Managing the Conflicting Interests of Workers and Shareholders: Evidence from Pension-Assumption Manipulations

Abstract

Whereas research on corporate governance typically attends to the conflicting interests between shareholders and executives, in practice executives must frequently adjudicate the demands of multiple stakeholders. To investigate how executives cope with the divergent interests of workers and shareholders, the author examines how much firms claim they will earn on the assets in their defined benefit (DB) pension plans. In a DB arrangement, employees forgo wages in the present in order to receive postretirement income, and they rely on executives to properly fund and manage plan assets. Executives, however, can increase the amount they expect the firm to earn on plan assets, which increases firm earnings in the current period but may undermine workers’ retirement security if expectations do not match actual returns over time. The author shows that the influence and interests of employees and shareholders as well as the decision-making schemas of the CEO affect whether executives exercise this discretion.

Keywords

corporate governance earnings management human resource practices ownership power retirement plans shareholder value

Scholars have long recognized that executives often confront conflicting demands from various groups making claims on the firm’s resources (e.g., Cyert and March 1963). An important dimension of this claims-making process is that each stakeholder group has its own interests that span different time horizons, allowing the firm to make promises with various timelines based on each group’s preferences (Pfeffer and Salancik 1978). One way firms do so is by transferring obligations from the present to the future. Accomplishing this conveyance across periods requires the firm to anticipate future states based on imperfect present information. When predictions about future states prove incorrect or when past promises conflict with influential stakeholders’ current expectations, keeping such promises can be challenging.

These conflicts frequently appear in the context of employment relationships, as the interests and time horizons of workers frequently differ from those of other stakeholders, such as investors. In fact, recent studies have focused on the role of financial market pressures in general and shareholders specifically in the reconfiguration of employment relationships in the United States (e.g., Appelbaum and Batt 2014; Jung 2015; Lin 2016; Cobb and Lin 2017), highlighting the often conflicting interests of these two stakeholder groups. We should expect particularly heightened conflicts between workers and shareholders when firms offer deferred compensation to employees. In such arrangements, employers must reconcile explicit promises for future compensation for work in the present with implicit promises of long-term employment (Lazear 1986), which frequently contradicts financial market pressures to minimize longer-term investments in workers and emphasize flexibility for firms (Cappelli 1999; Jacoby 2005).

This article addresses a critical question regarding how firms manage these conflicting demands: When executives must decide between the interests of two stakeholders with conflicting interests, which factors motivate them to favor one group over the other? I examine this question by focusing on how executives manage the financing of their defined benefit (DB) pension plans. Effectively, DB pension funds are workers’ deferred earnings; therefore, we can consider DB participants to have a claim on pension assets (Kruse 1995). Moreover, because firms must reserve enough funds to cover the future liabilities owed to their employees, workers must entrust executives to properly finance and manage DB plans in exchange for work done in the present.

Yet, many executives face pressures from investors to maximize near-term earnings (Bushee 2001); hence, financing long-term liabilities to workers with current earnings creates a potential conflict between the interests of workers and of shareholders. Executives can negotiate these conflicting interests through management of pension-fund financing. If executives make aggressive assumptions about the expected rate of return (ERR) on pension-fund investments, they can boost current earnings by pushing out pension costs into the future. Perhaps not surprisingly, scholars commonly refer to the manipulation of ERRs as a form of earnings management. Such activity may compromise the long-term health of the firm’s pension fund if expectations do not match actual returns over time, which undermines workers’ retirement security (Waring 2012).

When executives increase the ERR on a pension fund, the funds intended for the DB plan may be transferred to current reported earnings. Hence, this is an appealing context for assessing stakeholder conflicts, as executives have the discretion to allocate the firm’s rewards to two stakeholder groups—employees and investors—who differ in interests, sources of influence, and time horizons. Incorporating insights from corporate governance (e.g., Davis 2005), sociopolitical theories of the firm (e.g., Cyert and March 1963), and power-dependence perspectives (e.g., Pfeffer and Salancik 1978), I develop a theory of how executives adjudicate two stakeholder groups with conflicting demands on firms’ resources. My core argument is that because complex organizations have multiple goals that often diverge, which goals executives prioritize depends, in part, on stakeholder influence. Because executives have discretion in setting the ERR, I expect that their willingness to boost earnings by increasing ERR is guided by the influence and interests of employees and shareholders and by the decision-making schema of the firm’s chief executive officer (CEO).

Through a set of detailed analyses of a sample of large-scale US employers from 1992 to 2006, I examine the effects of union influence, shareholders with varying fiduciary duties, and CEO functional backgrounds on the propensity for firms to manipulate their ERRs. In so doing, I provide a more complete depiction of the corporate-governance process, whereby employees are also principals, shareholders have diverse interests, and executives rely on their own decision-making schemas to adjudicate the two groups. This study further elucidates the effects of labor power and firms’ financial strategies on employment practices, examining the role of organized labor in protecting retirement security for its members and for non-unionized workers as well. Moreover, industrial relations researchers have increasingly attended to the ways in which value accumulation and extraction operate under financial capitalism (e.g., Jacoby 2011; Appelbaum, Batt, and Clark 2013). By focusing more closely on how financial actors influence firms’ willingness to redistribute value from one stakeholder group to another, this study clarifies the ways in which finance increases the likelihood that firms will rewrite implicit contracts between firms and workers.

Corporate Governance and Multiple Principals

Corporate governance concerns “the structure of rights and responsibilities among the parties with a stake in the firm” (Aoki 2001: 11). In practice, corporate governance involves balancing the interests of various stakeholders, including shareholders, management, and labor (Aguilera et al. 2008), yet most governance research focuses primarily on the methods shareholders use to discipline management. The core problem motivating most of this work is that because the actions and interests of executives often conflict with those of shareholders, shareholders rely on various control mechanisms to monitor and motivate executives to meet their demands (Jensen and Meckling 1976). While greatly informing our understanding of governance problems and potential solutions, an agency-theoretic focus may overlook certain issues confronting executive decision-makers. In particular, because of its focus on the bilateral contracts between shareholders and executives, agency theory typically overlooks other stakeholder claims on the firm’s resources. In many firms, executives must balance the demands of multiple claimants (Asher, Mahoney, and Mahoney 2005). Goal conflicts, therefore, occur not only between executives and shareholders but can also arise when an executive must “maneuver through the tangled loyalties he or she owes to many different principals . . . to negotiate through their competing interests” (Shapiro 2005: 278).

By contrast, management scholarship has long advanced the idea that executives often face numerous conflicting demands from stakeholders (e.g., Cyert and March 1963). A negotiation process that occurs between coalitions, comprising groups with similar interests seeking to have the firm meet their preferences, helps to determine a firm’s strategies. Coalitions often have divergent interests, and typically no single group determines the organization’s goals. The power of the actors involved and the legitimacy and urgency of their demands determine, in part, which issues executives attend to and whose interests they prioritize (Mitchell, Agle, and Wood 1997).

