“The effect of internal control environment on the value relevance of earnings”

This study examines whether a suitable control environment increases the value relevance of earnings by providing greater assurance on the reliability of financial reporting. Specifically, the level of suitable control environment is assessed by considering the quantity and quality of IC personnel, which are closely related to the personnel integrity/ethical values, competence, and authority/responsibility. Using a sample of 1,834 firm-year observations of Korean listed companies covering 2005–2010, the author finds that earnings are more value relevant when the increase in the average work experience of IC personnel is greater. However, no evidence is found that the value relevance of earnings is positively associated with the increase in the proportion of IC personnel. The findings suggest that a suitable control environment, established by deployment of qualified IC personnel with more work experience, improves the IC effectiveness and, thus, provides greater assurance on the reliability of financial reporting to market investors.


INTRODUCTION
This study investigates the effect of internal control (IC) environment on the value relevance of earnings.The control environment is the foundation of the components in the IC framework for carrying out effective control system (COSO, 2013).A suitable control environment, represented by the personnel integrity/ethical values, competence, and authority/responsibility, is expected to improve the effectiveness of IC and thus provide reliable financial reporting.
According to the guidance to auditors on audit considerations, issued by the Public Company Accounting Oversight Board (PCAOB), the loss of employees integral to the operation of ICs may increase the risk of deficiencies in IC over financial reporting owing to the lack of segregation of duties or effective controls (PCAOB, 2008).In addition, a publication from Deloitte encourages the use of knowledgeable personnel with appropriate skills in control activities to make sure that the ICs are operating effectively (Deloitte, 2013).Therefore, I assume that the management who value the personnel integrity/ethical values, competence, and authority/responsibility would establish a suitable IC environment by allocating more employees and qualified ones with longer work experience.Then, the reported earnings of the firms with a suitable control environment would be perceived as more informative and reliable in the valuation process, thereby leading to an increase in the value relevance of earnings.
Based on the assumption above, I empirically test whether the value relevance of earnings is affected by a suitable control environment, using a sample of 1,834 firm-year observations of Korean listed companies covering 2005-2010.The level of suitable control environment is measured by the ratio of the number of IC personnel to the total number of employees in the firm and the average work experience of IC personnel (Choi et al., 2013; Shin et al., 2016).My findings can be summarized as follows.First, I find that earnings are more value relevant when the increase in work experience of the firm's IC personnel is greater.This suggests that allocation of more experienced personnel to the IC function would send a positive signal to the market that the firm's reported earnings are more reliable and informative.Second, I find no evidence that earnings are more value relevant when the proportion of IC personnel among the firm's employees is greater.This implies that a numerical increase in IC personnel alone does not provide greater assurance on the reliability of earnings.Taken together, my findings indicate that the firms with a suitable control environment composed of more qualified IC personnel is more likely to provide reliable financial reporting.
This study contributes to the literature in several ways.First, this study extends the literature on IC by exploring the association between IC environment and the value relevance of its accounting information.I provide compelling evidence that a suitable control environment is positively related to the value relevance of earnings by using data unique to Korean listed firms, which are required to disclose the number of IC personnel and their average work experience.Second, my findings suggest that a suitable control environment represented by the personnel integrity/ethical values, competence, and authority/ responsibility can be attributed to the quality rather than the quantity of IC personnel; this supports the guidance of Deloitte that emphasizes the use of knowledgeable personnel to ascertain the effectiveness of IC on financial reporting.This study has important implications for managers, auditors, regulators, and capital market participants who are interested in assessing effectiveness of IC, reliability of financial reporting, and the value relevance of earnings.
The remainder of the paper proceeds as follows.Section 1 describes the relevant literature and proposes the hypotheses.Section 2 introduces the study's research method and the model used to test the hypotheses.Section 3 presents the empirical results.Finally, last section concludes the paper.

