Public Pension Governance and Opportunistic Accounting Choice: A Politico-Economic Approach

Abstract

The selection of actuarial assumptions used to value state and local government pension liabilities is an important culprit of the looming state and local pension crisis in the U.S. Due to the impact these selection choices have on the value of pension liabilities and annual required contributions (ARC), pension plans are often said to make these choices opportunistically for purposes of freeing up budget resources and making pension funding look better. Using empirical data on 114 state-administered pension plans, this research shows that the likelihood of such opportunistic pension accounting choices (OPAC) increases when the plan is underfunded, organized as a cost-sharing plan, governed by a politically embedded fiduciary body, and when the sponsoring government is surrounded by a high degree of unionization, and is divided in terms of partisan control. The results also show that the likelihood of OPAC decreases when a pension plan is subjected to an audit by a Certified Public Accountant (CPA), suggesting that professional gatekeepers can play an important role in limiting the adverse effects of OPAC behavior, including insufficient ARC payments and reduced transparency of governmental financial reports.

Keywords

public pension governance opportunistic accounting pension funding discount rates

Introduction

Unfunded pension liabilities have become a pressing fiscal challenge for many state and local governments in the U.S. Recent analyses suggest that only modest improvements have been made over the last decade in terms of placing pension plans on a sustainable funding path (Munnell & Aubry, 2016; Prosek & Rodriguez, 2019). Moreover, those improvements are contingent on the extent to which pension plans are able to realize assumed long-term expected investment returns, which is a highly contested issue. A failure to realize these returns is problematic not only because it adversely affects investment revenue, but also because it reduces annual required contribution (ARC) amounts.¹ The latter follows from prevailing state and local government accounting standards, which require pension plans to use their assumed long-term expected returns when calculating the actuarial value of their pension liabilities. The selected rate, commonly referred to as the discount rate, has a significant impact not only on this value, but also on ARC amounts. The selection of the discount rate has, as a result, become one of the most frequently mentioned causes of the current public pension funding problem (Wang & Peng, 2018).

However, despite its influence on the value of pension liabilities and ARC amounts, systematic research aimed at determining the causes of discount rate changes has been scant. It has been confined to a limited number of studies that examines how various institutional, economic, and organizational factors affect the selection of the discount rate and other related actuarial assumptions (Brooks, 2019; Eaton & Nofsinger, 2004; Stalebrink, 2014; Vermeer et al., 2010; Wang & Peng, 2018). The evidence generated by these studies indicates that the selection of actuarial assumptions often is driven by the pursuit of political and economic goals, rather than accurate measurement. The literature refers to such selection choices as either “optimistic” (Eaton & Nofsinger, 2004; Vermeer et al., 2010) or “opportunistic” (Stalebrink, 2014).

The purpose of this research is to further the understanding of opportunistic pension accounting choices (OPAC), by developing and testing a politico-economic model that considers the role of professional gatekeepers in OPAC. The study is conducted in the context of U.S. state-administered defined-benefit (DB) pension plans. Explicitly stated, therefore, the following research question is examined: How do professional gatekeepers and politico-economic factors influence OPAC in the context of U.S. state-administered DB-plans?

The application of a politico-economic model is helpful when examining this question, because it captures not only the incentives of elected officials, but also the influence that other key political stakeholders might have on accounting choices. Moreover, by considering the role of professional gatekeepers, the research contributes toward focusing existing theory on a topic that has received scant attention in the public pension context. In the literature review conducted as part of this research, only one study was found that considered the role of auditors in combatting OPAC. Moreover, this study, by Vermeer et al. (2010) produced findings that, although weak, contradict the accepted theoretical idea that audits conducted by industry specialist auditors induce monitoring effects that deter organizations from engaging in opportunistic accounting choice (Deis & Giroux, 1992; Elder et al., 2015; Garven et al., 2018; Giroux & Jones, 2011; Lowensohn et al., 2007). Vermeer et al. (2010, p. 535) suggest that this may indicate that “. . . specialized auditors know how to follow the rules but may ‘help’ their clients [with OPAC] at the same time.” Our research adds clarity to this finding by examining a broader set of gatekeeper characteristics that are theorized to affect a gatekeeper’s ability to detect and prevent OPAC, including proxies for their level of competence and their ability to make decisions independently. Our research also controls for the potential effect that actuaries might have on OPAC. As noted by Chen (2018), state pension systems are assisted by actuaries who set assumptions in accordance with professional actuarial standards. Similar to auditors, they fulfill a gatekeeper function and are often required to adhere to high professional standards of expertise and independence to enter their profession (Gunz et al., 2009).

This research also makes improvements to the empirical testing of OPAC. To minimize endogeneity, we exploit the implementation of the Governmental Accounting Standards Board’s (GASB) Statements No. 67 and 68 (GASB 67 and GASB 68) as an exogenous shock that temporarily increased the prevalence of OPAC among state and local pension plans. Most important, GASB 67 required a large number of underfunded pension plans to make downward adjustments to their discount rates, unless they made sufficient improvements to their funding policies. Such downward adjustments are politically challenging because they increase the actuarially reported value of pension liabilities, and the size of ARC payments. ARC payments are typically funded via general funds and as such compete directly with government programs that serve more general and immediate needs than pension funding do (Coggburn & Kearney, 2010; Peng, 2004). The adoption of GASB 68 augmented the significance of GASB 67 from an OPAC perspective, by requiring sponsoring governments to place pension liabilities on their balance sheet. Prior to the implementation of GASB 68, these were left off the sponsoring government’s balance sheet and reported as part of the general long-term debt and asset account group section of the financial report. It should be noted that GASB 68 was implemented in fiscal year (FY) 2014/2015 and that GASB 67 was implemented in FY 2013/2014. In this research we assume that OPAC decisions made in 2013/2014 include consideration, at least partly, of the impact they will have on the 2014/2015 accounting period.

The article is organized into five major sections. The first section explains how the adoption of GASB 67 and GASB 68 produced new opportunities and incentives for pension plans and sponsoring governments to engage in OPAC, and relates these changes to prior literature. The second section introduces the politico-economic theory of accounting policy, and the hypotheses that are tested. The third section describes the empirical research approach, including data, variables and methods. The fourth section presents the results and the analysis. The concluding section offers a brief summary of the results and a discussion of their policy relevance and what practitioners can learn from the study.

Actuarial Assumptions and Opportunistic Accounting Choice

Consistent with prior research, we define opportunistic accounting choice as the deliberate use of accounting discretion for purposes of artificially distorting measures of financial performance, including financial position (i.e., financial health) and results (see Healy & Wahlen, 1999). While the large majority of empirical research on this topic has centered on private firms (Dechow et al., 2010; Healy & Wahlen, 1999), several empirical studies have also been produced in public sector contexts (e.g., Arcas & Marti, 2016; Cohen et al., 2019; Donatella, 2020; Ferreira et al., 2013; Greenwood et al., 2017; Stalebrink, 2007). Similar to studies conducted on private firms, these studies have typically relied on an agency or public choice framework, where it is argued that public officials have incentives to use discretion in accounting policy, because it aids them in the pursuit of economic and political goals.

In the public pension context, empirical work has been built around the argument that elected officials have incentives to engage in OPAC because it allows them to free up current budget resources, and artificially improve a government’s reported financial health. The main focus has been on determining whether public pension plans adopt actuarial assumptions for purposes of reducing the actuarial value of pension liabilities. For underfunded plans, such actions reduce the size of the unfunded actuarial liability (UAL), thereby making the plan appear better funded. Moreover, an improvement in a plan’s UAL increases available budget resources, by reducing the size of ARC amounts. In brief, the ARC is an estimate of the cost of pension benefits earned during a fiscal year (referred to as the “normal cost” of pensions), which is typically paid from the general fund. If a plan is underfunded it also includes an amortization amount to pay down the UAL.

Regardless of context, public pension accounting standards often extend substantial measurement discretion to entities. The intent is to improve communication by allowing professional judgment to enter the accounting measurement process. Such judgment allows managers to more effectively convey information about their entity’s performance. As explained by Healy and Wahlen (1999, p. 366), it allows managers to “. . . use their knowledge about the business and its opportunities and to select reporting methods, estimates, and disclosures that match the firms’ business economics, potentially increasing the value of accounting as a form of communication.”

However, there is always a risk that accounting measurement discretion is used opportunistically in a manner that results in accounting choices that do not reflect the entity’s underlying economics (Healy & Wahlen, 1999). In the U.S. public pension context, these discretionary choices include several types of actuarial assumptions, including the selection of the discount rate, amortization period, amortization method, and payroll growth assumption.² Due to its relative impact on the value of reported pension liabilities, particular attention has been given to the choice of discount rate. In brief, this choice refers to the rate used to discount pension obligations. Estimates produced by Winkelvoss (1993) suggest that a 1% increase in the discount rate can reduce the value of actuarial liabilities by as much as 16%. For U.S. state and local governments, the discount rate is determined based on assumptions made about the long-term expected rate of return of accumulated pension assets. Because changes in the discount rate are inversely related to the value of reported pension liabilities, prior research has assumed that OPAC behavior is reflected in decisions that reduce the discount rate (Eaton & Nofsinger, 2004; Stalebrink, 2014; Vermeer et al., 2010; Wang & Peng, 2018).

The other three actuarial assumptions are likely to have more modest effects on the size of the UAL and ARC amounts. Empirical research suggests that they sometimes are applied in combination with discount rate changes to achieve OPAC goals (Eaton & Nofsinger, 2004; Vermeer et al., 2010). Among the three remaining actuarial assumptions, the selection of amortization period refers to the number of years over which underfunded plans amortize their UAL. Prevailing accounting standards allow pension plans to use an amortization period that ranges between 1 and 30 years. The selection of a shorter amortization period increases the UAL and the ARC, while a longer amortization period reduces them. Prior research has therefore assumed that OPAC behavior is reflected in decisions that increase the length of the amortization period (Eaton & Nofsinger, 2004; Vermeer et al., 2010).

