Internal control weakness and corporate employment decisions: evidence from SOX Section 404 disclosures

ABSTRACT This study investigates the impact of material internal control weaknesses on corporate employment decisions. We find that, on average, ineffective internal control is significantly related to lower efficiency in employment decisions. We also find that firms with material internal control weaknesses are associated with both over-investment and under-investment in labour. Further analysis suggests that the negative impact of internal control weaknesses on employment decisions is predominantly driven by more severe types of weakness that have a pervasive effect on internal reporting and those related to core accounts. Moreover, our change analysis shows that the remediation of material weaknesses contributes to an improvement in labour investment efficiency. Finally, consistent with the effective monitoring role of high-skilled employees, our subsample analysis indicates that the negative impact of internal control weaknesses on labour investment efficiency is mitigated in firms with high reliance on human capital. Our findings are robust to various sensitivity checks including propensity score matching, entropy balancing, removal of observations during the financial crisis, various measures for the efficiency of investment in labour, and the adjustment for employing residuals as outcome variables. Overall, our study contributes to the ongoing debate on the net benefits of SOX 404 by highlighting the significant value of internal control systems to efficient human capital investment and provides timely implications for managers and regulators.


Introduction
In this paper, we examine the impact of internal control weaknesses on labour investment efficiency.Since the implementation of Sarbanes-Oxley Section 404 (SOX 404) in 2004, which requires managers and external auditors to comment on the adequacy of the internal controls over financial reporting, there has been a contentious and ongoing debate on the effectiveness of SOX 404 amongst academics, practitioners, and regulators.On the one hand, several studies have documented the economic benefits of effective internal controls, such as better financial reporting quality (Ashbaugh-Skaife et al., 2008;Doyle et al., 2007a;Epps & Guthrie, 2010) and lower cost of capital (Dhaliwal et al., 2011;Kim et al., 2011).On the other hand, other studies suggest that the disclosure of internal control weaknesses may not materially influence firms either financially or operationally (Alexander et al., 2013;Aobdia et al., 2020;Ogneva et al., 2007).On top of this, many scholars and practitioners have long been critical of the extremely high compliance costs associated with SOX 404 (DeFond & Francis, 2005), which casts further doubt on the net benefits of SOX 404.
Given the continuing controversy and inconclusive evidence on the real effects of SOX 404, in this study, we aim to contribute to this important debate by investigating the influence of effective internal control on a crucial corporate decision.Arguably, one of the most significant corporate decisions that firms have to make is their investment in labour. 1 Despite the proliferation of research on capital investment, few accounting studies have addressed the investment in labour, as a key factor of production (Falato & Liang, 2016;Jung et al., 2014;Pinnuck & Lillis, 2007).Numerous previous studies highlight the economic significance of the efficiency of investment in labour and the urgent need for firms to optimize it (Ben-Nasr & Alshwer, 2016;Ghaly et al., 2020;Ha & Feng, 2018;Jung et al., 2014;Khedmati et al., 2020;Zhang et al., 2020).Thus, deviations from optimal labour investment in the form of either over-or under-investment can be extremely costly to firms (Ghaly et al., 2020;Jung et al., 2014).On the one hand, the expansion of the workforce beyond the optimum leads to over-capacity issues and firms need to devote scarce corporate resources to cover the extra costs of excess labour.On the other hand, under-investment in labour means firms underutilize corporate resources, leading to insufficient growth and low productivity (Stein, 1989;Williamson, 1963).As human capital plays an increasingly important role in the economy (Zingales, 2000), maintaining efficient labour investment has become a top priority for modern firms.
In light of the contentious debate on the efficacy of SOX Section 404 (Aobdia et al., 2020;Feng et al., 2015;Kim et al., 2011;Ogneva et al., 2007) and the economic significance of labour investment, our study examines the impact of internal control weaknesses on the efficiency of investment in labour, a major internal stakeholder that is relatively underexplored in the accounting literature.We hypothesize that internal control ineffectiveness can have a significantly adverse impact on labour investment efficiency through two dimensions: (1) erroneous internal management reports used for operational decision-making and (2) information asymmetry arising from the imperfections in the capital market.First, ineffective internal control can influence labour investment efficiency via erroneous or stale internal management reports used for employment decisions (Cheng et al., 2018;Feng et al., 2015).Given managers rely on internal management reports for daily operational decision-making (Feng et al., 2015), an ineffective internal control system can undermine managers' operation decisions, including labour investment decisions.For instance, over-forecasting of sales or under-forecasting of operating expenses may misleadingly depict a high-growth trend that requires expansion of labour investment, which can potentially result in over-investment.In contrast, under-forecasting of sales or over-forecasting of operating costs may lead firms to reduce labour below the optimal level, thus resulting in under-investment.Second, we also argue ineffective internal control systems exacerbate information asymmetry, resulting in inefficient labour investment.On the one hand, managers of firms with ineffective internal control may engage in self-serving behaviours (e.g.empirebuilding activities) that are not in the best interests of shareholders (Jensen & Meckling, 1976), which ultimately lowers labour investment efficiency.On the other hand, firms with internal control weaknesses tend to have lower financial reporting quality and higher costs to finance labour (Altamuro & Beatty, 2010;Ashbaugh-Skaife et al., 2008, 2009;Cheng et al., 2013;Doyle et al., 2007aDoyle et al., , 2007b;;Feng et al., 2009;Kim et al., 2011), which may, in turn, lead to under-investment in labour.
While the aforementioned studies seem to support our prediction that ineffective internal control can impair labour investment efficiency, it is also possible that such weaknesses might not affect the efficiency of a firm's investment in labour.For instance, Alexander et al. (2013) show that managers do not believe that SOX 404 leads to better operational decisions.Moreover, Ogneva et al. (2007) find that internal control weaknesses are not directly associated with a higher cost of equity, which suggests that adverse selection might be less relevant to the relationship between internal control weaknesses and labour investment efficiency.Given the arguments above, whether internal control effectiveness can influence the efficiency of investment in labour is ultimately an empirical question.
Using a sample of 3,028 U.S. firm-year observations over the period from 2004 to 2016, we find that firms with internal control weaknesses are associated with inefficient labour investment2 and are likely to suffer both over-and under-investment.We also investigate the association between internal control weakness and labour investment efficiency by considering the severity of material weaknesses.Given the negative effect of internal control weakness on labour investment efficiency, severe material weaknesses are expected to play a key role.Following Cheng et al. (2018), we define two types of severe internal control weakness: (1) pervasive weakness and (2) core accountingrelated weakness.In line with these arguments, our results show that the adverse influence of internal control weaknesses on labour investment efficiency is primarily driven by those two types of severe internal weaknesses.