Pension Assumptions, Earnings Management, and Conflicting Stakeholder Interests

Firms structure their DB pensions to provide each participant a fixed, annuitized source of postretirement income. The firm commits to pay the cost of these promised benefits and, thus, incurs a liability equal to the present value of all future payments owed to its workers. The firm funds this liability with pension assets, investing them in financial vehicles such as debt, equities, and insurance assets.

Three main calculations determine the annual cost of DB plans: the first is service cost, which represents the increase in the pension obligation for workers who have accrued another year of service; second, firms calculate interest cost, which is the interest accrued on the pension liability arising from the passage of time;¹ the last factor is the ERR on plan assets, which estimates how much the firm’s pension assets will earn. Accounting rules allow for the use of the expected, as opposed to the actual, rate of return, to insulate firm earnings from year-to-year market swings. Under Accounting Standards Codification 715, employers have discretion in choosing their ERRs.² Over time, firms must reconcile the difference between the expected and actual returns, but this occurs over long amortization periods.³ By increasing its ERR, an employer can decrease reported pension expenses and shift its obligations to its employees further into the future. Doing so allows an executive to boost a firm’s market value, as evidence shows that earnings associated with changed pension assumptions are capitalized into the share price in a manner similar to operating earnings (Picconi 2006).

To illustrate the impact of ERR increases on firm earnings, consider the example of IBM Inc. from Bergstresser, Desai, and Rauh (2006). In 1997, the firm increased its ERR from 9.25% to 9.5%, which accounted for approximately 1.5% of its pretax income between 1997 and 1999. Despite poor equity market returns and declining bond yields, IBM raised the ERR again to 10% in 2000. As a result, its pretax income in 2000 and 2001 was nearly 5% greater than it would have been with an ERR of 9.25%. Between 1995 and 2001, IBM’s pretax income grew at an annual compound rate of 6.7%. Were it not for the two ERR increases, it would have grown at a rate of 5.6%.

To be clear, when a company states that it expects to earn a 10% return on its pension assets, it is claiming that, on average over time, the funds it sets aside and invests to meet its pension liabilities will earn 10% in returns. If we account for dividends and inflation, between 1870 and 2010 the S&P 500’s real return averaged approximately 6.5% per year. Moreover, given that firms invest large portions of their pension assets in bonds, which typically yield lower but less volatile returns, one might reasonably conclude that an expected return of 10% is too optimistic. In fact, Warren Buffet once critiqued firms’ use of high ERRs, stating, “I invite you to ask the CFO of a company having a large defined-benefit pension fund what adjustment would need to be made to the company’s earnings if its pension assumption was lowered to 6.5%” (Buffett and Loomis 2001).

We can therefore see a divergence in stakeholder interests in the context of DB pensions. Pension theory argues that workers effectively forgo present income, trusting that firms will manage and fund the plans and pay out the funds as a postretirement annuity (Kruse 1995). Thus, as Shleifer and Summers (1988) argued, ex ante risk-sharing between the firm and DB participants can lead to ex post opportunistic appropriation because DB plans do not clearly define the property rights held by pension participants. Because workers entrust the firm with their forgone wages and executives can use excess pension assets for their own ends, in a DB arrangement the employee–executive relationship represents a context rife with potential principal–agent conflicts (Eaton, Nofsinger, and Varma 2014).

Scholars commonly refer to the manipulation of ERRs as a form of earnings management. Examining this type of earnings management is useful for the study of executive decision-making for several reasons. First, executives have substantial discretion when setting the ERR, and rate changes reflect a conscious managerial choice. Second, unlike discretionary accruals and real-activity manipulations, ERRs are directly observable and largely unrelated to other dimensions of a firm’s performance that complicate analyses of these other forms of earnings management (Bergstresser et al. 2006). Finally, because ERRs directly relate to labor costs, aggressive ERR assumptions represent a transfer of assets from employees to the firm’s earnings, which potentially benefits investors and executives (Comprix and Muller 2011). Unlike other contexts in which we see this tradeoff, such as layoffs, the ERR transfer is largely unrelated to the firm’s operational strategy.

Prior research has focused chiefly on executives’ incentives to adjust ERRs for their own gain, finding, for example, that ERR increases when stock-based CEO compensation is higher (Bergstresser et al. 2006) and when the firm is likely to miss analysts’ earnings estimates (An, Lee, and Zhang 2014). In the sections below, I argue that employees can mitigate such manipulations and executives may be inclined to manipulate the rate to appease investors. I also argue that the CEO’s background influences how firms set their ERRs.

Union Influence

Although the corporate governance literature frequently neglects consideration of labor, employees can play a key role in the governance process through their ability to influence corporate decision-making and to control resources (Aguilera, Filatotchev, Gospel, and Jackson 2008). In a DB arrangement, firms invest workers’ forgone wages to provide an annuitized income in retirement. Because compensation tends to spike upward in a DB plan as a worker approaches retirement, she has a financial incentive to remain in the firm until her benefits have fully vested. Yet, workers’ willingness to accept a DB pension requires that firms match the explicit terms of the plan with an implicit promise of lifetime employment (Lazear 1986). We can therefore consider DB pensions to be both an incomplete contract that confers to workers a residual claim on a portion of the firm’s pension assets and an implicit promise of lifetime employment in exchange for effort and loyalty (Gustman, Mitchell, and Steinmeier 1994). By increasing the ERR, executives effectively push out pension costs, thereby increasing the funds necessary in the future to match the obligations owed to workers. In so doing, the firm increases the probability that it will fully renege on its pension promise by terminating or freezing its plan (Waring 2012). Consequently, higher ERRs can negatively affect workers.

Research shows that worker power influences key elements of the employment relationship, such as employment security and stability (Freeman and Medoff 1984). Although individual employees have little power to affect an organization’s strategies, when workers act collectively their influence is greater. The primary means by which employees do so is unionization, which provides workers an institutionalized apparatus to influence firms by engaging in or threatening strikes and slowdowns and by expressing demands through collective bargaining (Kochan 1980). Evidence has shown that unions play a crucial role in firms’ discretion to structure their employment relationships. For example, scholars have shown that unions deter firms’ ability to shift employment risk onto workers (Becker and Olson 1989; Gramm and Schnell 2001) and are associated with longer employment tenure (Jacoby 1985; Bidwell 2013) as well as improved benefits and wages (Freeman and Medoff 1984). In a union setting, the average worker’s interests predominate. Because the average worker is likely to be older and have higher exit costs, when workers bargain collectively they tend to seek more security, which DB pensions are designed to create (Allen and Clark 1986; Cobb 2015).

