Institutional Proprietorship, Audit Committee Size, and the Reputable Auditor Selection: Evidence from Indonesia

–This study is presented to investigate the effect of institutional proprietorship and audit committee size on the propensity of the companies to select outside auditors affiliated with the big four and know the accuracy of prediction based on these determinants. Following several previous scholars, this study uses the company size as the control variable. Furthermore, the logistic regression model and matrix classification are employed to answer the research intention by employing the Indonesian non-financial companies in the LQ45 index from 2014 to 2018 as the population. Moreover, to search for the total samples of 21 and take them from the populace, this study uses the Slovin formula with an inaccuracy border of 10% and a simple random sampling technique. Once examining the data, this study finds that institutional proprietorship and audit committee number positively affect the corporate propensity to choose the auditor affiliated with the big four. Similarly, this tendency happens in the firm size as the control variable. Additionally, the accuracy of grouping prediction of the firms using reputable and non-reputable auditors based on these determinants is 90.50%, 95.2%, 90.5%, 85.7%, and 85.7% for 2014, 2015, 2016, 2017, and 2018.


INTRODUCTION
Public companies in Indonesia must announce their financial report four times a year.Three times are for the report publication for three months, and the final time is for an annual report.The delivery of these reports cannot be late.If the firms are not on time to publish them, they will get the sanction, from a fine to the revocation of business license, set by the Financial Service Authority (Setyastrini & Kaluge, 2019).
Before publishing this annual financial report, the public-listed companies must appoint an independent external auditor to audit it (Setiadi, 2019).This auditor enhances financial reporting quality through its credibility (Alsmady, 2022).Therefore, many firms opt to hire trustworthy ones affiliated with the big four to get support from the market, improving their value in the capital market (Almashaqbeh et al., 2020;Wijaya, 2020).Besides getting a positive market response, another advantage of hiring the public accounting office associated with the big four auditors is earnings superiority (Agyei-Mensah & Yeboah, 2019;Lopes, 2018;Mollik et al., 2020).
The studies investigating the tendency determinants of choosing public accounting linked to the big four auditors are numerous, such as those carried out by Al-Hajri (2018), Alzeaideen and Al-Rawash (2018), Al-Hajaya (2019), Septiana and Khafid (2019), Sulistyawati and Agustina (2019), Gerged et al. (2020), and Ngo et al. (2020), Corten et al. (2021), and Thu and Khanh (2022).Based on their research, at the least possible, three contributing factors exist, i.e., institutional proprietorship and committee audit size becoming the primary variable and company size as the control variable.
Indonesia is a state adopting civil law in its legal system (Dimyati et al., 2021).In this system, the law is less robust in protecting public shareholders in the capital market (Sriwidharmanely et al., 2022).The controlling shareholders with a dominant position in the share portion tend to expropriate the public shareholders (Juniarti & Yuyetta, 2021) by optimizing discretion in line with their interests, called entrenchment (Anafiah et al., 2017).Furthermore, Anafiah et al. (2017) explain that these shareholders appoint reputable auditors to overcome this entrenchment effect.By hiring them, these shareholders signal that they protect the public shareholders by providing qualified financial reports.This explanation is confirmed by Alzeaideen and Al-Rawash (2018), Gerged et al. (2020), and Ngo et al. (2020), concluding the positive tendency of institutional proprietorship to use reputable auditors.By indicating these elucidations, this study proposes the first hypothesis (H1): Institutional proprietorship stimulates the propensity of the company to hire a respected auditor.
In firm governance, the audit committee is responsible to the supervisory board (Tambunan & Tambunan, 2021) for maintaining high-quality financial reporting (Al-Aamri et al., 2021).Because reputable external auditors can reduce earning management (Lopes, 2018), the audit committee recommends that the firm utilize them (Al-Hajri, 2018).In their investigation, Al-Hajri (2018) documents that the more people in the audit committee position, the higher propensity to employ the big four auditors.Furthermore, this positive tendency is confirmed by Al-Hajaya (2019).Also, Septiana and Khafid (2019) prove that the more influential the audit committee, the higher the possibility of choosing trustworthy auditors.By indicating these illuminations, this study proposes the second hypothesis (H2): the auditor committee size stimulates the propensity of the company to hire a reputable auditor.
The giant company utilizes the service of the external auditor with a reputation because of having numerous transactions.Besides, several possible mistakes become another consideration for this firm to use it (Sulistyawati & Agustina, 2019).In their investigation, Alzeaideen and Al-Rawash (2018), Al-Hajaya (2019), Sulistyawati andAgustina (2019), Gerged et al. (2020), Corten et al. (2021), andThu andKhanh (2022) verify that the greater the firm, the higher preference to employ respectable auditors.According to these explanations, the third hypothesis (H3) can be communicated this way: the size stimulates the propensity of the company to hire a reputable auditor.

