We investigate whether the presence of municipal audit committees is associated with internal control quality in the municipal setting. The evidence shows that only 20 percent of municipalities in our sample voluntarily maintained an audit committee during the sample period of 2001 through 2004. Our results highlight that municipalities with audit committees are associated with fewer internal control problems, which in turn suggests these cities should be less likely to experience future significant financial reporting failures. These results persist in specifications that use econometric procedures to address concerns about self-selection. Overall, our findings suggest that audit committee presence plays an important role in municipal financial oversight.
Audit committee oversight is often considered to be an important component of financial policies and decisions (e.g. Klein 2002; Agrawal and Chadha 2005; Armstrong, Guay, and Weber 2010). This is particularly germane in the government sector, where expenditures are large, agency problems are common (Gordon 2009), and voter oversight is often scarce (Zimmerman 1977). Since 1997, the Government Finance Officers Association (GFOA) has recommended that municipalities voluntarily form audit committees to enhance the credibility of government financial reporting (GFOA 2008).1 Despite the GFOA recommendations, few studies consider how audit committees influence governmental financial reporting.2 The purpose of this study is to investigate whether the presence and composition of municipal audit committees is associated with high quality internal control systems.
Our survey results highlight that 20 percent of sample observations maintained an audit committee during the sample period of 2001–2004. Of the audit committees in place, only 45 percent were fully independent, which contrasts with results from the corporate sector where regulation requires virtually all firms listed in the S&P 500 to maintain an audit committee (Klein 1998) consisting entirely of outside directors (Blue Ribbon Committee 1999). Furthermore, only 57 percent of audit committees had at least one member with financial expertise, which lags behind the roughly 80 percent level in the corporate sector (Zhang, J. Zhou, and N. Zhou 2007).
In our empirical analysis, we find that municipalities with audit committees are less likely to have internal control problems after controlling for other governance characteristics shown by prior literature to influence financial reporting. These results persist in specifications that use econometric procedures to address concerns about self-selection. One interpretation of our evidence is that localities with audit committees have stronger financial reporting systems than localities without audit committees, which in turn deters future significant financial reporting failures.
Given our findings of the significance of municipal audit committees, coupled with recent interest in governmental audit committee requirements (GFOA 2008), we also perform supplemental analysis to explore the impact of audit committee composition. Our results suggest that a completely independent audit committee, defined as one with no members that are full-time employees of the municipality, does not elicit improvements in terms of internal control weaknesses over those found for committees with at least one insider. This result supports assertions in Klein (1998) that insiders provide valuable information. We find similar results regarding audit committee financial expertise. More specifically, municipalities with an audit committee having at least one financial expert, defined as an individual with specific knowledge regarding governmental financial reporting, do not show stronger associations between audit committees and internal control quality beyond what is found for audit committees without a financial expert. However, we do find evidence that audit committees containing an insider with financial expertise are less likely to have internal control problems than audit committees with insiders lacking financial expertise, possibly suggesting that access to a financial expert is a motivation for maintaining a non-independent audit committee.
The study contributes to extant literature by providing preliminary evidence on the role of audit committee oversight in the municipal context. Few studies explore the role of boards in non-corporate settings, perhaps due to difficulty obtaining data (Adams, Hermalin, and Weisbach 2010). Such research is important for understanding the dynamics of municipal audit committees in a setting where there are currently no regulations requiring its presence, and where maximizing firm value is not a primary consideration.
Our findings have several implications for governance research in general. First, we highlight the importance of audit committees in monitoring financial reporting in the governmental setting where members of a board of city commissioners or city council may lack the expertise necessary to effectively manage a government's finances. Moreover, our findings that the audit committee independence and financial expertise do not influence associations between audit committees and internal control weaknesses, while non-independent audit committees with financial expertise relate to fewer internal control problems, has implications for regulators and others currently contemplating the importance of audit committee composition in the governmental sector.
Our findings also have broader implications for public policy. Municipal financial information is used to make important resource allocation, taxing, and financing decisions, and to set government fiscal policies that significantly impact the economy (e.g., with expenditures comprising 16 percent of GDP at all levels of government in 2012) (World Bank 2012). Evaluating the extent to which audit committees are associated with municipal internal control policies is potentially helpful for evaluating the conditions under which municipal officials more aptly fulfill their stewardship responsibilities, one of the primary roles of governmental accounting regulations (Government Accounting Standards Board [GASB] 2006).