Unions have also served as important monitors of firms’ corporate-governance practices, and anecdotal evidence suggests that unions have been outspoken against firms’ aggressive ERR assumptions (Solomon and Hawkins 2005).⁴ Firms may also decrease their ERRs prior to labor negotiations, to make the plan appear more costly, and use this as a lever when seeking wage and/or benefit concessions (Benmelech, Bergman, and Enriquez 2012). Therefore, I expect that when workers have greater levels of collective influence in a firm as evidenced by higher rates of unionization, its executives will use less aggressive ERR assumptions.

Hypothesis 1: As a firm’s level of unionization increases (decreases), its expected rate of return on pension assets will decrease (increase).

Shareholder Influence and Fiduciary Standards

Extensive research has examined how investors influence executive decision-making in order to enhance investor returns, arguing that the structure of equity ownership helps to shape the political dynamics within a firm and directly affects its goals and strategies (Useem 1996). There are, however, different types of shareholders with distinct goals, objectives, and time horizons (Bushee 1998). In this study, I examine how one point of differentiation among institutional investors—the stringency of their fiduciary responsibilities to clients—affects ERRs.

Although all institutional investors have discretion over how they invest their clients’ assets and the law treats them all as fiduciaries, the strictness of the standards of prudence differs based on the investors’ legal form. Prudent-man laws are designed to protect clients by allowing them to seek damages when a fiduciary fails to invest in the clients’ best interest. While institutions typically own a diverse array of stocks, court rulings have stipulated that the merits of each stock in the portfolio be assessed individually to determine prudence (Badrinath, Gay, and Kale 1989). This stipulation motivates investors held to stricter fiduciary standards to protect themselves from liability by tilting their portfolios toward firms with better current-earnings performance, which the courts often use to determine the prudence of an investment (Del Guercio 1996).

Although prudent-man laws were amended in the 1990s, observers still contend that institutions with stricter fiduciary standards prefer nearer-term earnings. Fiduciaries may restrict themselves to specific investments and strategies endorsed by similar institutions as a way to minimize personal liability (Hawley, Johnson, and Waitzer 2011). Because shorter-term investment strategies have predominated in the past several decades, fiduciaries of institutions with stricter standards may still prefer investments with better short-term earnings performance. Supporting this claim, research reveals that institutions with stricter fiduciary standards hold preferences for short-term earnings and governance quality that differ from those of their weaker-standards counterparts (e.g., Bushee 2001; Shin and Seo 2011; Bushee, Carter, and Gerakos 2014).

Since increasing the ERR immediately enhances current-year earnings, I expect that the stringency of fiduciary standards will play a role in how executives set ERRs. Although I do not necessarily expect that shareholders will specifically request that a firm raise its ERR, when shareholders value near-term returns and directly or indirectly pressure firms to provide it, I expect that firms will be more apt to manipulate their pension assumptions. To test the impact of fiduciary restrictions on ERRs, I categorize institutions based on the stringency of their fiduciary standards. Prior research has identified four types of institutions held to strong fiduciary standards: bank trusts, public pension funds, company pension funds, and endowments (Bushee 2001). In the Online Appendix, I provide additional detail on the laws that stipulate the fiduciary standards for each of these investor types and how these laws have changed over time.

One might expect investors to discount any returns garnered through ERR increases. Several studies have found, however, that investors do not effectively assess the impact of pension accounting on firm value (e.g., Coronado and Sharpe 2003; Picconi 2006). Researchers surmise that the complexity of pension accounting, in which footnotes contain much pertinent information, makes it challenging for investors to effectively value plan changes (Coronado, Mitchell, Sharpe, and Nesbitt 2008). It is also plausible that investors who place greater value on firms with better current-earnings performance are relatively indifferent to earnings generated from pension changes versus from operations.

Despite their transitory impact on reported income and the often arbitrary nature in which firms set these assumptions, higher ERRs result in increased earnings and share prices, providing shareholders with greater short-run profit potential (Coronado and Sharpe 2003). Considering the stringency of fiduciary standards for banks, public pension funds, company pension funds, and endowments, I expect that firms with higher levels of ownership by these types of institutions will be more likely to increase their ERRs.

Hypothesis 2: As a firm’s level of equity ownership by institutions with strong fiduciary standards increases (decreases), its expected rate of return on pension assets will increase (decrease).

Although I do not hypothesize a relationship between equity holdings by investors with weaker fiduciary standards (i.e., insurance companies, investment advisors, and miscellaneous institutions) and ERR changes, I follow Bushee (2001) and control for them in my analyses.

CEO Background and Decision-Making

The literature on pension-assumption manipulations has focused prominently on CEO incentives to manage earnings, finding, for example, that ERR increases occur when stock-based CEO compensation is higher (Bergstresser et al. 2006) and when the firm is likely to miss analysts’ earnings estimates (An et al. 2014). I complement and extend this work by examining how CEOs’ backgrounds affect their willingness to change the ERR.

When faced with similar circumstances, executives may make substantially different decisions based on their individual understanding of those situations. Thus, we can link, in part, firm-level characteristics and outcomes to the experiences and schemas of senior executives. Executives’ beliefs about the most effective tactics and strategies likely emerge from prior training and experience, which significantly affect individuals’ cognition and decision-making schemas (Dearborn and Simon 1958). The functional background of CEOs, therefore, likely influences how they resolve stakeholders’ divergent interests (Ocasio and Kim 1999).

CEOs with a background in finance are of particular interest, as at least two explanations suggest that they might be more likely than others to manipulate ERRs. First, scholars believe that training and early career experience in finance affect CEOs’ commitment to the primacy of markets and to the privileging of shareholders as the sole stakeholder to whom firms are beholden (Fiss and Zajac 2004). A shareholder-primacy view may encourage a shorter investment time horizon and justify the maximizing of rents to equity holders by reducing rents to labor. Scholars believe that, as a result, finance CEOs view labor as a cost to be minimized, and research has found that CEOs with a finance background are more likely to engage in downsizing (Jung 2015). Executives with a finance background will favorably view the minimizing of a firm’s long-term obligations to its workers, as these executives are likely to view the workers’ demands and concerns as less salient than shareholder concerns (Bundy, Shropshire, and Buchholtz 2013).

Second, evidence also suggests that CEOs with a finance background are more financially sophisticated. Research has found such CEOs to be better able to raise external capital, less likely to hold cash, and more likely to issue debt and to use more aggressive financial policies (Custódio and Metzger 2014). Thus, one might expect CEOs with a finance background to better understand the consequences of raising a firm’s ERR and to see this as a way to manage earnings without the risk of impropriety. Overall, given their predisposition toward a market orientation, attention to financial measures of performance, and their greater knowledge and understanding of financial instruments, I expect that when a firm has a finance CEO, its ERR will be higher.

Hypothesis 3: When a firm’s CEO changes from one with a background in non-finance (finance) to one with a background in finance (non-finance), the firm’s expected rate of return on pension assets will increase (decrease).

