the firm is audited by a "big," 0 otherwise; SPEC= 1 if the firm is audited by a specialist, 0 otherwise. EXPER = the number of consecutive years of auditor experience. TENURE= the numbers of consecutive years that the auditor proceeded to the audit of the firm; SIZE = Ln market capitalization of a firm; DEBT = ratio of the total debts with regard to the total assets.
VR= adjusted R-square for the following model :
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively.
Table 2: Descriptive statistics and univariate tests on the sub-sample of Tunisian financial listed companies
valeur
Tous les groupes
Intra-groupe
Inter-groupe
VR
Freq
Percent
Freq
Percent
Percent
Variable
SPEC
0
1
69
51
57.50
42.50
15
11
75.00
55.00
76.67
77.27
5.40***
(0.000)
BIG
0
1
66
54
55.00
45.00
15
13
75.00
65.00
73.33
69.23
5.37***
(0.000)
Mean
St. Deviation
Minimum
Maximum
VR
EXPER
Intra-groupe
Inter-groupe
17.14286
3.149712
1.951179
8
25
3.90***
(0.000)
TENURE
Intra-groupe
Inter-groupe
3.733333
1.132198
2.150222
1
9
0.74
(0.460)
SIZE
Intra-groupe
Inter-groupe
17.68415
.9410656
.4527693
14.187
20.009
0.94
(0.349)
DEBT
Intra-groupe
Inter-groupe
.28619
28.21619
65.75156
.0021
.93
1.51
(0.132)
The statistics reproduced in the table are the average (Mean), the standard deviation (St Deviation), the minimum (Minimum) and the maximum (Maximum), frequency (Freq) and percentage (Percent).
BIG = 1 if the firm is audited by a "big," 0 otherwise; SPEC= 1 if the firm is audited by a specialist, 0 otherwise. EXPER = the number of consecutive years of auditor experience. TENURE= the numbers of consecutive years that the auditor proceeded to the audit of the firm; SIZE = Ln market capitalization of a firm; DEBT = ratio of the total debts with regard to the total assets.
VR= adjusted R-square for the following model :
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively.
4.2. Tests on Panel Data
In Table 3, we also report the test of panel data. In the sub-sample of non financial companies, the results of the test of presence of individual effect allow to throw reject H0 and to accept the alternative hypothesis that is the hypothesis of presence of individual effects. However, in the sub-sample of financial companies, the results of this test allow to accept H0 and to verify the absence of the individual effects.
Consequently, the model to use for the sub- sample of non financial companies is the model of panel data; and in that case it is necessary to pass in the other tests of the panel data. However, for sub-sample of financial companies, we apply the Heteroscedasticity test. If this test verifies the hypothesis of presence of the problem of Heteroscedasticity, thus we use the model of panel data which allows correcting this problem. But if the Heteroscedasticity test shows the absence of a serious problem of Heteroscedasticity, thus we use a linear regression.
The individual effects can be either fixed, or random. Generally it is the Hausman test that allows determining if there is a statistical difference between the coefficients of both estimations (fixed and random). The Hausman test shows a value of 5,29 and one p-value worth upper to 10 % (50,67 %). The results of this test show while the individual effects are random and not fixed.
The result of the Heteroscedasticity test (Beusch Pagan test) applied to the sub-sample of non-financial companies verifies the absence of a problem of Heteroscedasticity. However, in the sub-sample of financial companies, the result of this test shows the presence of a problem of Heteroscedasticity.