The selection of amortization method refers to a choice between two available methods, including a closed and an open-basis method. Under the open-basis method, the amortization period starts over each valuation period. Under the closed method, the unfunded liability is paid off during the previously selected period. Given that the period starts over when the open-basis method is applied, it will result in a lower amortization amount, compared with the closed method. Prior research has therefore assumed that OPAC behavior is reflected in a preference for the open-basis method (Vermeer et al., 2010).

The selection of payroll growth assumption refers to actuarial decisions about the rate by which current and future beneficiaries and associated salaries will grow. A reduction in this growth rate will lower the actuarial value of the pension liability. Prior research has therefore assumed that OPAC behavior is reflected in reductions in salary growth assumptions (Eaton & Nofsinger, 2004; Vermeer et al., 2010).

Prior research has also considered the use of an accounting technique that affects the value of pension assets, referred to as asset-smoothing (Chen, 2018). Asset-smoothing is a technique that allows plans to smooth out variations in asset values across fiscal years. Chen (2018) notes that an increase in the asset-smoothing period during times of fiscal stress reduces current pension costs for governments. In this research, OPAC is therefore assumed to be associated with increases in the asset-smoothing period.

Existing empirical evidence has shown that the likelihood that one or some combination of the above actuarial assumptions are used in OPAC is dependent on a variety of economic, organizational and institutional factors. The evidence is relatively consistent in terms of the effects of economic factors. It indicates that underfunded pension plans, sponsored by governments that are experiencing fiscal stress are significantly more likely to engage in OPAC, compared with well-funded plans that are sponsored by fiscally stable governments (Chaney et al., 2002; Eaton & Nofsinger, 2004; Stalebrink, 2014; Vermeer et al., 2010; Wang & Peng, 2018). Examinations of the effects of institutional and organizational factors on OPAC have produced either mixed or inconclusive evidence, including the role of board composition (Brooks, 2019; Stalebrink, 2014; Wang & Peng, 2018), unionization (Brooks, 2019; Chaney et al., 2002; Vermeer et al., 2010; Wang & Peng, 2018), government ideology (Wang & Peng, 2018), and monitoring agents (Stalebrink, 2014; Vermeer et al., 2010).

Hypothesis Development

Similar to many other policy contexts, U.S. state and local pension accounting and reporting is a matter of wealth distribution. As no concentration or capitalization is permitted under public property rights (Zimmerman, 1977), the wealth transfer at stake is primarily confined to service allocations between different periods and generations (Stalebrink, 2014). In the public pension context, this is reflected in accounting choices that either reduce the size of unfunded pension liabilities and free up budget resources in the short-term, or that increase the size of pension liabilities and reduce available budget resources (Coggburn & Kearney, 2010; Peng, 2004).

Cheng (1992), Zimmerman (1977) and other proponents of the politico-economic theory of accounting policy argue that accounting choices tend to reflect a balance between the will and interests of key political stakeholders. As will be discussed below, the public pension context is surrounded by multiple stakeholders with competing political interests. When analyzed through the lens of the politico-economic theory, emphasis is therefore placed on how contextual factors affect the workings of accounting choice. Moreover, it assumes that the adoption of common accounting standards (i.e., de jure harmonization) by itself is unlikely to be sufficient for harmonization of pension accounting standards to occur (i.e., de facto harmonization; cf. Tay & Parker, 1990). In this research, we theorize that OPAC behavior is affected by a combination of factors, including economic, political and bureaucratic conditions, and audit factors.

Economic Conditions

A fundamental assumption underlying this research is that elected officials have incentives to engage in OPAC activities that free up budgetary resources and make pension funding and the financial condition of the sponsoring government look better. In theory, these incentives exist because they improve chances of re-election, career advancement and other perceived benefits (e.g., Baber, 1983; Zimmerman, 1977). However, to deviate from generally accepted accounting principles (GAAP) imply a political risk, because such choices can impose political costs on elected officials (Donatella, 2020). Considering that elected officials, in a politico-economic framework, are assumed to be rational utility maximizers (Zimmerman, 1977), they will weigh the risk of political costs associated with OPAC against the utility it brings.

Prior empirical research indicates that budgetary constraints increase the likelihood of OPAC (Chaney et al., 2002; Eaton & Nofsinger, 2004; Mitchell & Smith, 1994). A sponsoring government experiences budgetary constraints when its fiscal decisions and available resources are misaligned (Chapman, 2008). One repercussion of such constraints is that they hamper the government’s ability to meet balanced budget expectations. Frequent failures to meet these might negatively affect voters’ perceptions about elected officials’ ability to govern (Cohen et al., 2019; Donatella, 2020; Ferreira et al., 2013), thereby reducing chances of re-election, career advancement, and other benefits. By reducing ARC amounts, OPAC can be used to artificially reduce or eliminate a deficit, and thus enhance a government’s ability to meet balanced budget requirements. Given this, we expect governments that are struggling in terms of meeting their fiscal year obligations to be more prone to engage in OPAC. Explicitly stated, it is hypothesized that:

Hypothesis 1 (H1): Short-term budgetary constraints in a sponsoring government are positively related to OPAC.

Prior research suggest that short- and long-term fiscal conditions influence pension decisions differently (Chen, 2018; Levine et al., 1981). As explained by Chen (2018), a government’s long-term fiscal condition is the cumulative result of decisions that have been made over an extended period. Perhaps most important, these decisions are reflected in government’s overall debt burden. Given that OPAC can be used to make a government appear more financially healthy, prior research has theorized that government officials engage in OPAC activities to improve credit ratings and reduce interest rates (Stalebrink, 2014). It is argued that such outcomes are perceived as beneficial, because they reflect positively on government officials’ reputations, thereby enhancing their electability (Vermeer et al., 2010; Zimmerman, 1977). Consistent with this theory, we propose that a sponsoring government’s long-term fiscal constraints are positively related to OPAC. Explicitly stated, it is hypothesized that:

Hypothesis 2 (H2): Long-term financial constraints of a sponsoring government are positively related to OPAC.

Prior research has shown that underfunded pension plans are more likely to engage in OPAC (Chaney et al., 2002; Eaton & Nofsinger, 2004; Vermeer et al., 2010). Given that ARC amounts are partly determined by the size of the UAL, underfunding exacerbates the impact of OPAC. The utility of OPAC is therefore likely to increase with the level of pension underfunding. It is therefore hypothesized that:

Hypothesis 3 (H3): The level of pension underfunding is positively related to OPAC.

A previously unexplored factor that is related to H3 is that proximity to certain funding thresholds, used to assess financial health, might influence the likelihood of OPAC. While the origins and credibility of such thresholds have been questioned in the pension context (see, Downs et al., 2012) there appears to be some credence to the argument that credit rating agencies at least partially rely on certain common funding thresholds when making assessments of a pension plan’s financial health. Brainard and Zorn (2012, p. 2) notes that “S&P assigns a “strong” rating for funding levels above 90 percent; a rating of “above average” for levels between 80 percent and 90 percent; “below average” for funding levels between 60 percent and 80 percent; and “weak” below 60 percent.” Given that OPAC can be used to artificially improve the funding ratio, it is conceivable that sponsoring governments and pension plans that have funding levels in close proximity to these thresholds are more prone to engage in OPAC. We therefore hypothesized that:

Hypothesis 4 (H4): Proximity to common funding thresholds, used to assess the financial health of pension plans, is positively related to OPAC.

Political and Bureaucratic Conditions

Public pension plans are governed by fiduciary bodies that receive their authority from the government in the form of statue or constitution. These fiduciary bodies hold the ultimate authority over the selection of actuarial assumptions in the overwhelming majority of U.S. public pension plans (Stalebrink, 2014). Prior empirical research has shown that these fiduciary bodies, depending on their composition, can serve as enablers of OPAC. Based on a cross-sectional analysis of 88 U.S. state-administered defined-benefit pension plans, Stalebrink (2014) found that the proportion of political appointees serving on a fiduciary body increases the likelihood that plans select higher discount rates. However, more recent findings by Wang and Peng (2018) contradict this finding. Based on an analysis of discount rate changes made over time by 81 state-administered pension plans, they found that more “politically embedded” pension boards are more likely to reduce their plan’s discount rate. Given these contradictory empirical findings, directional clarity is not predicted. It is therefore hypothesized that:

Hypothesis 5 (H5): The proportion of political appointees serving on a fiduciary body is related to OPAC.

Prior literature suggests that public unions influence public pension affairs, due to their role in collective bargaining (Frandsen, 2016; Hoang & Goodman, 2018; Kearney, 2003). However, the empirical results have been mixed in terms of their influence on OPAC. Several studies suggest that pension systems surrounded by strong public unions are more likely to be poorly funded and make fiscally irresponsible decisions, such as adopting artificially high discount rates and paying a lower percentage of ARC (Anzia & Moe, 2017; Mitchell & Smith, 1994). In contrast to these findings, Vermeer et al. (2010) found that plans with strong unions are less prone to adopt “optimistic” actuarial methods and assumptions, due to increased monitoring of public pension affairs. Moreover, Wang and Peng (2018) did not find evidence in support of their hypothesis that union coverage plays a significant role in decisions to change the discount rate. Given these mixed findings, directional clarity is not predicted. It is therefore hypothesized that:

Hypothesis 6 (H6): The degree of unionization surrounding a pension plan is related to OPAC.

A common theoretical idea that has been backed with empirical evidence is that polarization between the Democratic and Republican Parties is tied with labor issues and fiscal issues related to spending and taxing (Jochim & Jones, 2012). As the public pension issue is both an important labor and fiscal issue, Anzia and Moe (2017, p. 35) suggest that it sets up for a polarized conflict between the two parties with “. . . Democrats, as the allies of unions and proponents of a more active government, strongly favoring more generous pensions, and Republicans, as union adversaries and proponents of a less active government, opposing them.” Based on this logic, it is possible that the Democratic party would be more likely to exhibit biases similar to unions. Furthermore, it is commonly argued that governments that are controlled by a relative powerful political party tend to be less transparent and less efficient, than governments that are subject to greater political competition (Oates, 1985). Combining these two theoretical propositions, it is hypothesized that:

Hypothesis 7 (H7): The relative strength and influence of the Democratic party are positively related to OPAC.