The remainder of the paper proceeds as follows. We present background information on municipal audit committees in the following section. Empirical results for internal control quality are presented in the third and fourth sections, with concluding remarks in the fifth.
BACKGROUND ON MUNICIPAL AUDIT COMMITTEES
The government is an important part of our economy. In 2011, the federal government provided $607 billion in grants to fund state and local governments' programs on health care, income security, education, and transportation. These funds account for 17 percent of federal outlays and finance about 25 percent of state and local government spending (Congressional Budget Office [CBO] 2013).
In 1997, the GFOA, a professional association that promotes financial management of state and local governments, recommended that governmental units establish audit committees or the equivalent (GFOA 2008). In 2002, 2006, and 2008, the GFOA refined these recommendations to be more consistent with corporate sector requirements in terms of audit committee size, independence, and financial knowledge (GFOA 2008). Although the GFOA continues to recommend that municipalities establish audit committees, formation continues to be voluntary. Typically stated functions of government audit committees include overseeing the financial reporting process, monitoring internal controls, and interacting with independent auditors (GFOA 2008).
Using survey data from federal, state, and local governments, David (2008) finds that many audit committees play a role in fraud prevention and risk management practices, further supporting the assertion that audit committees are determinants of municipal internal control quality. However, Helland and Sykuta (2004) discuss that some directors serve as window dressing in the corporate context, whereby their sole function is to avoid interfering with management. It is similarly plausible that, for some municipalities, audit committees provide political cover, such as to increase officials' election probabilities, appease voters, or further unelected bureaucrats' careers without providing meaningful oversight (Carpenter 1991; Carpenter and Feroz 1990). In the end, the relation between municipal audit committees and internal control quality is a matter to be resolved empirically.
MUNICIPAL INTERNAL CONTROL QUALITY
Prior literature establishes that high-quality accounting systems are important mechanisms in the functioning of efficient capital markets, in part because they reduce information asymmetry between firms and investors (Merton 1987; Healy and Palepu 1993; Diamond and Verrecchia 1991). The importance of monitoring mechanisms over financial reporting is well documented empirically in the private sector context (Dechow, Sloan, and Sweeney 1996; McMullen and Raghunandan 1996; Shleifer and Vishny 1997). In many of these studies, audit committees are presumed to be a particularly important part of the oversight of the financial reporting process. Prior corporate research finds that audit committee existence is inversely associated with fraud (Dechow et al. 1996) and positively associated with financial reporting quality (McMullen 1996; Wild 1996). These findings are diminished to an extent by regulations that require audit committee presence and dictate audit committee composition.
Despite prior studies in the corporate sector, relatively little empirical research considers how audit committees influence nonprofit or governmental financial reporting. One exception is Pridgen and Wang (2012), who analyze a sample of nonprofit hospitals subject to the Single Audit Act from 2001 to 2008. The authors find mixed results for the effectiveness of audit committees. In addition, a recent working paper by Elder, Feng, and Neely (2014) finds that formation of nonprofit audit committees relates to higher reported administrative expenses, suggesting a financial reporting impact. However, given the political and regulatory characteristics of governmental organizations that distinguish municipalities from both the corporate and other nonprofit enterprises (Vermeer, Raghunandan, and Forgione 2006), the consequences of municipal audit committees are not obvious ex ante. Therefore, whether and how internal control quality relates to audit committee existence in a largely unregulated setting is an empirical question.
Internal control systems are important elements of financial reporting quality in part because they predict (or prevent) future significant accounting and financial reporting problems. Prior literature generally suggests that corporate audit committee oversight influences the strength of internal control systems, where audit committee independence or financial expertise is associated with fewer internal control weaknesses (Krishnan 2005; U. Hoitash, R. Hoitash, and Bedard 2009).