Institutional background
In the United States, the 2002 Sarbanes-Oxley Act (SOX) was enacted to protect investors by improving transparency and reliability of financial reporting.It was in response to a series of corporate accounting fraud in the late 1990s.Section 404 of the SOX mandates listed firms to establish, maintain, and assess an IC system that provides reasonable assurance regarding the reliability of financial reporting.It also requires external auditors to provide an independent opinion on the adequacy of ICs for financial reporting.Following the passage of SOX in the United States, the Korean government released guidelines and regulations that re-quire firms to establish and maintain an effective IC system for reliable financial reporting.In particular, the guidelines, which were released by the Financial Supervisory Service in 2002, state that Korean listed firms should disclose reports on the operation of their IC system that contains detailed information about the number of IC personnel and their average work experience.This is presumed to reflect the expectation that the number and/or work experience of IC personnel is one of the important determinants of IC effectiveness on financial reporting.

Literature review
Because Section 404 of the SOX requires firms to report the effectiveness of IC on financial reporting, researchers have investigated the association between the weakness of IC and the quality of financial reporting.IC personnel reduces audit reporting lag, a proxy for the timeliness of financial reporting.However, they find no relation between the ratio of IC personnel and audit reporting lag, suggesting that an increase in the number of IC personnel does not necessarily lead to efficient audits.Their findings indicate that the qualitative aspect of IC personnel is one of the important determinants of timely financial reporting, because more experienced IC personnel can improve the efficiency of the audit procedure.

Hypotheses development
It is well known that earnings are reflected in the stock price, because it captures and summarizes the economic performance that determines a firm's value (Beaver, 1998; Francis & Schipper,  1999; Ohlson, 1995; Penman, 1998) 1 .The earnings become more value relevant when they are more reliable; this is because the market would perceive the earnings reported in higher quality of financial reporting to be more useful in evaluating the firm value, and thus place more reliance on them in the valuation process (Francis  & Ke, 2006; Frankel et al., 2002).In other words, the value relevance of earnings is influenced by the market's perception of the reliability of financial reporting.
To provide reasonable assurance for the reliability of financial reporting to market participants, listed firms are required to establish an effective IC system under Section 404 of the SOX.IC system can be effectively operated under a suitable control environment, which is founded on the personnel integrity/ethical values, competence, and authority/responsibility.The level of suitable control environment is closely related to the quantity and quality of IC personnel for the following reasons.First, in guidance to auditors on audit considerations, the PCAOB states that the loss of employees integral to IC function may increase the risk of deficiencies in IC over financial reporting owing to the lack of segregation of duties or effective controls (PCAOB, 2008).2013).In this vein, the Korean Sarbanes-Oxley Act (K-SOX) mandates listed firms to disclose a report on the operation of their IC system that contains information about the number of IC personnel, as well as their average work experience 2 .In addition, qualified IC personnel with more work experience can implement effective communication channels based on their understanding of a firm's structure, accounting procedures, and relevant laws/regulations, thereby leading to higher quality of financial reporting.Thus, I hypothesize that the value relevance 2 K-SOX, which is the Korean equivalent of SOX, is a set of regulations on the IC system in Korean listed companies.3 In Korea, external auditors evaluate the effectiveness of internal control system of firms and express their review opinion under K-SOX.Thus, the disclosure of internal control weakness would be used as a good proxy of a suitable control environment.However, unfortunately, I cannot use this proxy for the following reasons.First, the relation between internal control weakness and the quality of earnings/market reaction is already tested by other studies.Prior studies have documented that internal control weakness is negatively related to the quality of earnings (Ashbaugh-Skaife et al  , 2008).Second, the ratio of firms reporting internal control weakness is quite low.According to Lee (2015), the firms that are reported to have internal control weakness account for only about 3% of the total firms sampled during the period from 2005 to 2010.It means that the level of a suitable control environment in most of firms without internal control weakness is not precisely measured by this proxy.4 Value relevance is usually measured as the statistical association between accounting information and stock price/returns.In this study, I focus on the value relevance of earnings, which is among the most representative accounting information on firms' economic performance and widely used by market participants.of earnings will be positively associated with a suitable IC environment, measured by the quantity and quality of IC personnel 3 .
Hypothesis: The value relevance of earnings will be positively associated with a suitable IC environment.