To corroborate our main results, we follow extant literature (Cheng et al., 2018;Feng et al., 2015) and conduct a change analysis to investigate the influence of remediation of material weaknesses on the change in labour investment efficiency.If ineffective internal control systems contribute to lower efficiency of investment in labour, we expect efficiency to improve if firms remediate their internal control weaknesses.Consistent with our prediction, we find that the remediation of internal control weaknesses is associated with a significant increase in labour investment efficiency, lending additional support to our main finding that internal control weaknesses result in inefficient labour investment.
Subsample analysis using multiple proxies for a firm's human capital intensity provides consistent evidence that the impact of internal control weaknesses on labour investment efficiency is concentrated in firms in low-skilled industries, whereas the impact is mitigated in human-capital-intensive firms, consistent with high-skilled employees playing an effective internal governance role.
We conduct a battery of additional tests to alleviate endogeneity concerns and check the robustness of our findings.First, we employ propensity score matching (PSM) by matching firms (control group) with firms that report internal control weaknesses (treated group) based on the full set of control variables in our main model.For each treated firm, we select the nearest neighbour in the same industry (2-digit SIC) and the same year.We find the results support our main findings and show that firms with internal control weaknesses have significantly higher abnormal net hiring (i.e.inefficient employment decisions) than firms without internal control weaknesses.Second, to supplement our PSM analysis and ensure our results are not sensitive to a particular matching technique, we also employ an entropy balancing approach to ensure covariate balance between firms with internal control weaknesses and firms without internal control weaknesses.We obtain consistent results when repeating the analysis using an entropy-balanced sample.Third, to rule out the potential confounding effect that our results may be driven by non-labour investment or the occurrence of the financial crisis, we re-run our analyses and find our results remain unchanged after considering the potential effect of non-labour investments and the financial crisis.Fourth, our results are robust to various alternative measures of labour investment efficiency used in previous studies (Ben-Nasr & Alshwer, 2016;Ghaly et al., 2020;Jung et al., 2014).Fifth, Chen et al. (2018) show that employing residuals (i.e. the unexpected or abnormal component) as outcome variables leads to biased coefficient estimates and unreliable tstatistics, which can result in wrong inferences.In light of the solutions proposed by Chen et al. (2018), we include all the regressors of the two-step regression procedure in our model to alleviate the concern of employing residuals as the outcome variables.Our results still hold after the inclusion of additional regressors.Finally, we also use different fixed effects and clustered at both firm and year levels and we find our results are largely unchanged.
Our study contributes to the literature in multiple ways.First, it concerns the economic consequences of ineffective internal control and contributes to the ongoing debate on the net benefit of SOX 404.Although previous studies have highlighted the negative consequences of internal control weaknesses (Altamuro & Beatty, 2010;Ashbaugh-Skaife et al., 2008;Doyle et al., 2007aDoyle et al., , 2007b;;Feng et al., 2009), our study is noticeably distinct from their studies by extending our knowledge regarding the implications of internal control weaknesses to investment efficiency in labour, an important production factor and internal stakeholder within businesses.To the best of our knowledge, our paper is the first study that attempts to fill this void in the literature by focusing on the relationship between internal control weaknesses and labour investment efficiency.By doing so, our study adds to the stream of literature on internal control weakness by highlighting the adverse influence of internal control weaknesses on labour investment efficiency, which distinguishes the contributions of our study from previous studies.
Second, different from the prior literature, which largely focuses on the quality of external financial information, our study adds to the understanding of the unique role that internal financial information can play in facilitating efficient business decisions.Specifically, while a number of studies have shown how the external financial reporting quality may affect corporate investment decisions (e.g.Biddle et al., 2009;Biddle & Hilary, 2006;Jung et al., 2014), our study provides new insights into the value of internal information for an important corporate investment, that is human capital investment.
Third, previous research in accounting and finance primarily pays attention to capital investment but overlooks labour investment (Jung et al., 2014).Different from capital investment, as an important factor of production, labour investment typically represents two-thirds of economy-wide added value (Hamermesh & Pfann, 1996) and more than 5 times of the capital investment amongst US firms.Given the economic significance of labour, our study sheds light on how accounting information systems can influence labour investment decisions.Specifically, our paper shows that the negative impact of internal control weaknesses extends beyond capital investment, thus providing important insights regarding the efficient allocation of resources from the perspective of labour investment.
Finally, our study is highly relevant and timely to managers, accounting professionals, policymakers, and the wider capital market participants that have interests in internal information systems and corporate investment decisions.Given the critiques and inconclusive evidence on the value of SOX 404 as part of the regulatory framework, our paper offers a new insight that is highly relevant to policymakers.We directly contribute to the ongoing debate on the costs and benefits of SOX 404 reporting by revealing a significant and yet unexplored economic benefit of effective internal control, namely, efficient investment in human capital.Additionally, our finding also adds to the emerging literature on the potential positive spillover effect of regulation compliance (Cheng et al., 2018;Shroff, 2017) by showing that the process of compliance with regulatory change related to information systems (i.e.SOX 404) can improve the internal information quality and result in better investment decisions.
Furthermore, in light of the economic significance of labour as a factor of production, our study also has important implications for managers and practitioners.Most importantly, our study has profound implications for corporate internal control and efficient allocation of resources.Specifically, our findings suggest that managers and accounting professionals should devote more resources and effort to the timely identification and remediation of internal control weaknesses to improve investment efficiency in labour.
In particular, in the context of the severe disruption caused by the COVID-19 pandemic to both business operations and the way employees work and communicate internal information (e.g.Work from Home and virtual meetings), it is more important than ever that managers should invest in a more robust and reliable internal control system to help businesses to sail through this turbulent period and guide post-pandemic economic recovery.Thus, our study also serves a timely and relevant to the managers and regulators.For example, given the importance of internal control, the managers and regulators may consider appointing dedicated internal audit personnel to strengthen the internal control system.
The remainder of the paper is organized as follows.In Section 2, we review related literature and formulate our main hypothesis.In Section 3, we explain the data collection and empirical design.In Section 4, we discuss our empirical results, followed by a series of robustness tests in Section 5. Section 6 concludes the study.