Data and Methods

Analytic Approach

In this study, the unit of analysis is the firm, and the unit of observation is the firm-year. To examine the relationship between ERRs and the covariates, I used a pooled time-series regression analysis with firm-fixed effects. A fixed-effects model accounts for firms’ unobserved characteristics that do not vary over time and that may affect ERRs; thus, the model strengthens the inferences about the covariates’ effects on ERRs by ruling out the possibility that firms that adopted those rates had stable unobserved preferences for their value. A Hausman test indicated that the fixed-effects model was appropriate (χ² = 75.88, p < .001). Using variance inflation factor tests, I found that multicollinearity was not an issue.

Sample

I began my analysis by examining the years 1986 to 2006, a choice motivated by the passing of new Financial Accounting Standards Board (FASB) guidelines in 1985, making 1986 the first year in which a change in pension assumptions under the new guidelines could have occurred. When studying large firms, researchers commonly draw their samples from a single year, which can create survivorship bias. Thus, I elected to draw my sample from multiple years spanning the observation period. The initial sample included all publicly traded firms that had a DB plan and were included on the 1986 Fortune 500 list of largest industrials; the 1995 Fortune 1,000 list, consisting of the 500 largest industrial and 500 largest service firms; or the 2007 Fortune 500 list, consisting of the 500 largest firms. This selection generated an unbalanced sample of 938 firms. Because some of the CEO data were available from only 1992 onward, however, the primary analyses I present here use data for 1992 to 2006 only, yielding an unbalanced sample of 708 firms. A total of 41 firms were missing data on at least one control variable; thus, the final sample was 667 firms.

All firms appearing on any of those Fortune lists are in my sample for every year that they existed and had a DB plan between 1992 and 2006. Firms exited the sample if they were liquidated, acquired, if they terminated their DB plan(s), or if they hard froze their DB plan(s). In Online Appendix B, I describe the implications of firms dropping out of the sample and the steps I took to ensure that this had no substantive impact on my findings.

Variables

Expected Rate of Return

My dependent variable is the ERR on pension assets. The ERR data come from Compustat. For ease of interpretation, I multiplied the ERR by 100.

Unionization

I included a measure of union influence, unionization, which is the percentage of a firm’s DB participants covered by collectively bargained plans. These data come from Form 5500 reports. Each year, firms offering retirement plans to their workers must submit a Form 5500 report to the US Department of Labor that details information about the plan, including a field indicating whether a pension plan is part of a collective bargaining agreement. Unions often stipulate that some firms’ decisions, particularly those directly affecting employees covered by a collective bargaining agreement, require union participation (Freeman and Medoff 1984). Collectively bargained plans, therefore, limit firms’ ability to modify the plan’s terms unilaterally.

Institutional Share Ownership

To examine whether fiduciary standards are associated with ERR increases, I created two variables: strong fiduciary standards and weak fiduciary standards. Using the coding scheme from Bushee (2001), I took the sum of the percentages of the total number of shares outstanding for the focal firm owned by bank trusts, public and company pension funds, and endowments as my measure of strong fiduciary standards for each firm in each year. I took the sum of the percentages of the total number of shares outstanding for the focal firm owned by insurance companies; investment advisors, which includes investment companies and independent investment advisors; and miscellaneous institutions for my measure of weak fiduciary standards.⁵ I obtained the data from a publicly available website maintained by Bushee (1998), and I lagged the data one year. I also examined the effects of equity ownership by each institutional investor type independently and discuss those results below.

CEO Background

I coded the functional background of CEOs manually, based on their prior work experience. I coded individuals who served in an accounting or finance-related position, such as chief financial officer or treasurer, as a finance CEO. Because firms typically set the ERR early in the fiscal year, I considered the CEO of record to be the individual working at the end of quarter one of each fiscal year. In the analyses below, I compared finance CEOs with all other CEO background types. I collected the information from Who’s Who in Finance and Industry and supplemented with information from the Dun & Bradstreet Reference Book of Corporate Management and company proxy filings.

Control Variables

Other factors may confound the relationship between a firm’s ERR, on the one hand, and its levels of unionization, equity ownership by institutions with strong fiduciary standards, and the presence of a CEO with a finance background, on the other. To account for the possibility that firms’ size affects the relationship between the independent variables and ERRs, I included the lagged log of total assets. I included a measure of revenue growth, as it may relate to the firms’ ability to finance their DB plans. To account for the effect of performance on the relationship between the predicted relationships, I took the company’s previous-year return on equity (ROE). Analyses using return on assets yield similar results. Because this variable has some extreme values, I winsorized this figure at the 99th and 1st percentiles.⁶ I included a measure of long-term debt to total assets, lagged one year, as a measure of the firm’s relative covenant constraints. I also winsorized this figure at the 99th percentile, as it contained extreme values at the upper end. The employment and financial data come from Compustat. The size of a firm’s pension plan(s) may make it more willing to increase the ERR of its pension assets; therefore, I also controlled for the size of the firm’s DB pension plan(s), using the log total of all the assets the firm has in its DB plan(s) (DB assets). I took pension asset data from the Form 5500 filings, and the operating asset, revenue, performance, and debt data come from Compustat.

To account for the possibility that other governance mechanisms may affect the relationship between the independent variables and ERR changes, I included a measure of shareholder rights, which measures the number of practices the firm adopted that grant (or restrict) shareholder rights. I used the governance index (Gompers, Ishii, and Metrick 2003) obtained from the Investor Responsibility Research Center (IRRC) governance database and reverse-coded it so that higher values equal greater shareholder rights. I also included a control measure for CEO tenure, as CEOs who are more entrenched have greater influence and may be monitored less rigorously. Moreover, this measure helps to account for the fact that CEO changes could result from shareholder pressure (see Gillan and Starks 2007). Prior work has also found a positive relationship between option-based executive compensation and higher ERRs (Bergstresser et al. 2006). Hence, I included the ratio of stock-option pay to total pay (CEO pay from options) of the firm’s CEO, for which I use the Black-Scholes methodology of valuation to calculate the option value. For robustness, I also examined the log value of options; the impact on the hypothesized covariates was immaterial. I obtained the CEO pay and tenure data from the Compustat Executive Compensation database. Finally, because researchers speculate that the threat of acquisition disciplines management, I include a lagged count of mergers and acquisitions (industry mergers) at the two-digit Standard Industrial Classification (SIC) level to account for the fact that some firms face more active acquisition markets. These data come from the Securities Data Company Platinum database.

CEOs have incentives to report earnings that match or exceed those expected by analysts, as missing an earnings target has negative consequences for the firm’s stock price and CEO compensation. Although Picconi (2006) found that firms typically set ERR at the beginning of the fiscal year, An et al. (2014) found evidence that firms increase the ERR to meet or exceed analyst forecasts. To account for this possibility, I included two control variables. The first is a binary indicator, missed earnings, which I calculated by subtracting the firm’s reported earnings per share (EPS) from the median analyst forecast. The measure is equal to 1 when the firm missed its earnings target for the year.