Model specifications
For the value relevance of earnings, I apply a typical regression model to investigate the statistical association between stock returns and change in earnings 4 .In model

Less:
Missing observations in obtaining IC personnel data 393 Observations whose stock return and other financial data are not available in the KISVALUE database 144 Final sample firm-years 1,834

Note:
The variables are defined as in Table 2. http://dx.doi.org/10.21511/imfi.16(2).2019.16 TOP subgroup are higher than those of BOTTOM subgroup.It implies that the value relevance of earnings is positively associated with the level of suitable control environment, measured by the allocation of more experienced IC personnel.To ensure that multicollinearity is not driving my results, I check for the variance inflation factor (VIF) in all empirical tests as there are highly correlated variables.The VIFs are all less than 3, indicating that multicollinearity is not a serious concern in my analysis. 8 For the rest of analyses, I also have conducted a 3-step hierarchical regression and their results are unchanged.For brevity, only the result of step 3 regression is tabulated in the manuscript.
For more sophisticated analysis, I investigate the moderating effect of ΔIC1 (ΔIC2) on the value relevance of earnings by conducting a 3-step hierarchical regression analysis as follows 8 : .
The hierarchical results for the moderating effect of ΔIC1 (ΔIC2) are summarized in Table 5. Panel A of Table 5 shows that ΔIC1 has no moderating effect on the value relevance of earnings, because the introduction of interaction term (

E IC ∆ ⋅∆
) does not significantly increase R 2 and the coefficient of

E IC ∆ ⋅∆
is not significant in step 3. On the other hand, Panel B of Table 5 reveals that ΔIC2 moderate the relationship between ΔE and RET.In step 3, the introduction of interaction term (

E IC ∆ ⋅∆
) significantly increase R 2 , and the coefficient of is positively significant at 5% level.The R 2 change is tested with an F-test (p-value = 0.049 < 0.5).The coefficient of moderator variable (ΔIC2) is not significant, indicating that ΔIC2 is a pure moderator (Sharma et

E IC ∆ ⋅∆
) is still positive and statistically significant at the 5% level, whereas the coefficient of the interaction term between ΔE and ΔIC1 (

E IC ∆ ⋅∆
) is not.This confirms the work experience of IC personnel is a more important determinant of the improvement in the reliability of financial reporting through effective ICs than the number of IC personnel.
Additionally, I conduct the test by using an alternative measure of IC personnel quantity, ΔIC1_N, which measures the change in the ratio of IC personnel to total employees in the IC-related departments, such as accounting, finance, audit, and information technology.Column B of Table 6 reports the test results using the alternative measure.Overall, the results remain qualitatively similar when the alternative measure is applied.This confirms the view that a suitable control environment represented by the personnel integrity/ethical val-Table 6. Analysis on the effect of IC environment on the value relevance of earnings after controlling for the effect of the quantity and quality of IC personnel on each other ues, competence, and authority/responsibility can be attributed to the quality rather than the quantity of IC personnel.

Additional analysis
In this subsection, I conduct an additional analysis to mitigate the concerns on the variable related issues.First, the mean value of IC1 (0.077) is greater than the median value (0.033) as shown in Table 1.It indicates that the distribution of IC1 is skewed to the right because there are a few firms with extreme value of IC1 in the sample.To rule out the effect of these outliers, I additionally winsorize the upper 10% of the observations for IC1 9 .Second, in the main analysis, IC2 is defined as the log of one plus average work experience of the IC personnel in months following prior studies (Choi et al., 2013; Shin et al., 2016).In an ad-9 The upper 10% of values are set equal to the maximum value in the 90th percentile.After this winsorization, the mean (median) value of IC1 becomes 0.047 (0.033).
ditional analysis, I use modified IC2 calculated by the log of 1 plus average work experience of the IC personnel in years.Third, I also use modified STDSALES, which is scaled by total assets instead of the market value of equity to eliminate the problem caused by the fluctuation of the market value of equity.
Table 7 presents the results of the analysis using the modified variables (∆IC1, ∆IC2, and STDSALES).
The coefficient of the interaction term between ∆E and ∆IC2 is still positive and statistically significant, whereas the coefficient of the interaction term between ∆E and ∆IC1 (