There are three major types of public pension plans at the U.S. state and local level, including single-employer plans, agent multiple-employer pension plans, and cost-sharing multiple-employer pension plans. A single-employer plan is a pension plan that is adopted by a single sponsoring government that controls its own assets and is responsible for its own pension obligations. An agent multiple-employer pension plan is established by two or more employers. It is established for the purpose of pooling investments and sharing administrative costs. The sponsoring governments maintain separate accounts of their assets and obligations. Finally, a cost-sharing multiple-employer pension plan is established by multiple employers that share responsibility for both assets and obligations. However, the state commits to fund the benefits. In the latter case, each individual employer is required to recognize their proportionate share of the net pension liability of the total plan in their financial report. This means that a larger number of governments may be impacted by OPAC decisions, thereby increasing the stakes and potentially the pressure to select favorable actuarial assumptions. Given this, it is hypothesized that:

Hypothesis 8 (H8): Cost-sharing multiple-employer plans are more likely to engage in OPAC than single- and agent multiple-employer plans.

Audit Characteristics

Accounting choice will, more or less, always be subject to external restrictions (Cheng, 1992). In the public pension accounting and reporting context, such restrictions are partly imposed on pension plans through auditors tasked with ensuring that a reporting entity adheres to the spirit of GAAP. In theory, the presence of auditors deters OPAC by increasing the risk that OPAC attempts are detected. However, empirical findings indicate that the quality of the services provided by auditors is not homogeneous (e.g., Elder et al., 2015; Garven et al., 2018). The disciplinary effects that auditors impose on pension plans and sponsoring governments in terms of deterring OPAC behavior may therefore differ, depending on audit quality. Conceptually, it is typically argued that audit quality differences arise, as a result of differences in auditors’ competence and independence (DeFond & Zhang, 2014).

In this research, we explore the association between OPAC and three different audit firm characteristics that are likely to at least partly influence a firm’s level of competence and ability to conduct audits independently. These include the auditor’s organizational affiliation, whether or not the auditor is a certified public accountant (CPA), and whether the organization is specialized in the public pension plan audit market.

Approximately 25% of state-administered pension plans in the U.S. are audited by personnel affiliated with an internal unit of their government (e.g., a state auditor or comptroller’s office). Empirical findings reported by Giroux and Jones (2011) suggest that such an organizational affiliation influences auditors’ ability to conduct audits independently. Specifically, they find that auditors affiliated with an internal unit of a government is less likely to preserve their independence compared with a third-party service provider. Given that internal auditors operate in close proximity to the political process, they might be more vulnerable to pressures that will compromise their ability to conduct audits independently, thereby expanding pension plans’ discretion over accounting choice. Moreover, reputation incentives may also influence the behavior of these units to a lesser extent, given that they do not compete on an open market with other service providers. In this research, it is therefore hypothesized that:

Hypothesis 9 (H9): The use of auditors from internal government units is positively related to OPAC.

To enter and remain in their profession, CPAs are expected to adhere to a strict code of professional conduct, which include, but is not limited to serving the public interest, act with integrity, and making decisions independently and objectively. In this study, approximately 14% of the plans were audited by non-CPAs. Given that these plans are audited by individuals who are not formally bound by the profession’s code of conduct, they may be more vulnerable to OPAC. Given this, it is hypothesized that:

Hypothesis 10 (H10): The use of non-CPA auditors is positively related to OPAC.

Empirical findings suggest that industry specialization, proxied by market share, affects the likelihood of clients’ misuse of accounting discretion (Deis & Giroux, 1992; Elder et al., 2015; Garven et al., 2018; Giroux & Jones, 2011; Lowensohn et al., 2007). The theoretical argument underlying these findings is that specialized audit firms are better equipped to withstand pressure from clients, compared with non-specialized audit firms, because reputation losses, incurred as a result of poor audit quality, are likely to be costlier, relative to losing a single client (DeFond & Zhang, 2014). Moreover, an audit firm that serves a large number of clients in the same industry is likely to have more opportunities to realize competency gains from training and practical experience. Such opportunities may give rise to superior internal knowledge sharing, which is expected to have a positive effect on a firm’s competency (Reichelt & Wang, 2010). Given this, it is hypothesized that:

Hypothesis 11 (H11): Audit firm specialization is negatively related to OPAC.

Empirical Research Approach: Data, Variables, and Method

The above hypotheses were tested using binary logistic regression models and cross-sectional data. The use of a cross-sectional approach was deemed appropriate because it allowed us to exploit the adoption of GASB 67 and GASB 68 as an exogenous shock that temporarily increased the prevalence of OPAC among pension plans. Moreover, the implementation of GASB 67 and GASB 68 gave rise to a significant level of divergence between pension policy and accounting practice (see Weinberg & Norcross, 2017), which allowed us to develop a dependent variable that exhibited substantial variation. Due to a low frequency of changes in discount rates and other actuarial assumptions, historically, prior research has been conducted using regressors that made it methodologically difficult to produce precise estimates (see, Eaton & Nofsinger, 2004; Stalebrink, 2014; Vermeer et al., 2010; Wang & Peng, 2018).

GASB’s Pension Accounting Standards and Opportunistic Accounting Choice

GASB 67 and GASB 68 were adopted by GASB in 2012. An important motive behind adopting these was to address criticisms raised against previous accounting standards (i.e., GASB Statement No. 27), including their inability to sufficiently support the production of financial statements that provide a comprehensive and accurate picture of a government’s financial health and performance (Weinberg & Norcross, 2017). The amended standards responded to these criticisms, in part, by requiring sponsoring governments to include unfunded pension liabilities or asset balances on their balance sheets, and by requiring them to report the market value of assets invested for purposes of meeting pension obligations. State and local government pension plans were required to implement GASB 67 in financial reports beginning after June 15, 2013. Sponsoring governments were required to implement GASB 68 in financial reports beginning after June 15, 2014.

When viewed through accounting framework presented by Zimmerman (1977), the adoption of GASB 67 and GASB 68 reflects a change that altered the incentives for government officials to engage in OPAC behavior. The adoption of GASB 67 temporarily changed these by requiring underfunded pension plans to make downward adjustments to their discount rate. More specifically, it required them to discount the unfunded portion of their pension liabilities using a rate that reflects the return of a high-quality general obligation municipal bond, unless they made sufficient improvements to their funding policies. Prior to the implementation of GASB 67, pension plans selected their discount rate solely on the basis of assumptions made about the long-term expected rate of return on pension assets. At that time, these rates varied between 7.0% and 8.5% with a median of about 8% (Wang & Peng, 2018). By comparison, the 20-year municipal bond index rate was approximately 3.50% during the implementation period. Ceteris paribus, this discrepancy implies that underfunded pension plans that adopted a blended rate (i.e., a rate that combines a municipal bond rate and a long-term expected return rate) were required to make downward adjustments to their discount rates.

The adoption of GASB 68, in turn, altered the incentives by amplifying the effect and the significance of discount rate changes. As noted earlier, it required sponsoring governments to place pension liabilities on their balance sheet. Prior to the implementation of GASB 68, pension liabilities were left off the sponsoring government’s balance sheet and reported as part of the general long-term debt and asset account group section of the financial report. Although GASB 68 was implemented a year later than GASB 67, it is reasonable to assume that the perceived utility associated with OPAC decisions made in FY 2013/2014 includes consideration of these changes. This assumption is consistent with earlier research, which indicates that opportunistic accounting choices often are strategic in nature (e.g., Cohen et al., 2019; Donatella, 2020; Ferreira et al., 2013; Stalebrink, 2007).

Analyses of the implementation of GASB 67 and GASB 68 indicate that the changes gave rise to a significant level of divergence between pension policy and accounting practice. Analyses conducted prior to their implementation indicate that the large majority of public pension plans would be required to make downward adjustments to their discount rate, in response to GASB 67, due to underfunding (Mortimer & Henderson, 2014). Data available from the Public Plans Database (Center for Retirement Research at Boston College, 2019) indicate that 88% of U.S. state-administered pension plans had funding ratios of 90% or less in the year immediately preceding the implementation of GASB 67 with an average funding ratio of 74%. However, despite the prevalence of underfunded pension plans, relatively few plans ended up making downward adjustments to their discount rates during the implementation year (i.e., FY 2013/2014). In their study of state and local pension plans, Weinberg and Norcross (2017) reports that less than 10% of pension plans applied a blended rate in response to GASB 67. Using data available from the Public Plans Database, this percentage increases to 23% when local plans are left out. Specifically, data on long-term investment return assumptions show that 26 out of 114 state-administered pension plans reduced their discount rate during the transition year.

A portion of the underfunded plans that did not reduce their discount rate appears to have adopted changes to their funding policies to offset the need for discount rate reductions. Data on overall contribution rates (i.e., employer and employee rates) from the Public Plans Database shows that 21 plans made material increases to these rates (defined as a change that increased the overall contribution rate by 5% or more) during the transition year. Such funding changes may have moved some plans above the so-called “crossover point,” which GASB (2013) defines as the point when “. . . projected benefit payments for current employees and inactive employees exceed the projected plan net position related to those employees.” Any benefit payments projected to be made from that point forward needs to be discounted using the high-quality municipal bond interest rate (i.e., it would require a blended rate). The number of funded plans that appears to have adhered to the spirit of GASB 67 only account for about 35% of the plans, including sufficiently funded plans and plans that improved their funding policies.

Classification of OPAC Plans: Dependent Variable Measurement

Plans were classified into OPAC and non-OPAC plans, using data on actuarial assumptions available from the public plans database. As depicted in Figure 1, 72 out of 114 state-administered pension plans were classified as exhibiting OPAC behavior during the GASB 67 implementation year. OPAC plans were coded as “1” and non-OPAC plans as “0.”

Figure 1.

OPAC/non-OPAC plan classifications.