Table 3: Test on panel data
Modèle du sous-échantillon d'entreprises tunisiennes non financières cotées
Modèle du sous-échantillon d'entreprises tunisiennes financières cotées
Test de présence d'effet individuel
47.83***
(0,000)
0.81
(0.3667)
Test d'Hausman
5.29
(0.5067
Test d'hétéroscédasticité (test de Breusch-Pagan
1.62
(0.2032)
8.49
(0.0036)
The last test to be verified within the framework of the panel data is the one of the absence of a problem of co-linearity between the independent variables introduced into the same model. Spearman correlation coefficients are reported in table 4 and table 6. The result shows a serious problem of co-linearity between the auditor BIG (variable BIG) and the auditor specialization (variable SPEC). Also, according to the table 5 and 7 we let us notice that the VIFs (Variance Inflation factor) relative to the variable BIG and the variable SPEC exceed the value 2 while the VIF of the other variables post values lower than 2. The VIFs indicate a co-linearity problem exist with BIG and SPEC variable. To resolve this problem, these variables are introduced one by one into the model.
Table 4: Spearman matrix for the sub-sample of Tunisian non-financial listed companies
BIG
SPEC
EXPER
ANAUD
TAILL
SPEC
0.6894***
0.0000
EXPER
0.2467***
0.0097
0.1807*
0.0600
ANAUD
-0.1136
0.2396
0.1201
0.2135
0.2578***
0.0068
TAILL
-0.1013
0.2947
-0.0971
0.3154
0.1562*
0.1048
0.1296
0.1791
END
-0.0371
0.7018
0.0665
0.4920
-0.0813
0.4009
0.1609*
0.0947
0.4147***
0.0000
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively
Table 5: The VIFs of independent variables for the sub-sample of Tunisian non-financial listed companies
BIG
SPEC
EXPER
ANAUD
TAILL
END
VIFs
2.27
2.11
1.27
1.26
1.35
1.29
Table 6: Spearman matrix for the sub-sample of Tunisian financial listed companies
BIG
SPEC
EXPER
ANAUD
TAILL
SPEC
0.9501***
0.0000
EXPER
0.5447***
0.0000
0.5205***
0.0000
ANAUD
-0.0020
0.9828
0.0080
0.9316
-0.0241
0.7949
TAILL
0.0519
0.5750
0.0251
0.7867
-0.2534***
0.0054
-0.0092
0.9210
END
0.0845
0.3606
0.1165
0.2071
-0.0192
0.8359
0.0710
0.4427
0.4147***
0.0000
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively
Table 7: The VIFs of independent variables for the sub-sample of Tunisian financial listed companies
BIG
SPEC
EXPER
ANAUD
TAILL
END
VIFs
11.56
10.52
1.71
1.06
1.48
1.40
4.3. Empirical Results
This section contains the results of regressing return-earning relation on the auditor's quality variable. Our aim was to examine the effects of auditor's competence (BIG, SPEC and EXPER) and auditor's independence (TENURE) on value relevance of accounting and test whether hypothesis 1, 2, 3 and 4 are valid. The coefficient on BIG is positive and statistically significant (two-tailed p-value < 0.01) on the two sub-sample (financial and non-financial companies). Consistent with H1, value relevance of accounting is, on average, higher for firms audited by auditors' Big. Consistent with prior research our result indicates that large auditors provide audits of higher quality relative to other auditors. The Big auditor's firms have more resources, superior knowledge than the non-Big auditor's firms and product better quality audit. So, as advanced by Teoh and Wong (1993) the clients of Big audit firms having higher return-earnings relation.
Table 8 and 9 reveals that the auditor's specialization affects the earnings quality of return-earnings relation since it is found to be positive and statistically significant in all model specifications on the sub-sample of financial and non-financial companies. Thus hypothesis 2 can be accepted signifying that industry specialization auditor is an important issue that firms must consider since it is strongly associated with return-earnings relation. This found result converges on the study Balsam and al. (2003) and Jenkins and al. (2006) which suggest that industry expertise is related with an auditor's ability to identify problems within financial statements and positively correlated with earnings quality, as measured by the return-earnings relation.
The results in Table 8 and 9 support our H3 prediction of a positive relation between auditor's experience and value relevance of accounting. Indeed, the coefficient of auditor's experience influences positively and significantly the return-earnings relation of p-value 1 % on the sub-sample of non-financial companies and of p-value 10 % on the sub-sample of financial companies. This finding provides support that more the auditor is experimented, more it is able to estimating the complexity of a situation and more it is able to detecting the errors and the fraud contained in financial status.