E IC ∆ ⋅∆
) is not.Overall, Table 7 provides robust results and support the hypothesis that the value relevance of earnings is increased by the level of suitable control environment, established by qualified IC personnel with more work experience.Notes: The variables (except ∆IC1, ∆IC2, and STDSALES) are defined as in Table 2. *, **, and *** indicate significance levels at the 10%, 5%, and 1%, respectively, based on two-tailed tests.VIFs are all less than 3. ∆IC1 -change in the ratio of IC personnel to total employees in the firm; the upper 10% of the observations for IC1 is additionally winsorized; ∆IC2 -change in log of one plus average work experience of the IC personnel in years; STDSALES -sales variation over a 3-year period; standard deviation of sales from fiscal year t-3 to fiscal year t, scaled by the total assets at the beginning of fiscal year t.

CONCLUSION
This study examines the effect of IC environment on the value relevance of earnings.Specifically, I consider the quantity and quality of IC personnel to assess the level of suitable control environment which is determined by the personnel integrity/ethical values, competence, and authority/responsibility.
Using a sample of 1,834 firm-year observations of Korean listed companies covering 2005-2010, I find that earnings are more value relevant when the increase in work experience of the firm's IC personnel is greater.This suggests that the allocation of more experienced personnel to the IC function provides greater assurance on the reliability financial reporting.However, I do not find a significant association between the value relevance of earnings and the increase in the ratio of IC personnel.This indicates that simply adding more IC personnel does not provide greater assurance on the reliability of earnings.These results imply that the effectiveness of ICs on financial reporting is likely to be attributed to a suitable control environment, established by qualified IC personnel with more work experience.The limitation of this study is that I use the data during the period of financial crisis.However, this study provides compelling evidence that the value relevance of earnings is positively affected by qualitative investment in IC personnel by analyzing a set of data unique to Korean listed companies.Evidence from this study can help managers, auditors, regulators, and market participants better understand the determinants of effective ICs for financial reporting from a control environment perspective.