The relatively few studies in the nonprofit and government sector that examine the determinants of internal control problems show that nonprofit entities are more likely to experience internal control problems when they are smaller, audited by less experienced auditors, financially distressed, growing, more complex, or the subject of prior audit findings (Keating, Fischer, Gordon, and Greenlee 2005; Petrovits, Shakespeare, and Shih 2011). In the government sector, municipalities with internal control problems are shown to have higher audit fees (Raman and Wilson 1992), non-staggered council elections (Peterson 2014), and audits conducted by CPA firms as opposed to the state government (Lopez and Peters 2010). Furthermore, Jakubowski (1995) finds that more control problems are reported in counties than in cities, and the frequency of control weaknesses reported declined significantly over the first four reporting years under the Single Audit Act for city governments, but remained relatively constant for county governments.
If municipal audit committees similarly provide enhanced monitoring over the internal control environment, then we expect an inverse association between their presence and internal control weaknesses. As a result, we present the following hypothesis (stated in null form).
Internal control weaknesses are independent of whether municipalities have audit committees.
We propose the following ordered logit specification to investigate this relation:
where ICW INDEX identifies internal control system deficiencies; AUDIT COMMITTEE is an indicator variable that is coded as 1 when an audit committee is in place; GOVERNANCE is a series of municipal governance provisions; and CHARACTERISTICS represents an array of other municipal-specific characteristics that potentially impact the financial reporting environment. Following Lopez and Peters (2010), we construct ICW INDEX in a way that captures the severity of internal control weaknesses, where municipalities with at least one material weakness during our sample period of 2001 through 2004 are coded as 2, municipalities with at least one reportable condition (but no material weaknesses) are coded as 1, and municipalities with neither an internal control weakness nor a reportable condition are coded as 0.3 The analysis includes only one observation per municipality because internal control systems tend not to fluctuate, such that the dependent variable varies little from year to year.4
Admittedly, audit committees are only one of an array of governance attributes that potentially influence the financial reporting and internal control environments of local governments. To discern between the effects of oversight by audit committees versus other governance mechanisms, we include several municipal governance attributes that relate to the financial reporting system. Extant literature suggests direct democracy provisions that encourage citizen lawmaking, such as initiative and referendum provisions, are important components of municipal policies (e.g., Gordon 2009; Matsusaka 1995).5 For example, Compton, Gore, and Kulp (2013) find that voter activism influences financial policies such as severance payments to bureaucrats. From a financial reporting perspective, Baber et al. (2013) find that voter oversight mechanisms are associated with fewer accounting restatements. Following Baber et al. (2013) and Zhang (2012), we set the indicator variable BOTH PROVISIONS equal to 1 if the municipality has both initiative and referendum provisions in place, and NEITHER PROVISION equal to 1 if the municipality has neither of these provisions. Both indicator variables are set to 0 for municipalities with either, but not both, provisions. Additionally, RECALL ATTEMPT, which measures citizen activism, is set to 1 if citizens attempted to recall either the mayor or a council member in the preceding five years.
Evans and Patton (1983) highlight that the council-manager form of government potentially influences financial reporting decisions. Thus, we include COUNCIL MANAGER FORM as an indicator variable equal to 1 for municipalities where the city manager is the chief executive, and 0 for the strong mayor form of government. Next, given that Baber et al. (2013) find a negative association between auditor independence and municipal restatements, we distinguish municipalities audited by external audit firms (INDEPENDENT AUDITOR) as opposed to state auditors based on links between restatements and internal control (Rice and Weber 2012).
Lopez and Peters (2010) use a sample of city and county governments to highlight that municipalities employing Big 4 audit firms are more likely to disclose internal control weaknesses. Therefore, we set an indicator variable (BIG 4 AUDITOR) to 1 if the municipality engages a Big 4 auditor for more than half of the sample period (0 otherwise). We also consider the potential impact from auditor turnover, setting AUDITOR SWITCH equal to 1 if municipality switched auditors during the sample period. Next, given that auditors can classify certain audits as low risk as part of the A-133 audit process (thereby decreasing the percentage of program expenditures that need to be audited), we include a separate variable titled LOW RISK AUDIT to consider audits deemed to require less audit effort.
Prior literature also documents links between board size and performance (Yermack 1996; Coles, Daniel, and Naveen 2008), so we include the log of COUNCIL SIZE. Furthermore, results from Baber et al. (2013) show that municipalities with staggered council elections are more likely to experience a greater likelihood of restatements, while Peterson (2014) finds evidence that staggered council elections are associated with fewer internal control weaknesses. As a result, we include STAGGERED COUNCIL as an indicator variable equal to 1 for municipalities with staggered council election cycles.