The second indicator, avoid missed earnings, equals 1 when the firm met or exceeded its earnings forecast but would not have done so had it not increased its ERR. Following the procedure used by An et al. (2014), I calculated an adjusted EPS using the following formula:

EPS' = EPS - [(Δ ERR * FVPA * (1 - MTR)] / # of common shares outstanding

where EPS is the reported earnings per share, $Δ$ ERR is the change in the firm’s ERR from the previous year, FVPA is the fair value of pension assets, and MTR is the marginal tax rate. This calculation yields an estimate of the firm’s EPS had it not changed its ERR. I subtracted from this figure the median analyst earnings estimate to determine whether the firm would have missed its estimate had its ERR not changed. I obtained the ERR and pension data from the Form 5500 and Compustat. I obtained the MTR data from Blouin, Core, and Guay (2010), the EPS and common share data from Compustat, and the median analyst forecast data from Thomson Reuters’ Institutional Brokers’ Estimate System database.

Additionally, I controlled for the possible effect of industry peers’ retirement practices on a firm’s retirement policy, by creating the variable industry average ERR. This figure represents the average ERR among firms in my sample that share the same two-digit SIC code. To account for unobserved effects that may matter for a given year, I included year dummy codes.

Tables 1 and 2 present the descriptive statistics and correlations. Average ERRs steadily increased until 2002, at which time they decreased. I explain this trend in detail in the supplemental Online Appendix. Unionization rates in my sample were comparable to those of the overall private-sector labor force.⁷ The equity ownership rates were similarly comparable to those found by Shin and Seo (2011) with the exception of bank ownership, which was modestly higher in my sample. The proportion of firms with finance CEOs was approximately 31% during the period of observation. Compared to prior studies, the proportion in my sample is higher than the 19.2% found by Jung (2015), lower than the 42.7% found by Custódio and Metzger (2014), and comparable to the 32% found by Fligstein and Markovitz (1993).

Table 1.

Descriptive Statistics for Select Variables and Select Years

Variable	1993	1996	1999	2002	2005	Total
Firms	403	423	380	357	402	667
ERR (in 100s)	905.53	909.33	922.61	884.69	825.68	891.32
ERR (in 100s)	(91.86)	(81.43)	(77.25)	(64.97)	(59.83)	(80.37)
Unionization (%)	15.51	15.28	16.32	14.18	12.74	14.71
Unionization (%)	(30.22)	(29.99)	(31.11)	(29.34)	(28.78)	(29.70)
Strong fiduciary standards ownership_t-1 (%)	17.04	15.88	16.85	17.13	16.96	16.93
Strong fiduciary standards ownership_t-1 (%)	(8.35)	(6.73)	(6.54)	(6.41)	(6.13)	(6.89)
Bank ownership_t-1 (%)	11.63	11.72	12.74	12.89	13.32	12.45
Bank ownership_t-1 (%)	(7.24)	(5.91)	(5.84)	(6.05)	(5.60)	(6.92)
Public pension ownership_t-1 (%)	3.89	3.18	3.06	3.44	3.07	3.45
Public pension ownership_t-1 (%)	(2.42)	(1.94)	(1.17)	(1.11)	(0.84)	(1.67)
Company pension ownership_t-1 (%)	1.08	0.52	0.67	0.64	0.45	0.71
Company pension ownership_t-1 (%)	(1.76)	(0.81)	(1.41)	(0.87)	(0.61)	(1.19)
Endowment ownership_t-1 (%)	0.44	0.46	0.38	0.16	0.12	0.32
Endowment ownership_t-1 (%)	(1.07)	(1.14)	(1.01)	(0.26)	(0.24)	(0.81)
CEO finance	0.33	0.28	0.32	0.32	0.29	0.3
Weak fiduciary standards ownership_t-1 (%)	35.85	37.53	43.43	45.81	54.27	43.35
Weak fiduciary standards ownership_t-1 (%)	(14.21)	(15.25)	(15.31)	(16.04)	(16.31)	(16.49)
Insurance co ownership_t-1 (%)	4.92	5.44	5.49	5.11	4.13	5.13
Insurance co ownership_t-1 (%)	(3.71)	(3.42)	(3.70)	(3.67)	(2.77)	(3.52)
Investment advisor ownership_t-1 (%)	30.78	31.99	37.41	39.77	47.29	37.32
Investment advisor ownership_t-1 (%)	(13.00)	(14.13)	(14.06)	(15.10)	(15.96)	(15.33)
Miscellaneous institution ownership_t-1 (%)	0.15	0.09	0.23	0.92	3.24	0.96
Miscellaneous institution ownership_t-1 (%)	(1.37)	(0.88)	(1.45)	(2.00)	(3.22)	(2.25)
Total assets_t-1 (log)	8.37	8.38	8.65	8.91	9.26	8.73
Total assets_t-1 (log)	(1.33)	(1.39)	(1.41)	(1.47)	(1.55)	(1.47)
CEO tenure	5.63	5.42	4.98	4.30	4.78	5.07
CEO tenure	(6.27)	(6.33)	(6.04)	(4.75)	(4.63)	(5.08)
CEO pay from options (%)	19.19	25.92	37.57	38.21	32.46	29.92
CEO pay from options (%)	(19.63)	(22.87)	(25.67)	(26.33)	(34.65)	(25.50)
ERR in firms with at least one collectively bargained DB plan	903.55	903.08	918.81	884.27	835.83	887.29
	(103.49)	(76.59)	(80.65)	(65.49)	(60.23)	(84.14)
ERR in firms where ownership by strong fiduciary institutions > mean	923.54	917.23	928.39	888.94	828.66	895.23
	(94.31)	(73.35)	(64.84)	(65.46)	(56.96)	(82.36)
ERR in firms with finance CEO	907.99	910.88	925.23	886.03	834.95	893.12
ERR in firms with finance CEO	(102.79)	(102.05)	(74.04)	(62.32)	(62.28)	(83.74)

Notes: Standard errors in parentheses. DB, defined benefit; ERR, expected rate of return.

Table 2.