However, results do not support hypothesis 4 since the audit tenure variable found to be insignificant in all model specifications on the sub-sample of financial and non-financial companies. Consequently, audit tenure is irrelative to the value relevance of annual accounting earnings, a result which is inconsistent to Ghosh and Moon (2005).
Concerning the other control variables, such as the size and debt, have insignificant effect on value relevance of accounting in the sub-sample of non-financial companies. However, according to the table 9, the results show that the size affects positively the return-earnings relation.
Tableau 8: Linear regression on the sub-sample of Tunisian non-financial listed companies
Model 1.1
Model 1.2
Variable
SP
ßi
ßi
Intercept
-.708338
(-1.29)
-.734094
-1.31
BIG
+
.1528784***
(2.60)
-
SPEC
+
-
.1374528**
(2.03)
EXPER
+
.0316815***
(3.32)
.0387479***
(4.35)
TENURE
+
.0009876
(0.11)
-.0058666
(-0.67)
SIZE
+
.0265116
(0.88)
.0239798
(0.79)
DEBT
-
-.1300997
(-1.34)
-.1085754
(-1.12)
Number of observation =109, Number of year = 6
Model 1.1: Wald chi2= 37.09*** ; within = 0.2720; between = 0.2273; overall = 0.2747
Model 1.2: Wald chi2= 33.80*** ; within = 0.2678; between = 0.1755; overall = 0.2147
Model 1.1: We eliminated the variable « SPEC » ;
Model 1.2: We eliminated the variable « BIG ».
Are reproduced in this table the coefficients of the linear estimate (ßit) and the coefficients of regression (statistical of Wald chi2 `Z') relative to each variable included in the corresponding model. *, ** and *** indicate a bilateral critical probability to the threshold of 10%, 5% and 1% respectively. The statistics of khi-deux Wald and test and the number of observation are also presented. PS corresponds to the predicted signs.
BIG = 1 if the firm is audited by a "big," 0 otherwise; SPEC= 1 if the firm is audited by a specialist, 0 otherwise. EXPER = the number of consecutive years of auditor experience. TENURE= the numbers of consecutive years that the auditor proceeded to the audit of the firm; SIZE = Ln market capitalization of a firm; DEBT = ratio of the total debts with regard to the total assets.
VR= adjusted R-square for the following model :
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively.
Tableau 10: Linear regression on the sub-sample of Tunisian financial listed companies
Model 1.1
Model 1.2
Variable
SP
ßi
ßi
Intercept
-.2293287
(-0.85)
-.344911
-1.36
BIG
+
.1755349***
(3.74)
-
SPEC
+
-
.1640706***
(3.86)
EXPER
+
.0068176
(1.34)
.0081191*
(1.68)
TENURE
+
.0060767
(0.60)
.005456
(0.586)
SIZE
+
.0134289
(1.13)
.0186743*
(1.67)
DEBT
-
.0420059
(0.74)
.0541883
(0.98)
Number of observation =119, Number of year = 6
Model 1.1: Wald chi2= 31.62***
Model 1.2: Wald chi2= 31.55***
Model 1.1: We eliminated the variable « SPEC » ;
Model 1.2: We eliminated the variable « BIG ».
Are reproduced in this table the coefficients of the linear estimate (ßit) and the coefficients of regression (statistical of Wald chi2 `Z') relative to each variable included in the corresponding model. *, ** and *** indicate a bilateral critical probability to the threshold of 10%, 5% and 1% respectively. The statistics of khi-deux Wald and test and the number of observation are also presented. PS corresponds to the predicted signs.
BIG = 1 if the firm is audited by a "big," 0 otherwise; SPEC= 1 if the firm is audited by a specialist, 0 otherwise. EXPER = the number of consecutive years of auditor experience. TENURE= the numbers of consecutive years that the auditor proceeded to the audit of the firm; SIZE = Ln market capitalization of a firm; DEBT = ratio of the total debts with regard to the total assets.