Following Baber and Gore (2008), we consider the potential impact of state disclosure regulations on internal control quality by using indicator variables equal to 1 for municipalities located in states that require compliance with GAAP accounting standards (GAAP), and those with no municipal reporting requirements (UNREGULATED). Both variables are set to 0 for municipalities in states with hybrid or state-specific municipal accounting requirements. The influence of state monitoring in general is considered by including the amount of state support to localities (STATE LEVEL SUPPORT), defined as the average ratio of state revenue to total revenue for all municipalities in a given state. Finally, proxies for municipal characteristics include DEBT PER CAPITA as the total log of debt per capita, and municipal SIZE as the log of population. Values for independent continuous variables are computed using 2001 to 2004 means for each municipality, and are winsorized at the 1 percent (99 percent) tails of the distribution.
The population is 4,244 municipalities that respond to the 2001 Municipal Form of Government Survey conducted by the International City/County Management Association (ICMA). For comparability to prior research, we draw our sample from the 365 responding municipalities with population greater than 50,000 citizens (Copley, J. Gaver, and K. Gaver 1995; Baber et al. 2013). We obtain audit committee details during the sample period of 2001 through 2004 via a mailed survey and follow-up procedures (see Appendix A).6 After deleting municipalities with missing financial, governance, and audit committee data, the final sample consists of 240 unique municipalities (66 percent). Table 1 outlines the sample selection procedures.
Most other governance data are from the ICMA survey, while auditor details are from the Single Audit Database provided by the Federal Audit Clearinghouse (https://harvester.census.gov/facweb/default.aspx). Financial reporting data are from the U.S. Census Bureau's State and Local Government Finance Database and the U.S. Census Bureau's 2000 County and City Data Book, as well as Comprehensive Annual Financial Reports (CAFRs) obtained directly from municipalities. Classification of state reporting requirements for municipalities are from Baber and Gore (2008).
Table 2 provides summary statistics for variables classified by whether municipalities have an audit committee in place, along with tests of differences between the two subsamples. The univariate results suggest that internal control deficiencies (ICW INDEX) are less likely in municipalities with audit committees compared to those without audit committees, and that the difference between the subsamples is greater for material weaknesses (MATERIAL WEAKNESS). Furthermore, audit committees are more frequent in municipalities that engage a high-quality auditor (BIG 4 AUDITOR), suggesting a complementary relationship between the two governance mechanisms. On the other hand, audit committees are less prevalent in GAAP states (GAAP), and more prevalent in states without disclosure regulations (UNREGULATED), suggesting that state accounting regulations and audit committees are substitutes.
The absence of voter ability to initiate changes in municipal policies (NEITHER PROVISION) varies inversely with audit committee presence. Such evidence suggests that audit committee formation is less likely in municipalities lacking direct democracy provisions, which is considered by some to be correlated with management entrenchment (Bebchuk, Cohen, and Ferrell 2009). Finally, audit committees are more likely in large municipalities (SIZE) implying a relationship between complexity and committee formation. Audit committee presence does not vary with the amount of debt outstanding, the presence of an independent auditor, auditor turnover, staggered council elections, or the form of government.
Propensity Score Matching
One potential issue with our design involves the fact that audit committee formation in the municipal sector is voluntary, which creates the risk that selection bias influences associations. We therefore use propensity score matching to align each municipality employing an audit committee with a control municipality that does not have an audit committee (Rosenbaum and Rubin 1983). Following Lawrence, Minutti-Meza, and Zhang (2011), we estimate a propensity score using variables from a selection model, along with all control variables employed in the multivariate analysis.