Descriptive Statistics and Correlation Matrix, 1992–2006

Variable		Mean	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1	Expected rate of return (in 100s)	891.32	80.37	1
2	Unionization (%)	14.71	29.70	−0.03	1
3	Strong fiduciary standards ownership_t-1 (%)	16.93	6.89	0.03	−0.02	1
4	CEO finance	0.30	0.46	−0.01	−0.03	−0.05	1
5	Weak fiduciary standards ownership_t-1 (%)	43.35	16.49	−0.08	−0.06	0.05	−0.04	1
6	Total assets_t-1 (log)	8.73	1.47	−0.04	−0.12	0.14	0.23	−0.10	1
7	Revenue growth (%)	0.75	2.36	−0.05	−0.02	0.03	0.03	0.03	0.01	1
8	Return on equity_t-1 (%)	12.30	29.41	0.00	−0.01	0.07	0.01	−0.03	0.02	0.03	1
9	Debt-to-assets ratio_t-1 (%)	22.55	13.71	0.05	0.09	−0.14	−0.10	0.05	−0.15	−0.03	−0.07	1
10	DB assets (log)	19.96	1.65	0.12	0.02	0.16	0.00	−0.10	0.58	−0.05	0.04	−0.02	1
11	Shareholder rights	13.82	2.53	−0.05	0.04	−0.10	−0.03	−0.11	0.13	0.02	0.01	0.02	0.07	1
12	CEO tenure	5.07	5.80	−0.01	0.00	0.02	−0.02	−0.04	0.00	0.03	0.02	0.00	−0.08	0.06	1
13	CEO pay from options (%)	29.92	25.50	0.07	−0.05	0.11	0.00	0.16	0.13	−0.01	0.06	−0.11	0.10	−0.05	−0.09	1
14	Industry mergers_t-1	6.71	7.77	0.01	0.00	−0.03	0.12	−0.10	0.27	0.03	0.04	−0.06	0.21	0.00	−0.04	0.12	1
15	Missed earnings	0.43	0.50	0.01	0.00	−0.07	−0.02	0.06	−0.04	−0.14	−0.06	0.11	0.02	−0.01	−0.04	0.03	0.03	1
16	Avoid missed earnings	0.04	0.19	0.07	0.04	−0.01	0.01	−0.07	0.00	0.03	0.01	−0.02	0.06	0.03	−0.03	−0.02	0.01	−0.17	1
17	Industry average ERR_t-1 (%)	895.42	43.02	0.38	0.02	−0.03	−0.03	−0.13	−0.19	−0.09	−0.01	0.05	−0.05	−0.04	0.02	0.06	0.00	0.05	0.02	1

Notes: DB, defined benefit; ERR, expected rate of return; SD, standard deviation.

Results

Table 3 shows the results of the fixed-effects regressions. Model 1 contains the results for the independent variables. In model 2, I included only the control variables. Model 3 shows the results of the full model. As mentioned, firm-fixed effects capture only within-firm differences in the covariates. Although CEO tenure rates declined during the observation period, CEO turnover is still infrequent enough that CEO functional background does not vary greatly over time for most firms. Thus, in model 4, I ran the analyses using firm-random effects, with robust standard errors with industry dummies at the two-digit SIC level. This model captures cross-firm differences in the covariates.

Table 3.

Ordinary Least Squares Regressions on ERRs on Pension Assets, 1992–2006

Variable	Model 1	Model 2	Model 3	Model 4
Unionization (%)	−0.264**		−0.281**	−0.221***
	(0.082)		(0.091)	(0.057)
Strong fiduciary standards ownership_t-1 (%)	0.406**		0.360*	0.301*
	(0.141)		(0.159)	(0.148)
CEO finance	4.529*		6.095*	5.110*
	(2.306)		(2.464)	(2.284)
Weak fiduciary standards ownership_t-1 (%)			0.053	0.122
			(0.089)	(0.079)
Total assets_t-1 (log)		3.367	3.431	3.207+
		(2.217)	(2.241)	(1.687)
Revenue growth (%)		−0.142	−0.197	−0.216
		(0.299)	(0.299)	(0.296)
Return on equity_t-1 (%)		−0.053*	−0.054*	−0.049*
		(0.024)	(0.024)	(0.024)
Debt-to-assets ratio_t-1 (%)		−0.015	−0.006	0.075
		(0.092)	(0.092)	(0.085)
DB assets (log)		−0.195	−0.242	4.023***
		(1.491)	(1.493)	(1.172)
Shareholder rights		−4.157***	−3.808***	−2.580***
		(0.773)	(0.783)	(0.589)
CEO tenure		0.417**	0.427**	0.425**
		(0.154)	(0.155)	(0.148)
CEO pay from options (%)		−0.033	−0.035	−0.012
		(0.033)	(0.033)	(0.032)
Industry mergers_t-1		0.146	0.117	0.229
		(0.139)	(0.139)	(0.142)
Missed earnings		2.533	2.600	2.491
		(1.615)	(1.617)	(1.574)
Avoid missed earnings		22.722***	22.552***	21.563***
		(3.697)	(3.695)	(3.663)
Industry average ERR_t-1 (%)		0.115***	0.112***	0.108***
		(0.031)	(0.031)	(0.032)
Constant	917.428***	981.832***	974.590***	990.520***
	(2.549)	(11.694)	(11.905)	(31.196)
Observations	6,786	5,796	5,796	5,796
Firms	708	667	667	667
Year dummies	Yes	Yes	Yes	Yes
Firm-fixed effects	Yes	Yes	Yes	No
R-squared	0.309	0.348	0.353	0.258

Notes: Standard errors in parentheses. DB, defined benefit; ERR, expected rate of return.

***

p < .001; **p < .01; *p < .05; +p < .10.

Because employees want to ensure that their pension benefits remain secure, I predicted that employee influence would be negatively associated with ERRs. Lending support for Hypothesis 1, the results in models 1 and 3 show a significant, negative relationship between unionization and ERRs. To convey the magnitude of this effect, I find, using the results from model 3 and holding all covariates at a fixed value, that a one standard deviation increase in the unionization rate leads to an 8.51 basis-point decrease in the ERR. I also ran analyses with a binary indicator regarding whether the firm has at least one collectively bargained plan, and those results were significant. For firms with at least one unionized plan, on average 45% of its DB participants are members of a collectively bargained plan. Therefore, the presence of a collectively bargained plan in a firm helps participants in non-unionized DB plans.

Hypothesis 2 predicted a positive relationship between equity ownership by institutions with strong fiduciary standards and ERRs. Across models, we see a positive and significant relationship between these two variables, which supports Hypothesis 2. According to the results in Table 3, model 3 and with all covariates held at a fixed value, a one standard deviation increase in equity ownership by institutions with strong fiduciary standards would increase a firm’s ERR by 2.85 basis points on average. These results support the claim that executives seek to enhance current-earnings performance by increasing ERRs when institutions with strong fiduciary standards own more of the firm’s equity. Additionally, as I alluded to above and describe in detail in the Online Appendix, the law governing fiduciary responsibilities changed in 1997. At that time, the courts based prudence determinations on the entire portfolio rather than on each individual investment, which should have decreased pressure on fiduciaries to prioritize short-term earnings performance when choosing stocks. Thus, I created a dummy variable indicating years after 1996 and interacted this dummy with my measure of high fiduciary standards. The effect of fiduciary standards on ERRs did significantly weaken after this legal change. The results revealed a negative and significant interaction term (b = −0.776, p < .001) and a positive and significant main effect for high fiduciary standards (b = 0.796, p < .001).