The most substantial part of the classification process involved classifying underfunded plans that did not make downward adjustments to their discount rates (i.e., kept it stable) into OPAC and non-OPAC plans. To be classified as an OPAC plan the plan had to be (a) materially underfunded, and (b) show no evidence of actuarial changes inconsistent with OPAC during the transition year. A materially underfunded plan was defined as a plan that had a funding ratio below 90%. Funding below this level was assumed to be substantial enough to warrant at least a small adjustment to the discount rate, given the large difference between the rate that plans typically assumed prior to change and the rate of return associated with a high-quality general obligation municipal bond.

The review of actuarial changes included material increases in payroll growth rates (defined as an increase exceeding 5%), changes from an open to a closed discount method, material reductions in amortization periods (a change exceeding 5 years), and increases in the asset-smoothing period. As depicted in Figure 1, 17 of the plans that kept their discount rate stable during the transition year were either sufficiently funded or adopted actuarial changes that were inconsistent with OPAC. These plans were classified as non-OPAC plans. Among the remaining plans that kept their discount rate stable, 67 were materially underfunded. Among these 67 plans there was no evidence of accounting choices that were inconsistent with OPAC.

As depicted in Figure 1, four plans also increased their discount rate during the transition year. In this research, such changes are classified as OPAC, unless (a) they were accompanied with a stated justification in the plan’s most recent experience review study, or (b) if material changes had been made to the plan’s overall portfolio risk. A review of changes in investment goals, target allocations across asset classes, and related experience review studies offered no evidence that accompanying changes had been made to investment objectives and targets. These remained stable across FYs 2013 and 2014 in all four cases. Consistent with the definition of OPAC, these plans were therefore classified as OPAC.

Finally, 26 plans reduced their discount rates during the transition year, which is indicative of adherence to the spirit of GASB 67. However, one of the plans, Illinois SERS, was classified as OPAC, because the discount rate reduction was postponed until the year 2065, which significantly reduces the impact of the change. The plan’s 2014 CAFR explains: “. . . the long-term expected rate of return on pension plan investments was applied to projected benefit payments through the year 2065, and the municipal bond rate was applied to all benefits payments after that date.”

Independent Variable Measurements

Data for the independent variables were primarily gathered from CAFRs (FY 2013 and 2014) and the Public Plans Database. Other important sources included the U.S. Census Bureau, UnionStats, and the State Fiscal Ranking survey (see Table 1). These sources generated complete data for all 114 state-administered pension plans. These plans administer close to 95% of public pension assets nationwide and cover more than 90% of members of state and local pension plans in the U.S.

Table 1.

Summary of Variables.

Variable	Hypothesis (sign)	Measure	Data source	Measurement period (FY)
Dependent variable
Opportunistic pension accounting choice (OPAC)		Indicator variable coded as “1” if required discount factor adjustment was not made, otherwise “0”.	Public plans data^a	2014
Economic (independent) variables
Short-term budget solvency index, negative (SHORT_NEGATIVE)	H1 (+)	Based on Norcross’ (2015) budget solvency index. Indicator variable coded as “1” if the sponsoring governments were not able to meet their fiscal obligations during FY 2013, otherwise “0”.	State Fiscal Rankings Data	2013
Long-term budget solvency index, bottom quartile (LONG_BOTTOMQUARTILE)	H2 (+)	Based on Norcross’ (2015) long-term budget solvency index. Indicator variable coded as “1” if the sponsoring governments were one of the governments in the bottom quartile of the index, otherwise “0”.	State Fiscal Rankings Data	2013
Budget solvency index, short-term negative and long-term bottom quartile (NEGATIVE_BOTTOMQUARTILE)	H1/H2 (+)^a	Indicator variable coded as “1” if the sponsoring governments were not able to meet their fiscal obligations during FY 2013 and were one of the governments in the bottom quartile of the long-term budget solvency index, otherwise “0” (i.e., SHORT_NEGATIVE “1” and LONG_BOTTOMQUARTILE “1”).	State Fiscal Rankings Data	2013
Budget solvency index, short-term negative and long-term top three quartiles (NEGATIVE_TOP3QUARTILES)	H1/H2 (+)^a	Indicator variable coded as “1” if the sponsoring governments were not able to meet their fiscal obligations during FY 2013 and were not one of the governments in the bottom quartile of the long-term budget solvency index, otherwise “0” (i.e., SHORT_NEGATIVE “1” and LONG_BOTTOMQUARTILE “0”).	State Fiscal Rankings Data	2013
Budget solvency index, short-term positive and long-term bottom quartile (POSITIVE_BOTTOMQUARTILE)	H1/H2 (+)^a	Indicator variable coded as “1” if the sponsoring governments were able to meet their fiscal obligations during FY 2013 and were one of the governments in the bottom quartile of the long-term budget solvency index, otherwise “0” (i.e., SHORT_NEGATIVE “0” and LONG_BOTTOMQUARTILE “1”).	State Fiscal Rankings Data	2013
Budget solvency index, short-term positive and long-term top three quartiles (POSITIVE_TOP3QUARTILES)	H1/H2 (–)^a	Indicator variable coded as “1” if the sponsoring governments were able to meet their fiscal obligations during FY 2013 and were not one of the governments in the bottom quartile of the long-term budget solvency index, otherwise “0” (i.e., SHORT_NEGATIVE “0” and LONG_BOTTOMQUARTILE “0”).	State Fiscal Rankings Data	2013
Pension funding ratio (FUNDING)	H3 (–)	Plan assets as percentage of pension liabilities	Public plans data^b	2013
Proximity to financial thresholds (THRESHOLDS)	H4 (+)	Indicator variable coded as “1” if pension plans had a funding ratio within 59 to 61, 79 to 81 and 89 to 91, otherwise “0”.	Public plans data^b	2013
Political and bureaucratic (independent) variables
Unionization of public sector employees (UNION)	H5 (?)	Number of the public workforce in a jurisdiction that is union member, in relation to the total workforce in that jurisdiction.	Union stats data^e	2014
Proportion of political appointees (POLITICAL)	H6 (?)	Number of political appointees on the board of the fiduciary body, in relation to total board size.	CAFRs	2014
Unified Democratic government (UNIFIED_DEM)	H7 (?)	Indicator variable coded as “1” if government was governed by unified Democratic party, otherwise “0”.	Census data	Pre 2014
Pension plan type (COST_SHARING)	H8 (+)	Indicator variable coded as “1” when pension plan is cost-sharing multiple-employer plan, otherwise “0”.	Public plans data^b	2013
Audit (independent) variables
Internal auditors (INT_AUDIT)	H9 (+)	Indicator variable coded as “1” if pension plan was audited by internal auditors, otherwise “0”.	CAFRs	2014
Non-CPA auditors (NON_CPA)	H10 (+)	Indicator variable coded as “1” if pension plan was audited by a non-CPA auditor, otherwise “0”.	CAFRs	2014
Internal auditors that is non-CPA (INTAUDIT_NONCPA)	H9/H10 (+)^a	Indicator variable coded as “1” if there was an overlap between internal audit and non-CPA auditors (i.e., INT_AUDIT “1” and NON_CPA “1”), otherwise “0”.	CAFRs	2014
Internal auditors that is CPA (INTAUDIT_CPA)	H9/H10 (–)^a	Indicator variable coded as “1” if there was not an overlap between internal audit and non-CPA (i.e., INT_AUDIT “1” and NON_CPA “0”), otherwise “0”.	CAFRs	2014
Industry specialist audit firm (SPEC_FIRM)	H11 (–)	Indicator variable coded as “1” if pension plan was audited by one of the market leading audit firms, otherwise “0”.	CAFRs	2014
External auditors affiliated at an industry specialist audit firm (EXTAUDIT_SPECFIRM)	H11 (–)^a	Indicator variable coded as “1” if there was an overlap between external audit and specialist audit firm (i.e., INT_AUDIT “0” and SPEC_FIRM “1”), otherwise “0”.	CAFRs	2014
External auditors not affiliated at a non-industry specialist audit firm (EXTAUDIT_NONSPECFIRM)	H11 (–)^a	Indicator variable coded as “1” if there was not an overlap between external audit and specialist audit firm (i.e., INT_AUDIT “0” and SPEC_FIRM “0”), otherwise “0”.	CAFRs	2014
Control variables
Internal actuaries (INT_ACTUARIES)	(?)	Indicator variable coded as “1” if valuation of pension plans liabilities was conducted by internal actuaries, otherwise “0”.	CAFRs	2014
Industry specialist actuarial firm (SPEC_ACTUARIALFIRM)	(?)	Indicator variable coded as “1” if valuation of pension plans liabilities was conducted by the market leading actuarial firm, otherwise “0”.	CAFRs; Public plans data^b	2013, 2014
Partial payment of the ARC (ARC_FAIL)	(?)	Indicator variable coded as “1” if government failed to pay full ARC, otherwise “0”.	Public plans data^b	2014
Reform of pension fund structure (REFORM)	(–)	Indicator variable coded as “1” if government undertook material funding policy changes, otherwise “0”.	Public plans data^b	2013, 2014
Pension plan size (PLAN_SIZE)	(?)	Natural logarithm of average pension liability.	Government financial report; Census data	2013

Note. CAFRs = Comprehensive annual financial reports. ^aThe predicted signs are related to the reference variable used. For short- and long-term budget solvency index variables, POSITIVE_TOP3QUARTILES was used as reference variable, and for audit variables INTAUDIT_NONCPA was used as reference variable. ^bData available at http://publicplansdata.org.^c https://www.mercatus.org/statefiscalrankings. ^d http://unionstats.com and http://unionstats.gsu.edu/CPS%20Documentation.htm.

To measure budgetary constraints faced by sponsoring governments, we relied on a budget solvency index developed by Norcross (2015). This index measures the extent to which a sponsoring government is able to meet its fiscal year obligations, by combining its operating ratio and its surplus or deficit per capita. The measure is based on a method for assessing fiscal condition developed by Wang et al. (2007) and is inclusive in terms of dimensions of financial condition identified by public finance scholars (Hendrick, 2004; Kloha et al., 2005; Poterba, 1994). The indicator variable (SHORT_NEGATIVE) was developed by coding sponsoring governments that were able to meet their fiscal obligations during FY 2013 as “1,” and those that were not as “0.”