Under this approach, we incorporate the following determinants model of audit committee presence:
where AUDIT COMMITTEE is an indicator variable that is coded as 1 when an audit committee is in place; GOVERNANCE and CHARACTERISTICS are as defined previously, and OTHER represents other municipal characteristics proposed to be associated with the decision to form an audit committee. Variables included in OTHER are: PROFESSIONALS ON COUNCIL is the number of professionals (attorneys, business executives, etc.) that serve on the city council, which considers the impact of outside professional expertise. PAST FISCAL DISTRESS is an indicator set to 1 if the municipality experienced two or more operating deficits in the prior five years. If audit committees foster sound financial decisions, then cities with longstanding financial problems should be associated with the absence of audit committees. However, cities with problems may have recently adopted audit committees, so interpreting past fiscal distress is not straightforward. CORRUPTION follows Butler, Fauver, and Mortal (2009), and uses Department of Justice statistics to compile a state-level corruption measure based on the number of corruption convictions of local, state, and federal officials. If audit committees limit the ability of municipal managers to extract rents from citizens, then one consequence could be an inverse association with audit committee formation. Finally, OWN SOURCE REVENUE is a measure of internally generated funds, primarily coming from taxes and user fees (Berry and Gersen 2009). Given that municipalities can finance government operations in many ways—via taxes, new debt, or transfers from higher levels of government (Matsusaka 1995)—different sources of revenue may lead to differences in demand for additional financial oversight.
Results for the Model 2 specification of audit committee probability are in Table 3, which shows a pseudo R2 of 0.24. We make no predictions about the coefficient signs, as evaluating the determinants of municipal audit committee formation is beyond the scope of this paper. We find audit committees are more likely in states that provide high levels of funding to municipalities (STATE LEVEL SUPPORT) and that have no municipal disclosure requirements (UNREGULATED), but are less likely in states with formal GAAP requirements (GAAP) and high levels of corruption (CORRUPTION). These results suggest that state-level monitoring and the political environment impact the decision to form an audit committee.
The results in Table 3 also indicate that audit committees are less likely in jurisdictions with no citizen participation provisions (NEITHER PROVISION) and poor recent operating performance (PAST FISCAL DISTRESS), but more likely when municipalities utilize a Big 4 auditor (BIG 4 AUDITOR), and finance more of their operations via internally generated funds (OWN SOURCE REVENUE). These results provide some evidence that audit committees complement strong audit and operating environments.
We then match (without replacement) each municipality with an audit committee with the municipality not having an audit committee that has the closest propensity score, with a maximum distance of 3 percent (Lawrence et al. 2011). We can identify a sufficient match for 72 percent of the sample municipalities. Untabulated univariate results for the propensity score matched sample highlight that internal control deficiencies are less likely in municipalities with audit committees compared to those without audit committees (p-value < 0.01), but we find no statistically significant differences between the audit committee and matched samples for other independent variables.
Table 4 displays ordered logit specifications of the ICW INDEX capturing the severity of internal control weaknesses using Equation (1). Column  shows that the presence of an audit committee is associated with significantly fewer internal control deficiencies for the base model (z-statistic = −2.49).7 The odds ratio of 0.22 for this specification suggests that municipalities with audit committees are about one-fourth as likely to move up one category in terms of internal control weakness severity as those with no audit committee. Column  shows that results are robust to the inclusion of state fixed effects in lieu of state-level controls, as highlighted by the significantly negative coefficient on AUDIT COMMITTEE (z-statistic = −3.13). Column  presents results using our propensity score matched sample (Lawrence et al. 2011).8 The Column  evidence highlights that our primary findings are not driven by the choice to voluntarily adopt an audit committee, as highlighted by the negative coefficient on AUDIT COMMITTEE (z-statistic = −2.35).9
With respect to other governance attributes, we find that municipalities with greater oversight are less likely to experience internal control deficiencies. Specifically, we find that municipalities that employ the council manager form of government (COUNCIL MANAGER FORM) are less likely to disclose internal control weaknesses, which is consistent with prior literature (Evans and Patton 1983). In addition, we find municipalities designated as a LOW RISK AUDIT are also less likely to have internal control deficiencies. Finally, we find some evidence that the presence of Big 4 auditors (BIG 4 AUDITOR), municipalities that experience auditor turnover during the sample period (AUDITOR SWITCH), and staggered council elections (STAGGERED COUNCIL) are associated with fewer internal control weaknesses, and limited evidence that municipalities with independent auditors (INDEPENDENT AUDITOR) are generally associated with internal control weaknesses.