Fundamental Research Framework
The quantitative design is applied in this study.According to Creswell and Creswell (2018), this design aims to validate the hypotheses statistically.Because it involves statistical technique, by referring to Gujarati et al. (2019), this study utilizes the probability of Z-statistic to examine the coefficients in the logistic regression model.Additionally, the first, second, and third hypotheses are that institutional share proprietorship, audit committee size, and company size stimulate the propensity to hire a reputable auditor.Moreover, this model is available in figure one.

Research Variable Measurement
This research employs the dummy as the explained variable to measure the reputable auditor selection.If the company decides to hire the big-four auditor, one is given for the notation, and zero is set as the symbol if the company does not.Furthermore, the share portion of institutions for the first explaining variable is used to measure the institutional proprietorship at the end of the year.Moreover, the total number of people becoming an audit committee is utilized to measure its size at the end of the year as the second explaining variable.For the third explaining variable, the natural logarithm of total assets at the end of the year performs as the company size measurement.

Population and Sample
This research population is the twenty-six non-financial companies becoming the LQ45 index members from 2014 to 2018.To calculate the total samples (TS), we use the Slovin method utilizing the inaccuracy border (IB) of 10%.
Likewise, the Slovin formula is presented in the first equation.
Based on this formula, the total samples are = 20.63 ≈ 21 companies.Then, they are taken from the population by simple random sampling.To result in random numbers, we use Microsoft Excel, as Cheusheva (2022) explains.After that, these numbers guide selecting the company as the sample.Finally, twenty-one companies are obtained, and their name is available in Table 1.

Data collection technique
This study uses the secondary data collection technique by mentioning Bougie and Sekaran (2020).In this context, we employ the information from the annual reports of the associated firms on the Indonesian stock exchange as the sample (see Table 1) and the papers published in the linked journals as the reference to create hypotheses.

Model to analyze the data
This study adopts the logistic regression model to analyze the data by denoting the variable scales, as Gujarati et al.

RASit = ln
As a rule, the residual in the logistic regression model does not need to be normally distributed.Also, the variance of residuals between firms using the big-four auditor and the non-big-four auditor does not need to be homogeneous (Ghozali, 2021).To prove them, furthermore, the Jarque-Bera normality check (Gujarati et al., 2019) and homogeneous variance examination by M-Box are implemented (Hair Jr. et al., 2019).Before testing the regression coefficients, indeed, this model has to achieve the goodness-of-fit test.By denoting Hair Jr. et al. (2019), the examination of the additional variables is used.Moreover, this study uses the Z-statistic test for the logistic regression model regression coefficient (Gujarati et al., 2019).Likewise, the classification matrix is employed to get the grouping accuracy based on these explaining variables (Hair Jr. et al., 2019).

The descriptive statistics
The number of samples utilized is 21 companies; hence, the number of observations is 105.To statistically describe the chosen auditor by company, we employ the frequency, as Table 2 demonstrates.
Table 2.The frequency of the company as the user of auditor affiliated with the big four and non-big four

Description
The number of observations (frequency)

Company code
The company consistently employs an auditor affiliated with the big four.To statistically describe three explaining variables having the ratio scale, we use the total observations (N), the bottommost, the uppermost, average, and the standard deviation for IP, ACS, and LN(TA), as seen in Table 3.This table is the output of IBM SPSS 19.