To capture long-term financial constraints, we relied on the long-term budget solvency index developed by Norcross (2015). This index measures the extent to which the sponsoring government is able to meet its long-term obligations. Similar to the budget solvency index, it is based on a method for assessing fiscal condition developed by Wang et al. (2007). The index data suggest that the majority of governments were fiscally stable during the relevant time-period. To capture governments that experienced long-term fiscal constraints, we therefore created an indicator variable (LONG_BOTTOMQUARTILE) that included the bottom quartile of governments in terms of long-term budget solvency (FY 2013). Sponsoring governments in the bottom quartile were coded “1,” and the other three quartiles were coded “0.”

To measure the level of pension funding (FUNDING), we used the 2013 pension funding ratio available in the public plans database. This ratio measures a plan’s assets as a percentage of its liabilities. Moreover, the same data were used to measure proximity to common funding thresholds (THRESHOLDS). This indicator variable was coded as “1” if a plan’s funding ratio was within a close proximity of the three thresholds (i.e., 60%, 80%, and 90%), defined as +/– 1% of any of the three thresholds (i.e., 59–61, 79–81, and 89–91), otherwise “0.”

Similar to earlier studies (Stalebrink, 2014; Wang & Peng, 2018), a fiduciary body’s degree of political alignment (POLITICAL) was measured as the number of political appointees on the board, in relation to total board size. Similar to Munnell et al. (2011) the degree of unionization surrounding a particular pension plan (UNION) was measured as the proportion of the total public workforce in a specific jurisdiction that are union members. Both variables were calculated using data from the year 2014.

The indicator variable for unified democratic party (UNIFIED_DEM) was coded as “1” if the Democratic party held both the executive and the legislative branches in 2013, otherwise “0.” To capture plan type, an indicator variable was coded “1” for plans that are organized as a cost-sharing multiple-employer pension plan, otherwise “0” (COST_SHARING). Data on plan type was pulled from the public plans database (FY 2013).

Pension plans audited by an internal auditor were coded as “1,” otherwise “0” (INT_AUDIT). Pension plans audited by an auditor that was not a CPA were coded as “1,” otherwise “0” (NON_CPA). Plans that were audited by a principal auditor that was not a CPA were exclusively from plans that were audited by an in-house auditor (e.g., by a state auditor).

In prior literature, audit firm specialization has been operationalized in different ways. Whereas Lowensohn et al. (2007) and Vermeer et al. (2010) proxied specialization based on a threshold of at least 10% of the market share, others have used continuous market share measures (Deis & Giroux, 1992; Elder et al., 2015; Garven et al., 2018; Giroux & Jones, 2011). Given that there are two audit firms in the data set (i.e., KPMG and CliftonLarssonAllen [CLA]) that clearly differentiate themselves from their competitors in terms of market share, the relevant indicator variable was coded as “1” if KPMG or CLA audited a pension fund in FY 2014, otherwise “0” (SPEC_FIRM). When measured based on overall size of pension plans audited by third-party professional audit firms in 2014, KPMG had 25.9% of the overall market share and the firm CLA had 12.89% of the market. The third largest audit firm had 6.9% of the market share.

Control Variable Measurements

Controls were added to capture the potential effect that professional actuarial firms may have on OPAC. Similar to independent auditors, they fulfill a gatekeeper function and need to adhere to high professional standards of expertise and independence to enter their profession (Gunz et al., 2009). They are expected to adhere to professional standards such as guidelines developed by the Society of Actuary (SOA) and standards adopted by GASB (Chen, 2018). Moreover, some public pension plans use internal actuaries, rather than a third-party professional actuarial firm. To control for the possibility that actuarial firms impose similar effects on OPAC as are hypothesized about auditors, controls were added to capture the use of an internal actuary (INT_ACTUARIES) and actuarial firm industry specialization (SPEC_ACTUARIALFIRM). Plans receiving actuarial valuation services from an internal unit of the government (i.e., an internal actuary) were coded as “1,” otherwise “0.” Moreover, plans were coded as “1” if the actuarial service was provided by Gabriel, Roeder, Smith & Company (GRS) in FY 2014, otherwise “0.” Similar to the operationalization of the audit variable, the identification of specialist actuarial firms was based on the presence of a significant jump in market share. GRS differentiates itself from its competitors in that it holds a 30.8% share of the market, based on the overall size of pension plans to whom actuarial services were provided. By comparison, the second and the third largest actuarial firms had 19.4% and 15.7% of the overall market share.

An indicator variable was also included to capture substantive reform efforts, aimed at improving pension plan funding (REFORM). Such reforms, if substantive, increase the likelihood that a plan moves above the aforementioned “crossover point,” where projected benefit payments for current employees and inactive employees exceed the projected plan net position related to those employees. It is possible that such efforts occurred in conjunction with or in lieu of OPAC. Given this, it is not possible to predict a causal direction. To capture the effect of such changes, evidence of material changes in the overall contribution rate (i.e., employer and employee contributions in relation to overall payroll) across fiscal years 2013 and 2014 were tracked. Plans associated with sponsoring governments were the total contribution rate increased by more than 4% were coded as “1” (indicative of reform), others were coded “0.”

Finally, control variables were included for partial payment of the ARC and plan size. The indicator variable for partial payment of the ARC (ARC_FAIL) was coded as “1” if the sponsoring government failed to pay the full ARC in FY 2014 (a payment of 96% or less), otherwise “0.” Partial payments of the ARC are in breach with fiscal responsibility and in many cases also with state law. It achieves the same goal as OPAC (i.e., it reduces fiscal stress), albeit in a more visible way, given that governments need to report the proportion of ARC paid in their financial reports. This variable can therefore be interpreted as if the sponsoring government cares less about how it is perceived from a fiscal perspective. Governments that do not pay the full ARC can therefore be argued to be less inclined to favor OPAC. At the same time, however, it is possible that governments use a combination of both visible and invisible techniques. Given this, directional clarity is difficult to predict. Pension plan size (PLAN_SIZE) was controlled for using the natural logarithm of plans’ average pension liabilities for FY 2013. In prior literature, many studies have included size as a proxy for political visibility or administrative capacity.

Empirical Model

A complete summary of all the variables included in the resulting models is provided in Table 1. In Model 1, the effects of economic, political, bureaucratic and audit factors on OPAC are estimated as follows:

\begin{array}{l} {OPAC}_{i} = β_{0} + β_{1} SHORT_{NEGATIVE}_{i} \\ + β_{2} LONG_{BOTTOMQUARTILE}_{i} \\ + β_{3} {FUNDING}_{i} + β_{4} {THRESHOLDS}_{i} \\ + β_{5} {POLITICAL}_{i} + β_{6} {UNION}_{i} \\ + β_{7} UNIFIED_{DEM}_{i} \\ + β_{8} COST_{SHARING}_{i} \\ + β_{9} INT_{AUDIT}_{i} + β_{10} NON_{CPA}_{i} \\ + β_{11} SPEC_{FIRM}_{i} + β_{12} INT_{ACTUARIES}_{i} \\ + β_{13} SPEC_{ACTUARIALFIRM}_{i} \\ + β_{14} ARC_{FAIL}_{i} + β_{15} {REFORM}_{i} \\ + β_{16} PLAN_{SIZE}_{i} + ε_{i} \end{array}

(1)

In addition, we develop a slightly revised model (Model 2) to add further clarity into how fiscal conditions and audit characteristics affect OPAC. This model includes variables that capture the overlap between short- and long-term financial conditions, as well as the overlap between internal auditors that are or are not CPAs, and external auditors that are or are not affiliated with an industry specialist audit firm. It should be noted that no new data are introduced in Model 2. The only difference is how observations are being categorized for some of the categorical variables. These alternative categorizations include budget solvency index variables (β₁ and β₂ in both models) and audit variables (β₉, β₁₀, and β₁₁ in both models). In Model 2, the effects of economic, political, bureaucratic, and audit factors on OPAC are estimated as follows:

\begin{array}{l} {OPAC}_{i} = β_{0} + β_{1} NEGATIVE \\ _{BOTTOMQUARTILE}_{i} \\ + β_{2} NEGATIVE_{TOP3QUARTILES}_{i} \\ + β_{3} {FUNDING}_{i} + β_{4} {THRESHOLDS}_{i} \\ + β_{5} {POLITICAL}_{i} + β_{6} {UNION}_{i} \\ + β_{7} UNIFIED_{DEM}_{i} \\ + β_{8} COST_{SHARING}_{i} \\ + β_{9} INTAUDIT_{CPA}_{i} \\ + β_{10} EXTAUDIT_{SPECFIRM}_{i} \\ + β_{11} EXTAUDIT_{NONSPECFIRM}_{i} \\ + β_{12} INT_{ACTUARIES}_{i} \\ + β_{13} SPEC_{ACTUARIALFIRM}_{i} \\ + β_{14} ARC_{FAIL}_{i} + β_{15} {REFORM}_{i} \\ + β_{16} PLAN_{SIZE}_{i} + ε_{i} \end{array}

(2)

Results and Analysis

Descriptive statistics for all variables are presented in Table 2. Scatter dots and outlier statistics indicate that continuous variables are free from extreme values. Moreover, pairwise correlations (see correlation matrix, in Appendix) and the variance inflation factor (VIF; see Table 3) show that there are no evident problems with multicollinearity.

Table 2.

Descriptive Statistics.