Literature review and hypothesis development
Prior studies have investigated the implications and consequences of material internal control weaknesses (Altamuro & Beatty, 2010;Ashbaugh-Skaife et al., 2008, 2009;Cheng et al., 2013;Cheng et al., 2018;Costello & Wittenberg-Moerman, 2011;Dhaliwal et al., 2011;Epps & Guthrie, 2010;Feng et al., 2015).Firms with internal control weaknesses generally have been found to have significantly higher idiosyncratic risk, systematic risk, cost of capital and lower financial reporting quality (Ashbaugh-Skaife et al., 2008, 2009;Dhaliwal et al., 2011).Apart from the negative impact of ineffective internal control on corporate financing, some studies also examine the effect on operational decisions.For instance, Feng et al. (2015) consider the influence of ineffective internal control on inventory management and find that ineffective control over inventories leads to suboptimal order quantities, which further causes higher inventory levels and holding costs.Moreover, they find firms with inventory-related material weaknesses in internal control suffer lower inventory turnover and higher inventory impairments.Cheng et al. (2018) find that firms with material weaknesses in internal control have lower operational efficiency relative to firms without such weaknesses.
Building upon the theoretical framework used in the prior literature, we hypothesize that internal control weaknesses can affect labour investment efficiency through (1) poor internal information quality via erroneous internal management reports used for operational decisions; (2) information asymmetry arising from the imperfections in the capital market.Hence, in the following paragraphs, we elaborate on how internal control weaknesses can affect labour investment efficiency before formulating our main hypothesis.
First and most directly, ineffective internal control may affect labour investment efficiency via erroneous internal management reports used for operational decisionmaking (Cheng et al., 2018;Feng et al., 2009Feng et al., , 2015)).Unlike year-end financial statements that are externally audited, the internal management reports are rarely audited or hardly checked by audit committees.Therefore, any errors or incomplete information in the internal management reports due to ineffective internal control are likely to remain undetected in the short term and will directly undermine managerial operation decisions.For instance, flawed IT systems are more susceptible to erroneous records of raw transactions, and incompetent personnel of internal control would fail to provide relevant internal information that is needed for managers' decisions in a timely manner (Feng et al., 2009).Given internal management reports serve as a critical basis to facilitate managers' daily operational decision-making (Feng et al., 2015), internal control effectiveness can influence managers' operation decisions, including labour investment decisions, based on internal management reports.More specifically, Cheng et al. (2018) argue that internal control ineffectiveness can harm a firm's efficiency in operations via its influence on sales forecast reports.For instance, internal control weaknesses over revenue recognition can undermine the accuracy of sales forecasts, which could, in turn, lead to flawed budgeting and suboptimal resource allocation made by managers. 3Over-forecasting of sales can potentially result in over-investment in labour whereas under-forecasting of sales may result in under-investment in labour.In both cases, labour investment inefficiency arises due to the erroneous sales forecast reports caused by ineffective internal control.
Second, based on prior studies on the influence of ICW on operational decisions (Cheng et al., 2018;Feng et al., 2015), we argue that ineffective internal control can also influence labour investment through the misreporting of operating expenses.In particular, if labour cost is mistakenly reported because working hours and workload allocation are not accurately recorded, then it is likely that firms will have excessive or inadequate personnel, resulting in over-investment in labour (e.g.over-hiring) or under-investment in labour (e.g.under-hiring).Having an effective internal control system that generates more accurate and timely data related to labour investment, including workload, payroll, pension, and other fringe benefits, facilitates firms to manage their labour investment more efficiently and enables the proper functioning of human resources.In contrast, when a firm has ineffective control, the operational data used for labour investment will be inaccurate, contributing to suboptimal labour investment decisions.Taken together, when sales forecast and operating expenses are not accurately tracked, larger deviations are likely to occur between the optimal labour levels estimated based on the company's internal management reports and the actual level of labour currently working for the firm.Therefore, having an effective internal control system can also generate more accurate and up-to-date information regarding the day-to-day operations and future prospects of the firm, which enables managers to make corrective adjustments in resource allocation to optimize operations in a timely fashion, should unexpected events occur.
In addition, extant studies have shown that moral hazard and adverse selection are the two primary imperfections in the capital market that lead to firms' departure from the optimal investment level.On the one hand, in the absence of effective internal control and monitoring, managers can more readily engage in opportunistic behavior, which ultimately lowers labour investment efficiency.For instance, inefficient labour investment can happen if managers engage in empire-building and overinvestment in labour by growing their firms beyond the optimal levels (Bertrand & Mullainathan, 2003;Blanchard et al., 1994;Jensen, 1986;Lambert et al., 2007;Richardson, 2006;Stiglitz & Weiss, 1981).On the other hand, due to adverse selection, when facing weak monitoring and internal control, outside capital suppliers are more likely to charge a higher cost of capital for firms with material internal control weaknesses (Ashbaugh-Skaife et al., 2009;Dhaliwal et al., 2011).Hence, given that firms require external capital to finance their labour investment, adverse selection can also cause under-investment in labour, which results in lower labour investment efficiency.Therefore, without effective internal control, managers of firms with internal control weaknesses are potentially more prone to inefficient labour investment.
In sum, we hypothesize that a firm's weak internal control system can cause inefficient labour investment by impairing the operational decision-making and the quality of internal information, which causes firms to engage in the suboptimal level of labour investment.Thus, we develop the following hypothesis: Hypothesis: Internal control weakness is negatively associated with labour investment efficiency.