ADDITIONAL ANALYSIS ON AUDIT COMMITTEE COMPOSITION
An extensive literature studies the causes and consequences of audit committee composition in the corporate sector. Even so, empirical findings are mixed regarding the effectiveness of audit committee characteristics for constraining internal control problems. Krishnan (2005) shows that firms with independent audit committees and audit committees with financial expertise are less likely to report internal control weaknesses, and Goh (2009) finds a positive association between more timely remediation of internal control weaknesses and audit committee independence and financial expertise. Such results suggest that audit committee independence and/or audit committees with a member possessing financial expertise serve as a more effective monitor of the financial reporting system. However, Zhang et al. (2007) find no association between auditor independence and internal control weaknesses using a sample of firms disclosing material internal control weaknesses matched with firms disclosing no internal control problems. Nonetheless, the authors find that firms are more likely to have internal control weaknesses if their audit committees have less financial expertise.
Few papers in the nonprofit sector examine the determinants and consequences of independent audit committees. One example is Vermeer et al. (2006), who find that size, government grants, and engaging a Big 4 auditor are associated with the likelihood of having audit committees with solely independent directors. In addition, their results show that universities and hospitals are less likely to have audit committees with solely independent directors. Moreover, Vermeer et al. (2006) find that most of the nonprofits in their sample have at least one financial expert on the audit committee, and that organizations that receive government grants and have an internal audit function are more likely to have a financial expert on the committee. Callen, Klein, and Tinkelman (2003) find that major donors play an effective role in nonprofit efficiency. Specifically, they find that the ratio of total expenses to program expenses varies inversely with donor representation on board. However, there is currently a lack of evidence on whether audit committee independence/financial expertise impacts financial reporting in the municipal context.10
Given the recent emphasis by the GFOA regarding the importance of corporate audit committee characteristics, combined with inconclusive corporate sector findings, it is instructive to consider the impact of audit committee characteristics on our results. We use the following ordered logit specifications to investigate the impact of municipal audit committee financial expertise and independence on the incidence of internal control weaknesses:
In model (3) we partition the audit committee variable using two indicators—the first set to 1 if a “financial expert” according to GFOA (2008) guidelines is present on the audit committee, else 0 (denoted AUDIT COMMITTEE WITH FINANCIAL EXPERT), and the second set to 1 if a financial expert is not present on the audit committee, (denoted AUDIT COMMITTEE WITH NO FINANCIAL EXPERT), else 0.11 Note that the sample size remains at 240 observations in this analysis, and that municipalities with no audit committee are reflected in the constant term.
We use a similar approach for model (4). We partition our audit committee variable into two indicators—the first set to 1 if all audit committee members are independent (i.e., not full-time employees of the municipality), else 0 (denoted INDEPENDENT AUDIT COMMITTEE), and the second set to 1 if the audit committee contains at least one member identified as an insider (denoted NON-INDEPENDENT AUDIT COMMITTEE), else 0. This categorization is consistent with both the corporate sector, and GFOA (2008) guidelines.12
An advantage of this approach is that it enables direct comparisons of the magnitude of coefficients across audit committees conditional on independence or financial expertise. However, the analysis should be interpreted with some caution, since partitioning the audit committee variable results in substantially less variation in our specifications. That is, while approximately 20 percent of our sample municipalities have audit committees, 45 percent of these (or 9 percent of the total sample) are completely independent, and 57 percent (or 11 percent of the total sample) have at least one member with financial expertise.
A summary of the findings on audit committee characteristics is displayed in Table 5. For parsimony, we present only the two audit committee indicators and the pseudo R2 of the overall model. Estimates (unreported) for control variables are similar to those shown in Table 4. Panel A shows that audit committees with or without the presence of a financial expert are associated with fewer disclosed internal control weaknesses, as highlighted by the significant coefficients for AUDIT COMMITTEE WITH FINANCIAL EXPERT and AUDIT COMMITTEE WITH NO FINANCIAL EXPERT. Furthermore, we find no evidence of a statistical difference when we restrict coefficients on AUDIT COMMITTEE WITH FINANCIAL EXPERT and AUDIT COMMITTEE WITH NO FINANCIAL EXPERT to be identical. At a minimum, the evidence suggests that the presence of a financial expert, or the lack of a financial expert, does not alter interpretation of our audit committee effectiveness results in the municipal context.