Variable	M	SD	Min	Max
OPAC	0.632	0.485	0 (no)	1 (yes)
SHORT_NEGATIVE	0.781	0.416	0 (no)	1 (yes)
LONG_BOTTOMQUARTILE	0.246	0.432	0 (no)	1 (yes)
NEGATIVE_BOTTOMQUARTILE	0.246	0.432	0 (no)	1 (yes)
NEGATIVE_TOP3QUARTILES	0.535	0.501	0 (no)	1 (yes)
POSITIVE_BOTTOMQUARTILE	0.000	0.000	0 (no)	1 (yes)
POSITIVE_TOP3QUARTILES	0.219	0.416	0 (no)	1 (yes)
FUNDING	0.723	0.161	0.258	1.146
THRESHOLDS	0.219	0.416	0 (no)	1 (yes)
POLITICAL	0.496	0.338	0.000	1.000
UNION	0.327	0.177	0.058	0.673
UNIFIED_DEM	0.237	0.427	0 (no)	1 (yes)
COST_SHARING	0.816	0.389	0 (no)	1 (yes)
INT_AUDIT	0.254	0.437	0 (no)	1 (yes)
NON_CPA	0.140	0.349	0 (no)	1 (yes)
INTAUDIT_NONCPA	0.140	0.349	0 (no)	1 (yes)
INTAUDIT_CPA	0.114	0.319	0 (no)	1 (yes)
SPEC_FIRM	0.325	0.470	0 (no)	1 (yes)
EXTAUDIT_SPECFIRM	0.325	0.470	0 (no)	1 (yes)
EXTAUDIT_NONSPECFIRM	0.421	0.496	0 (no)	1 (yes)
INT_ACTUARIES	0.114	0.319	0 (no)	1 (yes)
SPEC_ACTUARIALFIRM	0.298	0.460	0 (no)	1 (yes)
ARC_FAIL	0.342	0.477	0 (no)	1 (yes)
REFORM	0.184	0.389	0 (no)	1 (yes)
PLAN_SIZE	7.204	0.519	6.061	8.574

Note. Statistics for all variables are based on 114 observations. For variable definition, see Table 1.

Table 3.

Logistic Regression Models.

	(1)			(2)
Variable name	Odds ratio	SE	p > z	Odds ratio	SE	p > z
SHORT_NEGATIVE	2.032	1.611	0.371
LONG_BOTTOMQUARTILE	0.131	0.138	0.054
NEGATIVE_BOTTOMQUARTILE				0.265	0.299	0.239
NEGATIVE_TOP3QUARTILES				2.032	1.611	0.371
FUNDING	0.002	0.004	0.003	0.002	0.004	0.003
THRESHOLDS	3.143	2.163	0.096	3.143	2.163	0.096
POLITICAL	12.242	12.045	0.011	12.242	12.045	0.011
UNION	257.088	605.240	0.018	257.088	605.240	0.018
UNIFIED_DEM	0.179	0.134	0.022	0.179	0.134	0.022
COST_SHARING	5.782	4.347	0.020	5.782	4.347	0.020
INT_AUDIT	0.111	0.115	0.033
NON_CPA	63.339	92.539	0.005
INTAUDIT_CPA				0.016	0.023	0.005
SPEC_FIRM	0.313	0.203	0.073
EXTAUDIT_SPECFIRM				0.044	0.053	0.010
EXTAUDIT_NONSPECFIRM				0.142	0.162	0.088
INT_ACTUARIES	0.703	0.720	0.731	0.703	0.720	0.731
SPEC_ACTUARIALFIRM	12.423	10.529	0.003	12.423	10.529	0.003
ARC_FAIL	0.470	0.345	0.304	0.470	0.345	0.304
REFORM	0.249	0.183	0.058	0.249	0.183	0.058
PLAN_SIZE	1.788	0.958	0.278	1.788	0.958	0.278
Constant	0.063	0.267	0.512	0.447	1.947	0.853
LR Chi²	56.10 (<0.001)			56.10 (<0.001)
Pseudo R²	0.374			0.374
Observations	114			114
Highest/mean VIF	2.55/1.61			3.11 /1.78

Note. VIF = variance inflation factor. For short- and long-term budget solvency index variables, POSITIVE_TOP3QUARTILES was used as reference variable, and for audit variables INTAUDIT_NONCPA was used as reference variable. Coefficients in bold are significant at the .05 level or less. For variable definition, see Table 1.

The regression results are presented in Table 3. Both Models 1 and 2 are significant (p < .001) with a Pseudo R² of 37.4%. As discussed above, differences between the two models are confined to how observations are being categorized in regards to the budget solvency indexes and audit characteristics. As a result of the different categorical variables used to estimate OPAC, there are some minor differences between the VIF values in Model 1 and 2. All other results produced by Model 1 and 2 remain similar.

In regards to economic conditions (i.e., H1, H2, H3, and H4), the regression results indicate that only one of the four variables, funding level, is statistically significant. Holding other factors constant, the results indicate that pension plans with lower funding ratios are more likely to engage in OPAC (odds ratio = 0.002). This finding is consistent with prior research, reported above, and underscores that actions aimed at avoiding the adoption of a blended rate are more significant for governments that sponsor plans that are substantially underfunded. The extent to which the discount rate has to be reduced is higher for such plans, given that more weight is placed on the high-quality municipal bond rate, when the blended rate is calculated. The regression results, thus, offer empirical support for H3.

Regardless of whether they are factually correct or not, commonly accepted perceptions about factors affecting credit ratings can help to predict and explain OPAC behavior. The regression results indicate that pension plans operating in close proximity to benchmark funding thresholds were more likely to engage in OPAC (odds ratio = 3.143), compared with plans that operate further away from such thresholds. However, it is important to note that this variable resides outside the limits of the 95% confidence interval and is only moderately significant at the 10% level (p = .096). H4 must therefore be rejected.

The regression results pertaining to fiscal constraints offer mixed signals. On one hand, the results suggest that short-term budget solvency, operationalized based on a government’s inability to meet its fiscal year obligations, is positively related to OPAC (odds ratio = 2.032). That is, pension plans associated with sponsoring governments that were having a difficult time meeting their fiscal obligations in 2013 were more likely to engage in OPAC. This is consistent with earlier research and the theoretical proposition that OPAC is used to free up budget resources. On the contrary, the variable long-run budget solvency is negatively related to OPAC (odds ratio = 0.131), which contradicts the theoretical proposition outlined above. While neither of the two variables are statistically significant, it is notable that the variable long-run budget solvency resides just outside the limits of the 95% confidence interval (p = .054). Considering these results, both H1 and H2 must be rejected.

Additional insights are gained into the relationship between a sponsoring government’s fiscal constraints and OPAC (see Model 2), when the overlap between restrictions relating to short- and long-term budget constraints is considered. When sponsoring governments that had a negative budget solvency index for FY 2013 are examined in combination with governments that are in the bottom quartile in terms of long-term budget solvency (i.e., NEGATIVE_ BOTTOMQUARTILE), the results indicate a lower likelihood of OPAC (odds ratio = 0.265), compared with the reference category POSITIVE_TOP3QUARTILES. Moreover, when sponsoring governments that had a negative budget solvency index in FY 2013, are examined in combination with being in the top three quartiles in terms of long-term budget solvency (i.e., NEGATIVE_ TOP3QUARTILES), the results show a higher likelihood of OPAC (odds ratio = 2.032), compared with the reference category. Although neither of the variables are significant, they highlight that sponsoring governments, facing both short- and long-term budget restrictions, may not be behaving in accordance with our theoretical predictions. One possible explanation for this is that OPAC is perceived to be insufficient in terms of signaling fiscal stability. As shown by Stalebrink (2007), governments that experience substantial levels of fiscal constraints may sometimes revert to a form of strategic accounting choice behavior that, in the short term, is contradictory to improving financial performance. Rather, they save accounting flexibility to future reporting periods when OPAC is likely to be more consequential.

All political and bureaucratic factors are statistically significant. Holding other factors constant, both the proportion of political appointees serving on a fiduciary body (odds ratio = 12.424) and the degree of unionization (odds ratio = 257.088) significantly increases the likelihood of OPAC. These results, thus, offer further clarity into the effects and association between a board’s political embeddedness and OPAC on one hand (H5), and unionization and OPAC on the other (H6). As noted above, earlier empirical findings have been mixed in regard to the effects of these two factors.

The result also offers clarity into the relative strength and influence that a unified democratic party has on OPAC (odds ratio = 0.179). Contrary to our prediction, the results indicate that such political conditions significantly decrease the likelihood of OPAC (H7). This suggests that a unified democratic government might be more prone to govern responsibly (i.e., without OPAC). Finally, cost-sharing and multiple employer plans (odds ratio = 5.782) are significantly more likely to engage in OPAC, thus supporting H8.

Audit characteristics also affect the likelihood of OPAC. Holding other factors constant, the regression results in Model 1 indicate that pension funds audited by non-CPA auditors were significantly more likely to engage in OPAC (odds ratio = 63.339). Contrary to our prediction, the results in Model 1 also indicate that pension funds audited by internal auditors were significantly less likely to engage in OPAC (odds ratio = 0.111). Although the expected negative association between industry specialist audit firm and OPAC is correct (odds ratio = 0.313), this result is only moderately significant at the 10% level (p = .073).

The overlap between internal auditors that are non-CPAs/CPAs and external auditors that are/are not affiliated with an industry specialist audit firm is shown in Model 2. In this model, internal auditors without a CPA are used as the reference category, given that pension plans audited by such auditors are theoretically predicted to have the highest likelihood of engaging in OPAC. Compared with the reference category, all other audit combinations decrease the likelihood of OPAC. Only internal auditors holding a CPA (odds ratio = 0.016) and external auditors affiliated with an industry specialist audit firm (odds ratio = 0.044) are significant. Given these findings, H9 was rejected, whereas H10 and H11 were supported.

One of the control variables, actuarial firm specialization (odds ratio = 12.423), was statistically significant. This result indicates that actuarial gatekeepers might be affecting OPAC differently than audit gatekeepers. Specifically, the results suggest that the risk of OPAC increases when the pension plan relies on a more specialized actuarial firm. This is consistent with Vermeer et al.’s (2010) idea that more specialized firms are more capable of aiding pension plans in executing OPAC decisions. It is also consistent with assertions made by several scholars who argue that sponsoring governments and pension plans have incentives to select “team players” that enable the selection of accounting policies and actuarial assumptions that minimize both the UAL and the ARC (Hess, 2005; Hess & Squire, 2009; Stalebrink, 2014). Finally, it is notable that governments and pension plans appear to have adopted changes to their funding policies in lieu of adopting a blended rate. Although it is not statistically significant, the variable, reform (odds ratio = 0.249), resides just outside the 95% confidence interval (p = .058).