VR= adjusted R-square for the following model :
Note: Significant at: *1, * * 5 and * * *10 percent levels, respectively.
5. Additional analyses
Prior literature suggest that the effect of auditor tenure on audit quality is controversial (Carcello and Nagy, 2004; Myers et al., 2003). Several prior researches argue that the length of auditor-client relationship reduces auditor independence and thus reduces audit quality. Recent studies provide evidence that audit tenure enhance higher earnings quality (Geiger and Raghunandan, 2002; Gul et al., 2007) and higher audit quality.
In the previous section, the results show that auditor Big and industry specialization of audit firms improve the value relevance of the accounting. However, audit tenure has no significant impact on the return-earnings relation. Indeed, and to deepen more our analysis, we verify in what follows, if the tenure of Big and specialists improves the value relevance of the accounting. Because the auditors Big and the auditor's specialists improve the value relevance, thus we wait in the fact that the tenure of auditor Big and auditor specialist strengthens the value relevance of accounting.
We further partition the sample into Big versus non-Big firms and specialist versus non-specialist firms and run separate regressions (with the same control variables) to tests if the value relevance of accounting is significant for each sub-grouping.
According to panel A with table 11, we observe that in the sub-sample of non-financial companies, the tenure of auditor's Big variable influences positively and significantly the value relevance of accounting of the p-value 1 %. We find that return-earnings relation is significantly positively associated with the tenure of auditor Big. However, the coefficient on the length of the auditor Big-client relationship is negatively but statistically insignificantly. Alternatively, results do not support hypothesis that the tenure of auditor's Big and auditor non-big effect return-earnings relation in the sub-sample of financial companies.
Besides, and according to panel C and D with table 11, we notice that in the sub-sample of not financial companies, the variable which relates to the length auditors specialists-client relationship has a positive and significant impact on the value relevance of the accounting of the p-value 1 %. However, the coefficient of the auditor's non-specialists tenure has a negative but not significant impact on the value relevance. Alternatively, results do not support hypothesis that the tenure of auditor's specialist and auditor non-specialist effect return-earnings relation in the sub-sample of financial companies.
In conclusion, the results show that in the sub-sample of non-financial companies, the length auditors Big-client relationship and the length auditors specialists-client relationship is beneficial and echoes positively on the quality of their service and consequently on the value relevance of the accounting.
Once again, our results confirm that auditor's Big and auditor's specialists produce a better quality of external audit what allows to improve the return-earnings relation.
Tableau 11 : Linear regression into Big versus non-Big firms and specialist versus non-specialist firms
Panel A: return-earnings relation for companies audited by Big auditor
Linear regression on the sub-sample of Tunisian non-financial companies
Linear regression on the sub-sample of Tunisian financial companies
variables
SP
ßi
Z
Z
intercept
-1.319202
-1.10
-.5336053
-0.84
TENURE
+
.0486012***
3.21
.0035729
0.20
SIZE
+
.0922026
1.35
.0462699
1.32
DEBT
-
-.1204561
-0.92
.1437041
1.27
Model on the sub-sample of Tunisian non-financial companies: number of observation = 62
Wald chi2= 13.40*** ; within = 0.2463 ; between = 0.0134 ; overall = 0.0430
Model on the sub-sample of Tunisian financial companies: number of observation = 51
Wald chi2= 3.28
Panel B: return-earnings relation for companies audited by non-Big auditor
Linear regression on the sub-sample of Tunisian non-financial companies
Linear regression on the sub-sample of Tunisian financial companies
variable
SP
ßi
Z
ßi
Z
intercept
.3940687
0.95
.1288569
0.91
TENURE
-
-.0100796
-1.14
.0053416
0.55
SIZE
+
-.0046776