Sample selection
In line with prior studies, we obtain information on firms' disclosures of their internal control from the AuditAnalytics database.To calculate the predicted value of labour investment from the model of Pinnuck and Lillis (2007) (Model 1), we obtain accounting data from COMPUSTAT and security price and return information from the Center for Research in Security Prices (CRSP).We also further require the availability of the labour investment variables and control variables used in our baseline regression.For control variables, we collect institutional ownership data from Thomson Financial Institutional Holdings (13f) database and obtain the industry-level rate of union membership and coverage data from UNIONSTATS.We exclude firm-year observations that are from the financial and utilities industries.Our final sample contains 3,028 firm-year observations from 2004 to 2016.

Measure of labour investment efficiency
Following prior literature (e.g.Ben-Nasr & Alshwer, 2016;Jung et al., 2014;Khedmati et al., 2020), we measure the expected level predicted by economic fundamentals following the model of Pinnuck and Lillis (2007) (Model 1).Our proxy for labour investment inefficiency, abnormal net hiring, is calculated as the absolute deviation of actual net hiring from its expected level.The higher the value of abnormal net hiring, the lower the labour investment efficiency. 4In the sensitivity test, we also use other alternative proxies for labour investment efficiency to ensure that our findings are robust.
where NET HIRE is the percentage change in employees; SALESGROWTH is the percentage change in revenue; ROA is the return on assets; RETURN is the annual stock return; SIZE is the percentile of firm size measured as the natural log of market value; LIQ is the ratio of cash and short-term investments plus receivables to current liabilities; LEV is the leverage ratio measured as long-term debt plus debt in current liabilities, scaled by total assets; and LOSSBIN1 to LOSSBIN5 are five dummy variables with each 0.005 interval of prior-year profitability from 0 to −0.025.For example, LOSSBIN1 is a dummy variable equal to 1 if the firm's prior-year ROA is between −0.005 and 0, and 0 otherwise.LOSSBIN2 is a dummy variable equal to 1 if the firm's prior-year ROA is between −0.010 and −0.005, and 0 otherwise.In all cases, i indicates the firm and t indicates the year.Following Jung et al. (2014), we winsorize continuous variables at the 1st and 99th percentiles to lower the impact of outliers.