The novel method used to develop the dependent variable in this study, makes it relatively challenging to compare the above results with specific results generated in prior research. The most relevant point of comparison would be the study by Vermeer et al. (2010), which employ a categorical dependent variable, in a logistic regression model that produced a Pseudo R² of 41.0%. Using this study as a benchmark, our models generated slightly lower explanatory power but was, nonetheless, still on a level indicating a good model fit. It is, however, important to underline that a straightforward pseudo R² comparison typically only is meaningful when the results have been generated from similar data. Obviously, this is not the case, because Vermeer et al.’s (2010) study was conducted in the context of local-administered pension plans from two states, including Michigan and Pennsylvania, not in the context of state-administered pension plans which is the scope of this study.

Summary and Conclusion

This study examined the role of economic, political, bureaucratic, and audit factors in OPAC, by exploiting the implementation of GASB 67 and GASB 68 as an exogenous shock that temporarily increased the incentives for elected officials to engage in OPAC. The results of the analysis show that state-administered pension plans are more likely to engage in OPAC when the plan is underfunded, organized as a cost-sharing plan, governed by a politically embedded fiduciary body, and when the sponsoring government is surrounded by a high degree of unionization, and is divided in terms of partisan control.

The results also show that CPAs can play an important role in combatting OPAC. Specifically, they show that pension plans that are subject to an audit by a CPA are less likely to engage in OPAC. Contrary to expectations, the results also indicate that the likelihood of OPAC decreases when pension plans are audited by an auditor that is internal to the government (e.g., a state auditor), compared with when the auditor is affiliated with a third-party professional auditing firm. However, compared with internal auditors that are not CPAs, auditors affiliated with firms that exhibit a higher degree of specialization decrease the likelihood of OPAC. This was also the case for CPAs that were affiliated with an internal unit of the government. The results also indicated that specialized actuarial firms might serve to enable OPAC, rather than combatting it.

These findings are important for several reasons. First, they are important for purposes of diagnosing and finding solutions to the fiscal challenges that many state governments face. These challenges were exacerbated by the great recession (Clark, 2015; Gorina et al., 2018), and goes well beyond the U.S. context (e.g., Barbera et al., 2017; Cepiku et al., 2016; García-Sánchez et al., 2012). As a result, the financial sustainability of subnational governments has become an increasingly prioritized item on the agenda of policy makers, practitioners and public administration scholars alike (Bolívar et al., 2018, 2019; Caruana et al., 2019; Navarro-Galera et al., 2017). Due to its substantial impact on the size reported pension liabilities and the ARC, the selection of an artificially high discount rate is often mentioned as a culprit of the problem (Wang & Peng, 2018). It can jeopardize a pension system’s financial health, and at worst, the financial health of an entire local or state government (Stalebrink, 2014; Vermeer et al., 2010; Wang & Peng, 2018).

The findings are also important from an accounting standard setting perspective. In the absence of competitive markets that discipline entities to make prudent and transparent accounting choices, it is logical for governmental accounting standards setting authorities, such as GASB, to respond to identified imperfections by adopting new and revised accounting standards that puts additional pressure on state and local governments to improve their financial reporting (cf., Chan, 2003; Stalebrink, 2014). However, it should not be assumed that the adoption of common accounting standards (i.e., de jure harmonization) automatically result in compliance with those standards (i.e., de facto harmonization; Tay & Parker, 1990). Rather, as our analysis, based on the politico-economic framework (e.g., Cheng, 1992; Zimmerman, 1977) suggests, the outcome of accounting choice is determined by a multitude of contextual factors. This illustrates that the intent of improving communication by allowing professional judgment to enter the accounting measurement process is associated with a risk of opportunism. From an accounting standards setting perspective, our findings suggest that it is critical to understand not only how contextual factors affect the workings of standards that rely heavily on professional judgment, but also whether and how professional gatekeepers can serve to offset misuse of the discretion. In the U.S. pension accounting and reporting context, GASB have pursued a route that relies relatively heavily on discretion to improve the quality of externally reported pension information, as is exemplified by the leeway the reporting entity enjoys when selecting the discount rate. The other major public sector accounting standards setting authority, the International Public Sector Accounting Standards Board (IPSASB), extends less discretion in this regard. In essence, it limits the reporting entity to selecting a rate on the basis of market yields on government bonds and high-quality corporate bonds. Given the large number and the diversity of the governments that are targets for the standards issued by the IPSASB, allowing for more limited discretion appears to be good practice, given the findings of this article.

Finally, while much uncertainty remains pertaining to the role of gatekeepers, the findings suggest that it might be fruitful to explore the adoption of strong recommendations relating to the involvement of external qualified actuaries and auditors when significant measurement discretion is extended to a reporting entity. An example of this is IPSAS 39. It includes language encouraging reporting entities to involve a qualified actuary in the measurement of material post-employment benefit obligations. It is, however, important to note that the extent to which these findings can be applied are highly dependent on their own unique contexts. It is also important to note that the manner by which the dependent variable was operationalized prevents insights into the extent to which the adjustments made by pension plans were sufficient. The dichotomous nature of the dependent variable does not distinguish between pension plans that made adjustments that were smaller than required. Hence, it is possible that the operationalization of the dependent variable understates the significance of the problem.

Footnotes

Appendix

Appendix.

Correlation Matrix.

Variables		OPAC	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.	13.	14.	15.
1.	SHORT_NEGATIVE	–.097
2.	LONG_BOTTOMQUARTILE	−.198	.302
3.	FUNDING	−.237	–.127	–.087
4.	THRESHOLDS	.053	–.180	–.155	.013
5.	POLITICAL	.190	−.248	–.001	.171	.028
6.	UNION	–.026	.185	.572	–.004	–.120	.086
7.	UNIFIED_DEM	–.045	.046	.353	–.161	–.046	–.017	.386
8.	COST_SHARING	.153	.022	.008	.110	–.076	–.006	–.056	–.055
9.	INT_AUDIT	.070	−.226	–.099	.139	.031	.120	.030	.101	–.034
10.	NON_CPA	.204	–.091	.063	–.045	–.092	.083	.159	.310	–.069	.692
11.	SPEC_FIRM	–.131	.096	.127	.015	–.005	.068	.180	–.034	.088	−.405	−0.280
12.	INT_ACTUARIES	−.241	.190	.436	.316	–.123	–.098	.285	–.005	.171	–.083	0.014	.164
13.	SPEC_ACTUARIALFIRM	.259	−.257	–.105	–.071	.025	–.026	–.155	.043	–.086	.192	0.178	–.165	−.234
14.	ARC_FAIL	–.101	–.109	.233	−.222	.154	–.148	.085	.251	−.277	.131	0.188	–.066	–.142	.136
15.	REFORM	–.059	–.076	.149	–.106	.076	.120	.232	.001	−.241	.086	0.199	–.039	.043	.086	.134
16.	PLAN_SIZE	–.002	.091	.312	–.112	–.008	.114	.174	.101	–.014	.052	0.067	.167	.030	–.043	.241	.068

Note. Pearson correlations is presented for all variables included in Model 1. Correlations in bold are significant at .05 level (two-tailed) or less. For variable definition, see Table 1.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a research forum grant awarded by the International Public Sector Accounting Standards Board (IPSASB).

ORCID iD

Odd J. Stalebrink

Notes

Author Biographies

Odd J. Stalebrink is an associate professor of Public Administration and Policy at the School of Public Affairs, Pennsylvania State University-Harrisburg. His research centers on public sector accounting, and governance and management of public pensions funds.

Pierre Donatella is a senior lecturer of Public Administration at the School of Public Administration, University of Gothenburg, Sweden. He received his PhD in Business Administration from the School of Business, Economics and Law, University of Gothenburg, Sweden, in 2016. His research focuses on public sector accounting and auditing.

References

Anzia

Moe

(2017). Polarization and policy: The politics of public-sector pensions. Legislative Studies Quarterly, 42(1), 33–62.

Arcas

M. J.

Martí

(2016). Financial performance adjustment in english local governments. Australian Accounting Review, 26(2), 141–152.

Baber

W. R.

(1983). Toward understanding the role of auditing in the public sector. Journal of Accounting and Economics, 5, 213–227.

Barbera

Jones

Korac

Saliterer

Steccolini

(2017). Governmental financial resilience under austerity in Austria, England and Italy: How do local governments cope with financial shocks? Public Administration, 95(3), 670–697.

Bolívar

M. P. R.

Galera

A. N.

Subirés

M. D. L.

Muñoz

L. A.

(2018). Analysing the accounting measurement of financial sustainability in local governments through political factors. Accounting, Auditing & Accountability Journal, 31(8), 2135–2164.

Bolívar

M. P. R.

Subires

M. D. L.

Muñoz

L. A.

Galera

A. N.

(2019). The financial sustainability of local authorities in England and Spain: A comparative empirical study. International Review of Administrative Sciences 87(1), 97–114. https://doi.org/10.1177/0020852319834721

Brainard

Zorn

(2012, January). What is the source of the 80-percent threshold as a healthy or minimum funding level for public pension plans? [Unpublished paper]. National Association of State Retirement Administrators (NASRA). https://www.nasra.org/files/Topical%20Reports/Funding%20Policies/80_percent_funding_threshold.pdf

Brooks

(2019). Board on the job: Public-pension governance in the United States (US) States. Journal of Public Policy, 39(1), 1–34.