Table 1 .
7he initial sample consists of companies listed on the Korea Stock Exchange during 2005-20107.To ensure the homogeneity of the analyzed companies, I exclude a sample of firms in the financial industry and firms whose accounting year does not end in December.I collected data on the IC personnel of firms from their annual reports in the Financial Supervisory Service website (http://dart.fss.or.kr).The stock price and financial data are obtained from the KISVALUE database (equivalent to the CRSP in the United States).Firms without appropriate IC personnel data are excluded; this removes 393 firm-year observations.I also eliminate 144 firm-year observations for which stock price and financial data are not available in the KISVALUE database.My final sample consists of 1,834 firm-year observations.The sample selection process is described in Table1.Sample selection procedure 1, market-adjusted return over 12 months, RET, is regressed on the change in earnings before tax (ΔE).The coefficient of ΔE captures the value relevance of earnings.The level of suitable control environment is measured by the change in the ratio of IC personnel to total employees (ΔIC1) and the change in the average work experience of a firm's IC personnel (ΔIC2).To verify the hypothesis, I partition the total sample into two subgroups by ranking observations based on ΔIC1 (ΔIC2) -one consisting of firms above the 5 th quintile (referred to as "TOP" subgroup) and the other of firms below the 1 st quintile (referred to as "BOTTOM" subgroup).The TOP subgroup is considered as the firms with the highest level of control environment, and the BOTTOM subgroup is considered as those with the lowest.I then estimate regression model 1 for each subgroup.If the value relevance of earnings is positively associated with the high level of control environment, the coefficient of ΔE in the TOP subgroup is expected to be greater than that of the BOTTOM subgroup.http://dx.doi.org/10.21511/imfi.16(2).2019.166STDSALESdenotes the variability of sales, calculated based on annual data(Francis et al., 2004).7Following the government's roadmap to adopt the International Financial Reporting Standards (IFRS), Korean listed firms were required to prepare their financial statements under Korean International Financial Reporting Standards (K-IFRS) from the beginning of 2011.Prior to the adoption of IFRS, financial statements were prepared in accordance with the Korean Generally Accepted Accounting Principles (K-GAAP).Thus, I use data until 2010 to rule out the effect of the change in accounting standards on the value relevance of earnings.(SIZE)andmarket-to-bookvalue (MTB) are significantly related to the stock return.The leverage (LEV) is used as a proxy of the riskiness of debt or default risk, which is negatively related to the stock return(Dhaliwal, 1991).As prior studies consider various financial ratios to estimate stock prices, ROA, GROWTH, and LOSS are added to the models (Fama & French, 2006; Nissim & Penman, 2001).Also, sales variability (STDSALES) is used as a proxy of business risk (Reilly & Brown, 2011; Rattiner, 2008) 6 .To control for the effects of ownership structure and external monitoring by foreign investors, OWN and FSH are included in the models (Choe et al., 1999; Lemmon & Lins, 2003).2.2.Sample selection procedure

Table 2 .
Descriptive statistics of variables t-3 to fiscal year t, scaled by the market value of equity at the beginning of fiscal year t; MTB -market to book ratio; OWN proportion of ownership held by the largest shareholder (including family members and other related parties) of the firm; FSH -proportion of ownership held by foreign investors.

Table 4
presents the results of model 1 estimated for TOP and BOTTOM subgroups.In specific, Panel A of Table4provides the results of model 1 estimated for two subgroups divided based on ΔIC1.The coefficient of ΔE is not statistically significant for BOTTOM subgroup, while it is significantly positive for TOP subgroup.It suggests that the value relevance of earnings is observed only in the firms with the highest level of suitable control environment, measured by the allocation of more IC personnel.Panel B of Table4provides the re- sults of model 1 estimated for two subgroups divided based on ΔIC2.The coefficients of ΔE are both significantly positive for two subgroups, but the coefficient of ΔE and its significance level for

Table 3 .
Correlations of variables

Table 4 .
Analysis on the value relevance of earnings for subgroups

Table 5 .
Analysis on the effect of IC environment on the value relevance of earnings

Panel B. Model 3 Variables Step 1 Step 2 Step 3 Coefficient t-stat Coefficient t-stat Coefficient t-stat
The variables are defined as in Table2.*,**,and***indicatesignificancelevels at the 10%, 5%, and 1%, respectively, based on two-tailed tests.., 1981).The results suggest that the earnings are more value relevant when the change in the average work experience of the IC personnel increases.This is in accordance with Ge andMcVay (2005), who find that the lack of expertise in IC can lower the quality of financial reporting.In sum, the results imply that the value relevance of earnings increases as the level of suitable control environment increases, even though the positive association is observed only when the level of suitable control environment is assessed in perspective of the quality of IC personnel.Table6presents the results of the analysis on the effect of IC environment on the value relevance of earnings, after controlling for the effect of the quantity and quality of IC personnel on each other.Column A of Table6provides the results of estimating model 3, which includes both ΔIC1 and ΔIC2.The coefficient of the interaction term be-tween ΔE and ΔIC2 ( Notes:http://dx.doi.org/10.21511/imfi.16(2).2019.16al

Table 7 .
Additional analysis on the effect of IC environment on the value relevance of earnings using modified variables