Empirical models
To examine the relationship between internal control weakness and labour investment efficiency, we develop our baseline regression as follows: As defined in Section 3.2, AB NETHIRE is calculated as the absolute deviation of actual net hiring from a firm's expected level; We set ICW equal to one if internal control weaknesses are reported in year t, and 0 otherwise; In line with previous studies on labour investment efficiency (Ben-Nasr & Alshwer, 2016;Jung et al., 2014;Khedmati et al., 2020), we also incorporate a group of explanatory variables that are likely to be associated with corporate labour investment efficiency, including market-to-book ratio, firm size, liquidity, leverage, dividend payouts, tangibility, loss dummy variables, labour intensity, the volatilities of cash flow, sales revenue and net hiring, unionization, abnormal non-labour investments, and institutional ownership respectively.We provide detailed definitions of these variables in Appendix A in this paper.We also include industry-by-year fixed effects and cluster standard errors at the firm level.In the robustness section, we also employ alternative fixed effects and cluster standard errors at both firm and year levels.

Descriptive statistics and univariate results
Panel A of Table 1 gives detailed descriptions of variables in our main model.The dependent variable, AB NETHIRE, has a mean of 0.146 and a median of 0.085 with one standard deviation of 0.195.We also decompose abnormal net hiring into two subgroups depending on the sign.Positive AB NETHIRE implies that a firm's observed value for labour investment is greater than the predicted value (i.e.over-investment in labour, OVER LABOR) whereas negative AB NETHIRE implies that observed value for labour investment is less than the predicted value (i.e.under-investment in labour, UNDER LABOR).The variable of interest, ICW, has a mean of 0.110 and a median of 0 with one standard deviation of 0.313.
In Panel B of Table 1, we compare the descriptive statistics of firms that have internal control weaknesses (ICW Firms) with firms without internal control weaknesses (Non-ICW Firms).The results show that firms with internal control weaknesses have a higher mean (median) abnormal net hiring of 0.184 (0.116) than those without internal control weaknesses, of 0.141 (0.083).These differences in the mean and median are statistically significant at the 1% level.
Table 2 provides the Pearson correlation coefficients for variables in our main model.Our results suggest that internal control weakness (ICW) is positively related to abnormal net hiring (AB NETHIRE), suggesting that firms with ineffective internal control are   more likely to have lower labour investment efficiency.The relations among other variables are generally in line with prediction.For example, our results show that firms with more investment opportunities (MTB), higher volatilities for cash flow, sales and net hiring (SD_CFO, SD_SALES, SD_NETHIRE) and higher abnormal non-labour investments (AB_INVEST) tend to suffer lower efficiency of investment in labour.In contrast, firms paying dividends (DIVD) and firms having higher institutional ownership and financial reporting quality are more likely to have higher labour investment efficiency.

The impact of internal control weaknesses on labour investment efficiency
Table 3 reports the main results.Column 1 presents the results of our main model using the absolute value of the residual, AB NETHIRE, as the outcome variable.We find ICW is positive and significantly associated with abnormal net hiring, suggesting that firms with material internal control weaknesses are inclined to suffer low labour investment efficiency.In the next two columns, we re-run our main model by dividing our sample into two subgroups: firms that over-invest in labour (i.e.positive residuals, observed net hiring greater than predicted) and under-invest in labour (i.e.negative residuals, observed net hiring less than predicted).We keep using the absolute values for both dependent variables for ease of interpretation and our results show that a firm with material internal control weaknesses suffers both over-and under-investment problems.In Column 4, we employ the Fama-MacBeth approach to estimate our main model and we find the results are similar to the main results in Column 1.