Table 5, Panel B presents evidence on audit committee independence. We find that the presence of at least one insider on audit committees is associated with fewer disclosed internal control weaknesses, as indicated by the significant coefficient for NON-INDEPENDENT AUDIT COMMITTEE. Furthermore, we find no evidence of a statistical difference in the magnitude of the coefficients between INDEPENDENT AUDIT COMMITTEE and NON-INDEPENDENT AUDIT COMMITTEE. At a minimum, the evidence suggests that the presence of at least one insider does not negatively impact audit committee effectiveness. This evidence is consistent with characterizations in Klein (1998), who finds the presence of insiders on finance committees is positively associated with improved corporate performance. Klein (1998) interprets such evidence as consistent with the Fama and Jensen (1983) notion that inside directors provide valuable information to boards, and consequently, boards strategically place inside directors on board committees. In the municipal setting, it is plausible that municipalities consider the audit committee's advising versus monitoring roles, and conclude that the benefits of placing insiders on the audit committee in an advisory capacity outweigh the potential costs of less independent monitoring (Linck, Netter, and Yang 2008).
A potentially competing explanation is that results for non-independent audit committees are driven by financial expertise. To investigate further, we partition NON-INDEPENDENT AUDIT COMMITTEE into two indicators, where NON-INDEPENDENT AUDIT COMMITTEE WITH FINANCIAL EXPERT is set to 1 if at least one member of the audit committee is employed by the municipality and is a financial expert, and NON-INDEPENDENT AUDIT COMMITTEE WITHOUT FINANCIAL EXPERT is set to 1 if at least one member of the audit committee is employed by the municipality, but none are financial experts. We find that of the 26 audit committees that have at least one inside director, 18 (69 percent) have at least one insider with financial expertise, while the remaining eight (31 percent) have insiders with no financial expertise. Although both coefficients are significantly negative in Column , we find a statistically significant difference in the magnitude of the coefficients between NON-INDEPENDENT AUDIT COMMITTEE WITH FINANCIAL EXPERT and NON-INDEPENDENT AUDIT COMMITTEE WITHOUT FINANCIAL EXPERT (p-value = 0.01). This result suggests that having at least one inside audit committee member with financial expertise strengthens associations in terms of internal control weaknesses over those found for audit committees containing insiders without financial expertise. One possible interpretation of this result is that access to financial expertise in an advisory role motivates including insiders on audit committees.
We perform a series of additional tests to verify that our results are robust. One possible driver of our multivariate results involves the definition of internal control weaknesses. To consider an absolute measure of internal control weaknesses, we also use a dependent variable designed to capture multiple instances of internal control weaknesses. In this specification, we performed a Poisson regression with the count of internal control weaknesses between 2001 and 2004 as the dependent variable (ranging from 0 to 4), and noted very similar results (untabulated) to those presented in Table 4. Additionally, we substitute revenue per capita in lieu of debt per capita to consider municipal wealth, and note results are consistent with those in Table 4.
We consider the possibility that audit committees established more recently—when increased attention has been centered on corporate and municipal governance—operate differently than longstanding committees. To consider the impact of audit committee formation dates, we collect audit committee formation year for 38 of the 47 municipalities with audit committees in our sample.13 Results indicate an average formation year of 1994, with 26 (12) audit committees formed before (after) 1997, the year of the first GFOA recommendation on audit committees (GFOA 2008). Next, we create three indicators to distinguish (1) 26 audit committees formed before 1997, (2) 12 audit committees formed after 1997, and (3) 9 audit committees where we cannot identify a formation date. Findings (untabulated) show that the coefficients for all three audit committee formation indicators are negative and significant, but the coefficient on the pre-1997 formation indicator is more negative than the post-1997 formation indicator (p-value = 0.08) suggesting that recently formed audit committees do not drive our results.
We investigate relations between municipal audit committees and internal control quality as a first step toward understanding the role of financial oversight in the public sector context. Because audit committees are not mandated, the public sector offers an interesting setting to examine such relations. Our analysis is also motivated by the large economic significance of local government finances, combined with recent concerns over government reporting quality (GFOA 2008). We characterize our analysis as exploratory given the lack of a well-developed theory to guide the investigation.