The correlation matrix
The correlation assesses the power of the relationship among research variables (Gujarati et al., 2019).The correlation employed in this section is the Pearson coefficient, and the rule to interpret the value is explained by Ananda and Fadhli (2018): If the coefficient is below 0.2, the correlation is unavailable.Suppose the coefficient is from 0.2 to 0.4 and 0.4 to 0.7; the correlation is unconvincing and medium, respectively.Finally, if the value is from 0.7 to 0.9 and 0.9 to 1, the correlation is strong and exceedingly powerful, one-to-one.The fourth table shows the correlation matrix consisting of the coefficient between RAS and IP of 0.360, RAS and ACS of 0.263, RAS and LN(TA) of 0.220, and IP and LN(TA) of 0.291.Because these values are between 0.20 and 0.40, a weak correlation exists by denoting Ananda and Fadhli (2018).Besides, the coefficient between IP and ACS is 0.062, less than 0.2; hence, the correlation is absent based on Ananda and Fadhli (2018).Meanwhile, the correlation of ACS with LN(TA) is 0.405.Because this value is near 0.40, as Ananda and Fadhli (2018) depict, the relationship is almost weak.

The result of the logistic assumption testing
The fifth table shows the initial verification for the logistic regression model, i.e., normality and variance homogeneity.The normality test result based on the Jarque-Bera statistic shows a probability of 0.00000, above a 5% significance level.Hence, the residuals are not normally distributed.The variance homogeneity test result of the M Box statistic shows the F-statistical probability of 0.004, exceeding a 5% significance level.Therefore, heterogeneous variance exists.By denoting two results demonstrating non-normality and heterogeneity, this model already achieves the preliminary testing for this model.

The estimation result of the regression model
After attaining two opening tests in the previous section, we check the goodness-of-fit model.To begin, we estimate two models.The first model consists of intercept (restricted model), and the second model contains intercept and regression coefficients β1, β2, and β3 (unrestricted model).Then, we calculate the probability of the log-likelihood ratio due to the additional explaining variables: IP, ACS, and LN(TA) into the model by E-Views 6.Fortunately, the Chi-Square statistical probability is 0.0001, lower than the 5% significance level (see the unrestricted model in Table 6).Therefore, adding three explaining variables is significant so that the model fits with the data.In the model without restriction in Table 6, the Z-statistical probability for IP, ACS, and LOG(TA) is 0.0027, 0.0883, and 0.0798, respectively.This situation means a significant positive sign at a significance level (α) of 5% for IP, 10% for ACS, and LN(TA); hence, all the null hypotheses are prohibited.Instead, the first, second, and third alternative hypotheses are affirmed.In other words, it can be statistically decided that institutional proprietorship, audit committee size, and company size positively affect the propensity to utilize a reputable auditor.

The classification matrix result
Besides performing estimation, the logistic regression model provides the classification matrix to assess the grouping accuracy of the firms using reputable auditors and non-reputable auditors based on the explaining variable.In this study, we get this matrix based on the output of IBM SPPS 19.This research covers five years; thus, each year's accuracy level can be counted (see Table 7).Furthermore, the degree for 2014 and 2015 is 90.5% and 95.2% (see Table 7A), and for 2016, 2017, and 2018 is 90.5%, 85.7%, and 85.7%, respectively (see Table 7B).