Severity of material weaknesses: pervasive ICW and core accounts-related ICW
In this section, we further investigate whether the negative effect of ICW on labour investment efficiency shown in our baseline results varies with the severity of material weaknesses.Prior literature (Cheng et al., 2018) shows two types of severe material weaknesses in internal control systems: (1) pervasive weakness and (2) core accountingrelated weakness.Generally, pervasive weaknesses are those pertaining to the security of information technology and the competency of accounting personnel, which can lead to systematic errors that fundamentally undermine the reliability of all accounting information reported both internally to managers and externally to investors and other market participants (Cheng et al., 2018).Therefore, the repercussions of such material weaknesses are considered most damaging and pervasive, casting doubt potentially on the entire information system of a particular firm.Another type of severe material weakness is ICW affecting individual core accounts.These core account weaknesses are identified within specific accounts that report vital information (e.g. sales revenue, inventory) that managers rely heavily upon for operational decisions, in contrast with weaknesses affecting other accounts that are not as relevant. 5These two types of severe ICW are likely to harm labour investment efficiency to a greater extent, relative to other weaknesses.Similar to Cheng et al. (2018), we classify internal control weaknesses into pervasive weaknesses (PERASIVE_ICW), and core account-related weaknesses (CORE_ICW), based on the reason keys recorded in the AuditAnalytics database. 6In addition, we construct an indicator variable BOTH_ICW equal to one if there is at least one pervasive weakness and at least one core-account-related weakness for each firm-year observation.To enable comparison of the effect between the severe weaknesses and other weaknesses, we also include an indicator variable for other weaknesses, OTHER_ICW, which is coded as one if neither pervasive nor core account-related weaknesses are reported for the firmyear observation, in the regressions.
Table 4 presents the results.We find consistent evidence that severe material ICW is associated with significantly higher inefficiency in labour investment, as proxied by abnormal net hiring.Specifically, in Column 1, we find that PERVASIVE_ICW is positive and statistically significant at 1%, suggesting that firms with pervasive weaknesses have, on average, poorer efficiency in labour investment.Similarly, CORE_ICW in Column 2 is positively associated with abnormal net hiring, showing that firms with weaknesses related to core accounts are also subject to lower labour investment efficiency.It is also important to note that the effect of PERVASIVE_ICW is stronger than CORE_ICW in terms of both statistical significance and magnitude, which is consistent This table reports the results of regressions of abnormal net hiring (AB_NETHIRE) on internal control weakness (ICW) based on the severity of material weaknesses.Column 1 shows the results regressing abnormal net hiring on internal control weaknesses that have a more pervasive effect on internal reporting (PERVASIVE_ICW).Column 2 shows the results regressing abnormal net hiring on internal control weaknesses related to core accounts that are more likely to lead to errors in the internal reports that managers rely on to make operational decisions (CORE_ICW).Column 3 shows the results regressing abnormal net hiring on internal control weakness that at least contains one pervasive weakness and one core accounts weakness (BOTH_ICW).OTHER_ICW is the internal control weaknesses that are neither pervasive weaknesses nor core accounts weaknesses.*, **, *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.All variables are defined in Appendix A.
with Cheng et al. (2018) in that pervasive weaknesses can have a more profound effect on firms than core-accounts-related weaknesses.Finally, in Column 3, we repeat the analysis using BOTH_ICW to consider cases where both types of severe weakness are identified within a firm for a particular year and our results remain robust to this additional specification.
Notably, the positive and significant effect of severe ICW is in contrast with the insignificant, albeit positive, coefficient for OTHER_ICW in all three models, suggesting that our main results are predominantly driven by the more severe weaknesses in internal control systems.Overall, in line with our prediction that severe internal control weaknesses are more detrimental to firms, we document consistent evidence that firms with severe weaknesses have significantly lower labour investment efficiency.

Remediation of internal control material weaknesses
So far, our results show that firms with material internal control weaknesses are likely to have lower labour investment efficiency.We then follow the specification employed in prior studies (Cheng et al., 2018;Feng et al., 2015) to conduct a change analysis to investigate the influence of remediation of material weaknesses on the change in labour investment efficiency.If lower labour investment efficiency is a result of internal control weaknesses, the remediation of internal control material weaknesses is expected to increase the efficiency of investment in labour.The change analysis for the remediation test can not only corroborate the main results but also uses the same firm as its own control and thus alleviates the omitted correlated variable concern by controlling for time-invariant firm characteristics (Cheng et al., 2018).To test the influence of remediation of material weakness on labour investment efficiency, we estimate the following change regression: We follow Feng et al. (2015) and Cheng et al. (2018) and measure ΔAB_NETHIRE as the change in labour investment efficiency from year t to year t + 2 because Feng et al. (2015) suggest that using one-year changes makes it unclear when in year t + 1 the material weaknesses were remediated.REMEDIATION is set to one if material weaknesses are disclosed in year t but no material weaknesses in year t + 1 (i.e.material weaknesses have been remediated).All the control variables are also the changes from year t to t + 2. Following Feng et al. (2015), we also include abnormal net hiring from year t−1.
Table 5 reports the results of the remediation test.We find that the coefficient on REMEDIATION is negative and significant.Thus, the results suggest that the remediation of internal control weaknesses leads to an improvement in labour investment efficiency, which corroborates our main results.