-0.21
-.0001441
-0.02
DEBT
-
-.091502
-0.46
-.0192019
-0.60
Model on the sub-sample of Tunisian non-financial companies: number of observation = 47
Wald chi2= 1.94 ; within = 0.0479 ; between = 0.0283 ; overall = 0.0303
Model on the sub-sample of Tunisian financial companies: number of observation = 69 Wald chi2= 0.8
Panel C: return-earnings relation for companies audited by specialist auditor
Linear regression on the sub-sample of Tunisian non-financial companies
Linear regression on the sub-sample of Tunisian financial companies
variables
SP
ßi
Z
ßi
Z
intercept
-1.305482
-0.93
-.3779642
-0.95
TENURE
+
.0464075***
2.64
.0065305
0.39
SIZE
+/-
.0869279
1.06
.0371056
1.69
DEBT
+ / -
.0176567
0.11
.1218349
1.29
Model on the sub-sample of Tunisian non-financial companies: number of observation=42
Wald chi2= 10.92*** ; within = 0.2875 ; between = 0.0001 ; overall = 0.0518
Model on the sub-sample of Tunisian financial companies: number of observation = 54 Wald chi2= 5.01
Panel D: return-earnings relation for companies audited by non-specialist auditor
Linear regression on the sub-sample of Tunisian non-financial companies
Linear regression on the sub-sample of Tunisian financial companies
variables
SP
ßi
Z
ßi
Z
intercept
.1891156
0.30
.0556279
0.38
TENURE
-
-.0001875
-0.02
.0041859
0.41
SIZE
+
.0094482
0.27
.0042118
0.50
DEBT
-
-.1861161
-1.31
-.016784
-0.52
Model on the sub-sample of Tunisian non-financial companies: number of observation= 67
Wald chi2= 1.74; within = 0.0395 ; between = 0.0001 ; overall = 0.0254
Model on the sub-sample of Tunisian financial companies: number of observation = 65 Wald chi2= 1.35
6. Conclusion
The quality of the external auditor's and their effect on the earnings quality has been an issue of international research in recent years. In this paper we provide evidence on the determinants of auditor independence and competence return-earnings relation, as a proxy for value relevance of accounting, and therefore earnings quality.
Based on a sample of 20 financial and 20 non-financial companies listed in the Tunisian Stock market, over the years 2000-2005, we find that the auditor's Big, the auditor's industry specialization and auditor's experience is significantly positively associated with return-earnings relation. In contrast, audit tenure is insignificantly associated with value relevance of accounting.
We further estimate our regressions separately for four size groups to explore whether the relation between audit tenure and value relevance of accounting differs for different size firms (Big versus non-Big firms and specialist versus non-specialist firms). We find that return-earnings relations vary with the length auditor's Big-client relationship and the length auditor's specialists-client relationship in the sub-sample of non-financial companies and do not vary in the sub-sample of financial companies. We conjecture that auditor's Big and auditor's specialists enhance higher quality of external audit and improve the value relevance of accounting.
Our study contributes to the auditing literature by examining whether the auditor competence and independence affects return-earnings relation. Although prior studies provide some evidence on the relation between auditor reputation, audit tenure and earnings quality (Teoh and Wong, 1993; Johnstone and al., 2002; Carcelle and Nagy, 2004; Ghosh and Moon, 2005; Carey and Simnett, 2006; Gul and al., 2006; Stanley and DeZoort, 2007; Habib and Azim, 2008), they do not examine whether auditor Big and industry specialization of auditors affect the association between auditor tenure and return-earnings relation. Moreover, these studies compare the association between auditor quality and return-earnings relation in the two sub-ample; sub-sample of non-financial companies and the sub-sample of financial companies. By examining the return-earnings relation as determinants of value relevance of accounting with two sub-sample of financial and non-financial companies, covering the years 2000 to 2005, our paper provides additional market-based empirical evidence on the influence of auditor's Big and industry specialization of auditors on the association between auditor tenure and value relevance of accounting.