Discussion
Based on the statistical examination result, the first hypothesis is acknowledged: the institutional proprietorship stimulates the propensity of the company to hire a reputable auditor.Although having to pay a premium fee for audit services, the institutions controlling the firms keep doing it because the benefits taken are more incredible than the costs.Reputable auditors can be trusted to limit and cut earnings management, leading to the quality of financial reporting.Besides, another motive of controlling shareholders to hire them is to decrease the entrenchment effect.By owning the positive impact, this study supports Alzeaideen and Al-Rawash (2018)  After denoting the statistical testing result, the second hypothesis is acknowledged: the audit committee size stimulates the propensity of the company to hire a reputable auditor.It means the higher the total number of members of the audit committee, the greater the tendency of companies to select qualified external auditors.The big fours are proven to assist the audit committee members in ensuring their audit quality.Therefore, this study confirms Al-Hajaya (2019) investigating 23 insurance firms in the Amman stock exchange between 2013-2017, documenting a positive relationship between the number of the audit committee and reputable auditor selection.Although using different measurements of the audit committee, i.e., score based on activities, total members, and competence, the investigation by Septiana and Khafid (2019), with 89 financial companies in the Indonesia capital market in 2017 as the samples, supports this tendency.
The third hypothesis is recognized by mentioning the statistical testing result: The firm size as the control variable is significantly proven to have a positive tendency to select a reputable auditor.Hence, this result is in line with Alzeaideen and Al-Rawash (2018) and Al-Hajaya (2019) examining this relationship using the Jordanian publiclisted firm data.After testing the data of seven companies with governance perception index between 2010-2014 in the Indonesian stock exchange, Sulistyawati and Agustina (2019)

CONCLUSION
The selection of reputable auditors needs to be considered by firms listed on the capital market because of favorable market reactions and costly fees.This circumstance motivates this study to investigate the impact of institutional proprietorship and audit committee size as the foremost cause and company size as the control variable on this decision.Furthermore, 21 non-financial companies of the LQ45 index on the Indonesian stock exchange from 2014 to 2018 are employed as the sample.By utilizing the logistic regression model with pooling data, this study reveals a positive effect of institutional ownership and audit committee size on reputable auditor selection.Also, the company size shows the same indication.Based on this determinant, the grouping accuracy rate of companies hiring reputable and non-reputable auditors is 90.50% for 2014, 95.2% for 2015, 90.5% for 2016, 85.7% for 2017, and 85.7% for 2018.As a limitation, this study only uses the non-financial firms becoming LQ45 index constituents, five years as an observational period, and the affecting factors.These circumstances open the chance for the following scholars to directly utilize non-financial firms in the Indonesian capital market to enlarge the population scope and extend years of observation to 10 years.The other primary determinants suggested for their research model are foreign ownership, the supervisory board size, supervisory board independence, managerial ownership, and leverage.

Figure 1 .
Figure 1.The research model to examine the tendency of institutional proprietorship, audit committee size, and company size to select reputable auditor used a big-four auditor from 2014 until 2016 and in 2018.However, in 2017, it hired a non-bigfour auditor.JSMR utilized a non-big-four auditor in 2014 but employed a big-four auditor between 2015 and 2018.
after investigating the data of 132 firms in the Amman stock exchange from 2005 to 2016.Furthermore, this positive tendency is acknowledged by Gerged et al. (2020) when studying 180 companies becoming members of the Financial Times Stock Exchange 350 Index in the London capital market from 2012 to 2017.Also, this positive propensity is confirmed by Ngo et al. (2020) after examining the data of 511 non-financial firms listed on the Vietnamese stock exchange between 2015 and 2017.
affirm this evidence.Besides, this study confirms Gerged et al. (2020) testing this association by employing 180 corporations becoming the Financial Times Stock Exchange 350 Index in London between 2012 and 2017.Additionally, Corten et al. (2021) confirm this evidence after investigating 316 private non-financial companies in 2015 in Belgia.After using 596 non-financial firms in the Ho Chi Min capital market from 2014 through 2017, Thu and Khanh (2022) demonstrate similar proof.

Table 3 .
The total observations, minimum, maximum, mean, and standard deviation of the explaining variables

Table 4 .
The Pearson Correlation Matrix Result

Table 5 .
The test result of normality and variance homogeneity

Table 6 .
The estimation result of the logistic regression model

Table 7A .
The result of the classification matrix for 2014 and 2015

Table 7B .
The result of the classification matrix for2016, 2017, and 2018