Caruana

Brusca

Caperchione

Cohen

Rossi

F. M.

(2019). Exploring the relevance of accounting frameworks in the pursuit of financial sustainability of public sector entities: A holistic approach. In Caruana, J., Brusca, I., Caperchione, E., Cohen, S., & Rossi, F. M. (Eds.), Financial sustainability of public sector entities (pp. 1–18). Palgrave Macmillan.

10.

Center for Retirement Research at Boston College. (2018). Public plans database (2001–2018).

11.

Cepiku

Mussari

Giordano

(2016). Local governments managing austerity: Approaches, determinants and impact. Public Administration, 94(1), 223–243.

12.

Chan

J. L.

(2003). Government accounting: An assessment of theory, purposes and standards. Public Money & Management, 23(1), 13–20.

13.

Chaney

B. A.

Copley

P. A.

Stone

M. S.

(2002). The effect of fiscal stress and balanced budget requirements on the funding and measurement of state pension obligations. Journal of Accounting and Public Policy, 21, 287–313.

14.

Chapman

(2008). State and local fiscal sustainability: The challenges. Public Administration Review, 68, S115–S131.

15.

Chen

(2018). Understanding decisions in state pension systems: A system framework. American Review of Public Administration, 48(3), 260–273.

16.

Cheng

R. H.

(1992). An empirical analysis of theories on factors influencing state government accounting disclosure. Journal of Accounting and Public Policy, 11(1), 1–42.

17.

Clark

B. Y.

(2015). Evaluating the validity and reliability of the financial condition index for local governments. Public Budgeting & Finance, 35(2), 66–88.

18.

Coggburn

Kearney

(2010). Trouble keeping promises? An analysis of underfunding in state retiree benefits. Public Administration Review, 70(1), 97–108.

19.

Cohen

Bisogno

Malkogianni

(2019). Earnings management in local governments: The role of political factors. Journal of Applied Accounting Research, 20(3), 331–348.

20.

Dechow

Schrand

(2010). Understanding earnings quality: A review of the proxies, their determinants and their consequences. Journal of Accounting and Economics, 50(2–3), 344–401.

21.

DeFond

Zhang

(2014). A review of archival auditing research. Journal of Accounting and Economics, 58(2–3), 275–326.

22.

Deis

D. R.

Giroux

G. A.

(1992). Determinants of audit quality in the public sector. The Accounting Review, 67(3), 462–479.

23.

Donatella

(2020). Is political competition a driver of financial performance adjustments? An examination of Swedish municipalities. Public Money & Management, 40(2), 122–130.

24.

Downs

Cohen

Hanna

Fuerst

Goldfarb

(2012, July). The 80% pension funding standard myth [Issue Brief]. American Academy of Actuaries.

25.

Eaton

T. V.

Nofsinger

J. R.

(2004). The effect of financial constraints and political pressure on the management of public pension plans. Journal of Accounting and Public Policy, 23(3), 161–189.

26.

Elder

R. J.

Lowensohn

Reck

J. L.

(2015). Audit firm rotation, auditor specialization, and audit quality in the municipal audit context. Journal of Government & Nonprofit Accounting, 4(1), 73–100.

27.

Ferreira

Carvalho

Pinho

(2013). Earnings management around zero: A motivation to local politician signaling competence. Public Management Review, 15(5), 657–686.

28.

Frandsen

B. R.

(2016). The effects of collective bargaining rights on public employee compensation: Evidence from teachers, firefighters, and police. Industrial and Labor Relations Review, 69(1), 84–112.

29.

García-Sánchez

I. M.

Cuadrado-Ballesteros

Frías-Aceituno

J. V.

Mordan

(2012). A new predictor of local financial distress. International Journal of Public Administration, 35(11), 739–748.

30.

Garven

S. A.

Beck

A. W.

Parsons

L. M.

(2018). Are audit-related factors associated with financial reporting quality in nonprofit organizations? Auditing: A Journal of Practice & Theory, 37(1), 49–68.

31.

GASB. (2013). GASB’s new pension standards: Setting the record straight. https://www.gasb.org/jsp/GASB/Page/GASBSectionPage&cid=1176163527830

32.

Giroux

Jones

(2011). Measuring audit quality of local governments in England and Wales. Research in Accounting Regulation, 23(1), 60–66.

33.

Gorina

Maher

Joffe

(2018). Local fiscal distress: Measurement and prediction. Public Budgeting & Finance, 38(1), 72–94.

34.

Greenwood

M. J.

Baylis

R. M.

Tao

(2017). Regulatory incentives and financial reporting quality in public healthcare organizations. Accounting and Business Research, 47(7), 831–855.

35.

Gunz

McCutcheon

Reynolds

(2009). Conflict of interest and the actuarial profession. Journal of Business Ethics, 89(1), 77–89.

36.

Healy

P. M.

Wahlen

J. M.

(1999). A review of the earnings management literature and its implications for standard setting. Accounting Horizons, 13, 365–383.

37.

Hendrick

(2004). Assessing and measuring the fiscal health of local governments: Focus on Chicago suburban municipalities. Urban Affairs Review, 40(1), 78–114.

38.

Hess

(2005). Protecting and politicizing public pension fund assets: Empirical evidence on the effects of governance structures and practices. University of California Davis Law Review, 39, 187–224.

39.

Hess

F. M.

Squire

J. P.

(2009, November). The false promise of public pensions. Policy Review. https://www.hoover.org/research/false-promise-public-pensions

40.

Hoang

Goodman

(2018). Public pensions and collective bargaining rights: Evidence form state and local governments. Public Administration Review, 78(5), 772–784.

41.

Inklebarger

(2011, June 30). Colorado, Minnesota COLA suits thrown out. Pensions & Investments. http://www.pionline.com/article/20110630/DAILYREG/110639983

42.

Jochim

A. E.

Jones

B. D.

(2012). Politics in a polarized congress. Political Research Quarterly, 66(2), 352–369.

43.

Kearney

R. C.

(2003). The determinants of state employee compensation. Review of Public Personnel Administration, 23(4), 305–322.

44.

Kloha

Weissert

C. S.

Kleine

(2005). Developing and testing a composite model to predict local fiscal distress. Public Administration Review, 65(3), 313–323.

45.

Levine

C. H.

Rubin

Wolohojian

G. G.

(1981). The politics of retrenchment: How local governments manage fiscal stress. Beverly Hills, CA: SAGE.

46.

Lowensohn

Johnson

L. E.

Elder

R. J.

Davies

S. P.

(2007). Auditor specialization, perceived audit quality, and audit fees in the local government audit market. Journal of Accounting and Public Policy, 26(6), 705–732.

47.

Mitchell

O. S.

Smith

R. S.

(1994). Pension funding in the public sector. The Review of Economics and Statistics, 76(2), 278–290.

48.

Mortimer

J. W.

Henderson

L. R.

(2014). Measuring pension liabilities under GASB statement no. 68. Accounting Horizons, 28(3), 421–454.

49.

Munnell

A. H.

Aubry

J.-P.

(2016). The funding of state and local pensions: 2015–2020. No. 50: Center for Retirement Research at Boston College.

50.

Munnell

A. H.

Aubry

J.-P.

Hurwitz

Quinby

(2011). Unions and public pension benefits. No. 19: Center for Retirement Research at Boston College.

51.

Navarro-Galera

Lara-Rubio

Buendía-Carrillo

Rayo-Canton

(2017). What can increase the default risk in local governments? International Review of Administrative Sciences, 83(2), 397–419.

52.

Norcross

(2015). State Fiscal Rankings. https://www.mercatus.org/statefiscalrankings

53.

Oates

W. E.

(1985). Searching for leviathan: An empirical study. The American Economic Review, 75(4), 748–757.

54.

Peng

(2004). Public pension funds and operating budgets: A tale of three states. Public Budgeting and Finance, 24(2), 59–73.

55.

Poterba

J. M.

(1994). State responses to fiscal crises: The effects of budgetary institutions and politics. Journal of Political Economy, 102(4), 799–821.

56.

Prosek

Rodriguez

(2019). Wilshire report on state retirement systems: Funding levels and asset allocation. Wilshire Associates Incorporated.

57.

Reichelt

K. J.

Wang

(2010). National and office-specific measures of auditor industry expertise and effects on audit quality. Journal of Accounting Research, 48(3), 647–686.

58.

Snell

(2009). State pensions and retirement legislation 2009. http://www.ncsl.org/issues-research/labor/2009-enacted-pension-legislation.aspx

59.

Snell

(2010). Pensions and retirement plan enactments in 2010 state legislatures. http://www.ncsl.org/issues-research/labor/2010-pension-and-retirement-enacted-legislation.aspx

60.

Stalebrink

O. J.

(2007). An investigation of discretionary accruals and surplus-deficit management: Evidence from Swedish municipalities. Financial Accountability & Management, 23(4), 441–457.

61.

Stalebrink

O. J.

(2014). Public pension funds and assumed rates of return: An empirical examination of public sector defined benefit pension plans. American Review of Public Administration, 44(1), 92–111.

62.

Tay

J. S.

Parker

R. H.

(1990). Measuring international harmonization and standardization. Abacus, 26(1), 71–88.

63.

Vermeer

T. E.

Styles

A. K.

Patton

T. K.

(2010). Are local governments adopting optimistic actuarial methods and assumptions for defined benefit plans? Journal of Public Budgeting, Accounting & Financial Management, 22(4), 511–542.

64.

Wang

Peng

(2018). Political embeddedness of public pension governance: An event history analysis of discount rate changes. Public Administration Review, 78(5), 785–794.

65.

Wang

Dennis

Y. S.

(2007). Measuring financial condition: A study of U.S. States. Public Budgeting & Finance, 27(2), 1–21.

66.

Weinberg

Norcross

(2017). GASB 67 and GASB 68: What the new accounting standards mean for public pension reporting. State and Local Policy. https://www.mercatus.org/publications/gasb-67-68-public-pension-reporting.

67.

Winkelvoss

(1993). Pension mathematics with numerical illustrations (2nd ed.). University of Pennsylvania Press.

68.

Zimmerman

(1977). The municipal accounting maze: An analysis of political incentive. Journal of Accounting Research, 15(2), 107–144.