Subsample analysis: role of human-capital-intensive firms
In this section, we conduct a subsample analysis to examine whether the relation between internal control weakness and labour investment efficiency varies with firms' reliance on high-skilled labour.Prior literature has established that employees can play an important role in the corporate governance mechanism through intense scrutiny of managerial behavior and corporate decisions (Chyz et al., 2013;Faleye et al., 2006;Fauver & Fuerst, 2006;Huang et al., 2017;Lin et al., 2018).As internal stakeholders directly participating in firms' day-to-day operations, employees have a long-term financial claim in the form of salaries and pensions (Campello et al., 2018).Therefore, they have a strong incentive to closely monitor the management (Huang et al., 2017) and the financial transparency of their employers (Hamm et al., 2018;Siu et al., 2009).
Relative to low-skilled employees, several studies have shown that high-skilled employees are more effective in deterring managerial misbehaviour due to stronger participation in corporate decisions (Aguilera & Jackson, 2003;Kim et al., 2018) and greater financial incentives (Hochberg & Lindsey, 2010;Kroumova & Sesil, 2006).Hence, given high-skilled employees' greater information demand and better educational background than low-skilled counterparts, we argue that high-skilled employees are more motivated and equipped to identify potential irregularities and ensure the reliability of information reporting systems.Furthermore, Goh (2009) shows that non-accounting financial expertise and education are relevant attributes to facilitate timely remediation of material weakness, thus effectively suppressing the adverse effect of internal control weaknesses on firms.Hence, we predict that, in the event of material weaknesses, high-skilled employees, who possess higher levels of education and expertise, as well as financial interests in the firms, would contribute both directly and indirectly to faster remediation of material internal control weaknesses, relative to low-skilled employees.Therefore, the negative effect of ICW on labour investment decisions is expected to be at least partially mitigated by more timely remediation and intense monitoring within in human-capitalintensive firms whose workforce are primarily high-skilled employees.In other words, we argue that impact of ICW on labour investment efficiency should be more pronounced in low-skilled firms, where employee monitoring is perceived to be weaker.
However, we concur that it is also possible that the negative influence of ICW on labour investment efficiency could be exacerbated in human-capital-intensive firms.Prior studies point out that labour adjustment costs are significantly higher for human-capital-intensive firms due to higher costs involved in recruitment, retention and training (Cao & Rees, 2020;Dixit, 1997;Ghaly et al., 2017;Ghaly et al., 2020).Therefore, firms that are highly dependent on high-skilled labour would require a significant amount of capital to finance their investment in human capital (Campello et al., 2010;Michaels et al., 2019).However, several studies suggest that firms with internal control weaknesses are penalized by both investors and lenders and thus face higher costs of capital, due to perceived poor financial reporting quality and higher information asymmetry (Dhaliwal et al., 2011;Kim et al., 2011;Ogneva et al., 2007).In addition, Jung et al. (2014) find that firms with better financial reporting quality enjoy higher labour investment efficiency.Therefore, faced with higher labour adjustment costs, human-capitalintensive firms with material internal control weaknesses are likely to suffer more as a result of rising financing costs, which would significantly impede them from investing in their human capital more efficiently.Overall, the above arguments point to two competing predictions on how human capital intensity might alter the relation.
Table 6 presents the results of our subsample analyses on firms' reliance on human capital.In Columns 1 and 2, we divide our sample based on the median of human capital reliance measured as the ratio of R&D expenditure to total sales.Thus, firms are considered human-capital-intensive if they have an above-median ratio, whereas those with a below-median ratio are categorized as firms with lower reliance on human capital.As we can see from the first two columns, there is a positive and significant relationship between ICW and abnormal net hiring in the low human-capital- intensity group (0.0572) in Column 2, where employees tend to play a much weaker role in monitoring.In contrast, such an influence is effectively mitigated in human-capitalintensive firms, as evidenced by the insignificant and much smaller coefficient (0.0016) in Column 1.We find that the coefficients of ICW are significantly different at the 5% level between the two subsamples.Furthermore, to ensure that our findings are not driven by a particular measure of human capital intensity, we use another four alternative proxies to check robustness.First, we partition our sample into two subgroups, based on whether a firm operates in an industry with high reliance on human capital.Consistent with previous studies (Ben-Nasr & Ghouma, 2018;Cao & Rees, 2020;Ghaly et al., 2015), we consider firms belonging to healthcare, high-tech, and telecommunications industries7 as humancapital-intensive firms.As illustrated in Columns 3 and 4, the coefficient of ICW is positive and significant only in low-skilled industries (Column 4), in contrast with the insignificant result in human-capital-intensive industries (Column 3).We also test the coefficient difference and find that the coefficient of ICW is statistically different at the 1% level across the two subgroups.
Second, Eisfeldt and Papanikolaou (2013) suggest that firms with high levels of organizational capital tend to be more human-capital-intensive and invest heavily in key talents.Following prior literature (Chen et al., 2016;Eisfeldt & Papanikolaou, 2013), organization capital is defined as capitalized selling, general and administrative (SG&A) expenses scaled by total assets.We thus categorize firms into high-organization-capital firms and low-organization-capital firms based on the sample median of organization capital.Similarly, we find that ICW remains positively significant in firms with low organization capital (Column 6) whereas its coefficient becomes insignificant in high-organization-capital firms.
Third, we partition our sample based on the knowledge capital of a firm, as defined by Peters and Taylor (2017).According to results in Columns 7 and 8, we find that the negative impact of ICW on labour investment efficiency is indeed more pronounced in firms with below-median knowledge capital (Column 8).Finally, in Columns 9 and 10, we use the industry-specific labour skill index developed by Ghaly et al. (2017) to proxy for a firm's reliance on high-skilled labour and document consistent evidence that the effect of ICW on abnormal net hiring is significant only in low-skilled industries (Column 10).
Overall, using various proxies for a firm's reliance on human capital, our subsample analyses present robust evidence that the negative effect of internal control weakness is attenuated in firms with high reliance on human capital.In comparison, the negative impact of ICW on labour investment efficiency is concentrated in firms in lowskilled industries. 8Taken together, the above findings are consistent with the view that high-skilled employees play a stronger monitoring role, which significantly mitigates the negative impact of internal control weaknesses on the efficiency of labour investment.