I also examined bank, public pension, company pension, and endowments as separate covariates in the model. The results indicate a positive and significant relationship between ownership by bank trusts and ERRs (b = 0.400, p < .05). They also reveal a positive relationship between public pension ownership and ERRs, although the effect is only modestly significant (b = 0.924, p < .10). The estimates for company pensions and endowments were insignificant. As Table 1 indicates, bank trusts are the largest category of institution with strong fiduciary standards. Thus, bank-trust ownership primarily drives this result.

Hypothesis 3 predicted that, compared to other CEOs, the presence of a CEO with a finance background will positively relate to ERRs. The results support the argument that because finance CEOs prefer to maximize short-term earnings and/or are more financially sophisticated, they are more willing than non-finance CEOs to increase ERRs. Based on the findings in Table 3, model 3 and with all other covariates held at a fixed value, were a firm with a CEO whose background was not in finance to transition to one with a finance background, its ERR would increase by 6.10 basis points on average. These findings reinforce those from other studies suggesting that finance CEOs make decisions that favor shareholder welfare over that of other constituents (e.g., Jung 2015).

Across models, several controls significantly related to ERRs. ROE has a modestly significant and negative impact on ERRs, indicating that when a firm’s financial performance declined, its executives increased the ERR in the subsequent year. Shareholder rights also negatively and significantly relate to ERRs. CEO tenure also positively and significantly relates to ERR, and executives are significantly more likely to increase the ERR when it allows the firm to avoid missing its earnings target. Finally, firms in industries with higher ERRs in the previous year are also likely to increase their ERR.

Supplemental Analyses

In observational studies of this type, establishing a causal relationship between the covariates and the dependent variable can be challenging, and I am aware of concerns about endogeneity affecting these results and inferences about causality. I attempted to deal with this concern empirically by using a two-stage least-squares analysis and a triple-difference analysis.

Two-Stage Least-Squares Model

To attempt to address the concern that ownership by institutions with strong fiduciary standards is endogenous with the dependent variable, I ran a two-stage least-squares (2SLS) regression with firm-fixed effects. In a traditional ordinary least-squares framework, a basic assumption is that the error term is independent of the predictor variables. The 2SLS helps to correct issues that arise from violation of this assumption. In the first stage of the analysis, I regress the endogenous regressor on the instrument variable (IV) and the covariates, which isolates the variation in the predictor that does not correlate with the error term. In the second stage, I then use the fitted value from the first stage in place of the endogenous regressor, allowing us to interpret the coefficient of the endogenous regressor as “capturing a covariate-adjusted causal effect” (Bascle 2008: 294).

I identified two possible instruments. First, I included the number of stock analysts covering the firm as an IV for equity ownership by institutions with strong fiduciary standards. I expect that analyst coverage will correlate with equity ownership by institutional investors, but the number of analysts covering a firm should not be subject to reverse feedback from variations in a firm’s ERR. Prior research has used the number of analysts as an IV to predict the impact of institutional ownership on firm performance (Cornett, Marcus, Saunders, and Tehranian 2007) and on opportunistic ERR increases (Eaton et al. 2014). Second, I included the annual average holdings by institutions with strict fiduciary standards at the focal firm’s two-digit SIC level, excluding the focal firm (industry-level strong fiduciary standards ownership), as a second IV. Prior research has documented a positive correlation between firm-level and industry-level institutional holdings (Choi and Sias 2009), and observers have noted that institutions with stricter fiduciary standards converge around similar investments (Hawley et al. 2011). However, the average level of institutional holdings in each firm’s industry is plausibly exogenous to the focal firm’s ERR. The Sargan overidentification test statistic was insignificant; thus, I cannot reject the null that the IVs are exogenous. The Wald chi-square test was significant.

The results of the 2SLS regressions appear in Table 4, model 5. As the results indicate, the relationship between ERRs and equity ownership by institutions with strong fiduciary standards remains positive and significant. In the second stage of these models, unionization is not significant, nor are several other control variables that were previously significant, including ROE and CEO tenure. It is possible that the lost efficiency inherent in these analyses has affected the relationships between these other covariates and the dependent variable (Bascle 2008). If so, a more strongly identified model would likely improve these results.⁸

Table 4.

Two-Stage Least Squares Regression on ERRs on Pension Assets, 1992–2006

Variable	Model 5
Unionization (%)	−0.194
	(0.123)
	First stage
Analysts_t-1	0.158***
	(0.020)
Industry-level strong fiduciary standards ownership_t-1	0.076***
	(0.016)
Strong fiduciary standards ownership_t-1 (%)	3.476**
	(1.333)
Finance CEO	4.910+
	(2.682)
Weak fiduciary standards ownership_t-1 (%)	0.165
	(0.116)
Total assets_t-1 (log)	0.726
	(3.018)
Revenue growth (%)	−0.196
	(0.340)
Return on equity_t-1 (%)	−0.051*
	(0.024)
Debt-to-assets ratio_t-1 (%)	0.067
	(0.111)
DB assets (log)	−2.795
	(1.855)
Shareholder rights	−2.679*
	(1.244)
CEO tenure	0.179
	(0.167)
CEO pay from options (%)	−0.031
	(0.039)
Industry mergers_t-1	−0.025
	(0.143)
Missed earnings	1.251
	(1.795)
Avoid missed earnings	23.800***
	(4.171)
Industry average ERR_t-1 (%)	0.103*
	(0.043)
Observations	5,557
Firms	642
Year dummies	Yes
R-squared	.030

Notes: Standard errors in parentheses. DB, defined benefit; ERR, expected rate of return.

***

p < .001; **p < .01; *p < .05; +p < .10.

Triple-Differences Analysis

Difference-in-difference (DD) approaches are a way to address endogeneity concerns, using observational data. They are particularly useful for estimating the effects of policy interventions that do not affect all actors similarly. In December 2002, the Securities and Exchange Commission (SEC) issued a warning that it might challenge ERRs set above 9%. Although the SEC issued the warning to all firms, we should expect this intervention to have differentially affected firms’ willingness to lower their ERRs after 2002 based on whether they had an ERR greater than 9% in 2002. As Figure 1 shows, the propensity of firms in my sample to maintain an ERR greater than 9% declined markedly after 2002, and the rate of decline in ERRs for firms with a high ERR was steeper than for firms with a low ERR.⁹

Figure 1.

Average ERR, 2000 to 2004

To examine whether the relationships between the hypothesized covariates and ERRs were sensitive to this change, I used a triple-differences (i.e., difference-in-difference-in-difference) design. The first difference estimation compares the difference in outcomes for firms before and after the SEC warning (post-SEC warning). The second compares the difference in ERRs between groups affected by the SEC warning (high ERR), which were firms with an ERR greater than 9% in 2002, and groups unaffected by the warning (i.e., those with an ERR less than or equal to 9% in 2002). Finally, I estimate the differences in the observed relationship between equity ownership by institutions with strong fiduciary standards and finance CEOs (third difference) and ERRs, after the SEC warning, for firms that had and did not have an ERR greater than 9% in 2002.¹⁰ My expectation is that for a firm with an ERR greater than 9% in 2002, if its level of ownership by institutions with strong fiduciary standards increases or if it transitions to having a CEO with a finance background, it will decrease its ERRs to a lesser degree post-SEC warning than it would have otherwise. Table 5 presents the results of the triple-differences models.