Primary findings are that municipal audit committees are associated with fewer internal control deficiencies. This suggests that audit committee oversight promotes high quality internal control structures, which in turn should strengthen accounting systems in a way that deters future financial reporting failures. Supplemental analysis suggests that audit committee financial expertise does not strengthen associations over those found for audit committees without financial expertise. Moreover, we find no evidence that audit committee independence strengthens associations beyond those found for audit committees containing insiders. However, we do find some evidence that associations are greater for audit committees where insiders have financial expertise, suggesting that access to financial expertise is one motivation for including insiders on municipal audit committees. These results are important for regulators and others contemplating the importance of audit committee independence and financial expertise in the governmental sector.
One exception is Baber, Gore, Rich, and Zhang (2013), who find that the presence of audit committees mitigates subsequent interest cost penalties following municipal financial restatements. Accounting restatements represent material failures of external financial reporting, and as such, measure very poor quality accounting.
A reportable condition is defined as a significant deficiency in the design or operation of internal control that could adversely affect the financial statements, while a material weakness is a reportable condition so severe that the internal control components cannot reduce the risk of material noncompliance to an acceptable level. It should be noted that the Statement on Auditing Standards No. 112 revised the language surrounding disclosure of internal control weaknesses for audits conducted after December 15, 2006, with one specific change involving replacement of the term “reportable condition” with “significant deficiency” (American Institute of Certified Public Accountants [AICPA] 2006). However, this change took place after the end of our sample period.
Specifically, initiatives allow citizens to propose new legislation, which may be enacted into law by a majority popular vote, while popular referenda allow citizens to approve or reject statutes previously adopted by elected representatives.
Follow up procedures consist of additional mail requests and telephone calls to non-responding municipalities. To test for non-response bias we follow Oppenheim (1992) and assume that late respondents are similar to non-respondents. We compare the rate of audit committee existence for early and late responders, and note no statistically significant difference (t-statistic = 0.90).
Note that our results are also robust to measuring internal control weaknesses using two levels, where municipalities with at least one reportable condition are coded as 1, and those with no internal control weaknesses coded as 0.
Results (untabulated) are similar using a Heckman (1979) two-stage procedure that considers the possibility of selection bias related to the presence of an audit committee. The first stage models the likelihood a municipality maintains an audit committee, which we use to extract the inverse Mills ratio for the second stage regression.
Variance inflation factors (VIF) suggest that multicollinearity is not an issue in our analysis. We find no value above 2 for any independent variables.
The GFOA also includes guidance regarding audit committee size, but we chose to omit this from analysis given that 91 percent of our sample audit committees meet or exceed the suggested size of three members.
Audit committee independence details are gathered as part of the survey described in the “Municipal Internal Control Quality” section.
These collection procedures involved a combination of website reviews, as well as inquiries of finance personnel.
Municipal Audit Committee Survey
5. Does your municipality have an audit committee?
(If yes, please proceed to question 6. If not, then please check “no,” and return this page of the survey in the enclosed envelope.)
6. Does your audit committee have a charter defining roles and responsibilities?
7. For each year, please indicate how many people were on the audit committee.
8. For each year, how many audit committee members were full-time municipal employees?
9. For each year, how many audit committee members would qualify as “financial experts” (see definition below)?
** According to the GFOA, a “financial expert” should, through both education and experience, and in a manner specifically relevant to the government sector, possess an understanding of generally accepted accounting principles and financial statements, as well as experience in:
1. The preparation or auditing of financial statements of comparable entities.
2. The application of such principles in connection with the accounting for estimates, accruals, and reserves.
3. Internal accounting controls and an understanding of audit committee functions.
10. For each year, how many audit committee members were both full-time municipal employees and qualify as financial experts?
The authors thank two anonymous referees and William R. Baber (editor) for their helpful comments and suggestions. The authors are grateful to Royce Burnett, Randy Elder, Angela Gore, Chris Jones, Terry O'Keefe, Jeremy Schwartz, Thomas Vermeer, and Jayaraman Vijayakumar for comments and advice. Workshop participants at The George Washington University, Jyväskylä University, Loyola University Maryland, Tampere University, Virginia Commonwealth University, and the 2013 AAA Government and Nonprofit Section Midyear and Annual Meetings offered useful suggestions. The authors also appreciate the assistance of Lei Gao in data-collection efforts.