Robustness tests
To ensure the robustness of our main result, we conducted a series of additional tests.We further employ the PSM procedure by matching control firms with firms reporting material internal control weaknesses based on the full set of explanatory variables in our main model (Shipman et al., 2016).Our PSM results support our main results, suggesting ineffective internal control weakness is associated with labour investment inefficiency.Second, to corroborate our results based on propensity-score-matched samples and ensure our results are not sensitive to the choice of a particular matching algorithm, in this section, we resort to entropy balancing 9 (Chahine et al., 2020;Hainmueller, 2012;Kim & Valentine, 2021;King & Nielsen, 2019;Wilde, 2017) and find consistent results.Third, to further alleviate the concern regarding whether labour investment is merely a reflection of corporate non-labour investments 10 , we also follow prior studies (Ben-Nasr & Alshwer, 2016;Jung et al., 2014) and empirically demonstrate how our study is distinct from previous studies developed in the context of investments and that our results are not solely driven by non-labour investments.Our results suggest that labour investment is not merely a byproduct of other nonlabour investments and the influence of internal control weaknesses on labour investment efficiency we documented earlier is not driven by non-labour investments.Fourth, our results are also robust to various alternative measures of labour investment efficiency and model specifications.Finally, to eliminate the potential confounding effect due to the financial crisis, we repeat our main analysis by removing the observations during the financial crisis period of 2007-2009 and we find our results are similar.Full details of the robustness tests in Section 5 can be found in the Online Appendices.

Conclusion
In this study, we investigate the impact of material internal control weaknesses on firms' investment efficiency in one of the important production factors, namely human capital.Using a sample of US firms from 2004 to 2016, we document robust evidence that firms with internal control weaknesses are associated with inefficient employment decisions, arising from both over-investment and under-investment in labour.Further analyses reveal that the detrimental effect of internal control weakness is primarily driven by the most severe types of material internal control weakness, namely pervasive weakness and core-accounts-related weakness.To further support our main finding, our change analysis confirms that the remediation of material weaknesses significantly improves labour investment efficiency.Moreover, our subsample analyses present consistent evidence that the negative impact of internal control weakness is mitigated in humancapital-intensive firms where the monitoring effect of employees is perceived to be stronger.Finally, our results are robust to a PSM approach, an entropy balancing approach, controlling for the role of non-labour investment and financial crisis, as well as various alternative labour investment efficiency measures and fixed effects specifications.
Our study contributes to the literature on several fronts.First, by investigating the role of effective internal control in labour investment efficiency, our study directly contributes 9 We thank the anonymous reviewer for suggesting this test. 10We thank the reviewer for raising this issue that inspires us to empirically test the role of non-labor investment.
to the contentious debate on the net benefits of SOX 404.In particular, while several previous studies have examined the adverse consequences of internal control weaknesses (Altamuro & Beatty, 2010;Ashbaugh-Skaife et al., 2008;Doyle et al., 2007aDoyle et al., , 2007b;;Feng et al., 2009), our study highlights the value of effective internal control by investigating the impact of material internal control weaknesses on investment efficiency in human capital, as a valuable intangible asset.
Second, prior empirical studies in accounting and finance fields primarily focus on capital investment but neglect labour investment (Biddle et al., 2009;Biddle & Hilary, 2006;Jung et al., 2014).Considering the economic significance of labour, our study provides empirical evidence that suggests how accounting information systems shape corporate labour investment decisions.By highlighting that the negative influence of internal control weaknesses extends beyond capital investment, our study offers important insights into the efficient allocation of resources from the perspective of labour investment.
Finally, in light of the economic significance of labour as a factor of production, our study has multiple important and practical implications.Our results suggest that managers and accounting professionals should devote more resources to the timely identification and remediation of internal control weaknesses to improve corporate efficiency.In addition, our study also adds to the ongoing debate on the costs and benefits of SOX 404 reporting from the human capital investment perspective.Overall, our study has timely implications for managers, accounting professionals, policymakers, and the wider capital market participants that have interests in accounting information systems and corporate investment decisions.Particularly, in the context of the severe disruption caused by the COVID-19 pandemic to both the actual business operations and internal information communication within the firms, our study suggests that managers should dedicate more resources to ensure an effective internal control system, which would generate valuable internal information to inform the post-pandemic economic recovery.

Table 1 .
Descriptive statistics of variables in the baseline regression.Table 1 presents the descriptive statistics of the main variables.All variables are defined in Appendix A. Panel B of Table 1 compares the average values of the firm characteristics for firms with ineffective internal control systems (ICW firms) and firms without material weaknesses (Non-ICW firms).The significance of the differences in means and medians between ICW firms and Non-ICW firms.All variables are defined in Appendix A.

Table 3 .
Internal control weakness and abnormal net hiring.
This table reports the results of regressions of abnormal net hiring (AB_NETHIRE) on internal control weakness (ICW).Column 1 shows the results of regressing abnormal net hiring on internal control weakness and control variables.Column 2 shows the results regressing labour over-investment (OVER_LABOR) on internal control weakness and control variables.Column 3 shows the results regressing labour under-investment (UNDER_LABOR) on internal control weakness and control variables.Column 4 presents the results of a Fama-MacBeth regression.*, **, *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.All variables are defined in Appendix A.

Table 4 .
The severity of material weaknesses: Internal control weakness and abnormal net hiring.

Table 5 .
Change analysis: The effect of remediation of material weaknesses.This table reports the results from regressing the change of abnormal net hiring (AB_NETHIRE) on the remediation of material weakness and the change in the determinants of labour investment efficiency.

Table 6 .
Human capital intensity, organization capital, knowledge capital and labour skills.reports the results of OLS regressions for subsamples based on human capital intensity, organization capital, knowledge capital, and labour skills.The robust standard errors clustered at the firm level are reported in parentheses.*, **, *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.All variables are defined in Appendix A.