Table 5.

Triple-Difference Regressions on ERRs on Pension Assets, 1992–2006

Variable	Model 6	Model 7	Model 8
Post-SEC Warning	−90.155***	−88.461***	−92.278***
	(5.740)	(6.134)	(6.233)
High ERR	−25.459*	−23.280+	−21.785+
	(12.213)	(12.278)	(12.418)
High ERR×Post-SEC Warning	−20.435***	−21.494***	−24.843***
	(3.410)	(4.037)	(4.201)
Unionization (%)		−0.297**	−0.303***
		(0.090)	(0.091)
Strong fiduciary standards ownership_t-1 (%)		0.941***	0.312*
		(0.199)	(0.158)
Strong fiduciary standards ownership×High ERR		−1.184**
		(0.373)
Strong fiduciary standards ownership×Post-SEC Warning		−1.512***
		(0.312)
Strong fiduciary standards ownership×High ERR×Post-SEC Warning		0.975*
		(0.486)
Finance CEO		5.439*	4.885
		(2.444)	(3.175)
Finance CEO×High ERR			−3.068
			(6.099)
Finance CEO×Post-SEC Warning			0.045
			(4.276)
Finance CEO×High ERR×Post-SEC Warning			14.860*
			(7.484)
Constant	986.517***	976.625***	979.579***
	(11.962)	(12.175)	(12.215)
Observations	5,809	5,800	5,800
Firms	667	667	667
Year dummies	Yes	Yes	Yes
R-squared	0.355	0.361	0.358

Notes: Standard errors in parentheses. ERR, expected rate of return; SEC, Securities and Exchange Commission.

***

p < .001; **p < .01; *p < .05; +p < .10.

In Table 5, model 6, I regressed onto firms’ ERRs all control variables and a dummy code for the post-SEC warning, a dummy variable indicating whether a firm had a high ERR in 2002, and an interaction of these variables. As expected, the results indicate that after 2002, ERRs decreased significantly, and this effect was greater for firms that had a high ERR in 2002. In model 7, I included the triple difference by interacting rates of equity ownership by institutions with strong fiduciary standards with the high ERR and post-SEC warning indicators. The positive and significant coefficient suggests that among firms with a high ERR in 2002, those with higher rates of ownership by institutions with strong fiduciary standards reacted less strongly to the SEC warning. Thus, despite a strong impetus to lower their ERRs, firms with higher rates of ownership by institutions with strong fiduciary standards lowered their ERRs to a lesser extent.

For model 8, I repeated the same steps for finance CEOs. The results reveal that firms with a finance CEO and a high ERR in 2002 decreased their ERRs in subsequent years to a lesser degree than did firms that had non-finance CEOs. I ran these analyses with firm-random effects, and the results are consistent with those presented here.¹¹

Although this intervention is not a pure experiment given that a firm’s choice of ERR in 2002 was not exogenous, it allowed me to examine whether institutional ownership and CEO background differentially affected the response of firms with a high ERR in 2002 to the SEC warning. Taken together, these results show that, despite a strong incentive to lower the ERR, executives did so to a lesser degree when their firms had higher rates of ownership by bank trusts and public pension funds and when they had a finance CEO. This finding provides additional insight on the purported causal relationship between these variables and ERRs.

Discussion

A recent issue facing many corporations is how to pay for past promises. We see this dilemma clearly with DB pensions. A DB plan allows firms to remunerate workers in the future for labor completed in the present. DB pensions require that firms make good on their promise to pay for these distant benefits through current contributions and the earnings on those contributions. Because the ERR is sensitive to manipulation, however, stakeholders may wish to influence its value for their advantage. The purpose of this study is to examine whether stakeholder power dynamics and CEO decision-making schemas influence how executives adjudicate the demands of two principals with conflicting interests, as evidenced by firms’ ERRs on pension assets.

Drawing insights from sociopolitical theories of the firm, power-dependence perspectives, and the literature on corporate governance, I argue that when executives serve as agents to two principals with conflicting interests, power dynamics and executives’ functional expertise influence which principal receives priority. The findings provide support for the hypotheses, showing first that a firm’s level of unionization negatively associates with ERRs. Although the literature has traditionally overlooked the importance of labor in corporate governance, the results here point to the influence of employee power on corporate decision-making. The results also generally reveal that ownership by bank trusts and public pension funds is positively associated with firms’ ERRs. That equity ownership by pension funds undermines DB funds by encouraging earnings manipulations speaks to some of the inherent contradictions of workers’ capital as a mechanism that protects broader labor interests (see Appelbaum and Batt 2014). Finally, although agency theory tends to treat agents’ interests as uniform, my results indicate the important effect of executive background on those interests. My findings support the idea that CEOs’ functional backgrounds affect how firms resolve stakeholders’ diverging interests, as the interests of CEOs with a finance background seem more closely aligned with those of investors preferring better current-earnings performance.

My study is not without limitations that indicate directions for future research, however. First, for example, I do not have data on wages and other fringe benefits. Therefore, it is possible that as firms increase their ERRs, they use the proceeds to remunerate workers in other ways, such as increasing wages or paying for rising health insurance premiums. The literature on DB plan terminations and freezes does not suggest that the altering of firms’ retirement plans led to greater worker remuneration, however (see Rauh, Stefanescu, and Zeldes 2017). Hence, although the compensating-differential rationale for ERR increases seems unlikely, I cannot rule it out.

Second, although scholars believe that increasing ERRs have negative consequences for workers by making pension plans more volatile (Waring 2012), a firm increasing its ERR might, at least temporarily, benefit workers by allowing the firm to remain solvent or to avoid cost-cutting strategies such as layoffs. However, if firms raised ERRs to deal with short-run declines in performance, we would expect them to lower their ERRs once performance improved. If this occurred systematically, it is improbable I would find significant effects because I did control for performance. Additionally, this possibility would only raise issues in the unlikely case that a firm’s desire to raise ERRs was negatively correlated with unionization rates (i.e., firms have lower levels of unionization) and positively correlated with ownership by institutions with strong fiduciary standards and with the firm having a finance CEO. Although I cannot rule out the possibility that employees at times benefit indirectly from rising ERRs, no evidence suggests that executives increase ERRs with employee welfare in mind.

Third, although common in observational studies, I cannot assess directly the influence of unionization, equity ownership, and CEO background. Although the data patterns largely support the claims made herein, the information one can infer from these results is limited. Furthermore, while the supplemental analyses help to address concerns with endogeneity, they cannot completely rule it out.

Finally, there is no way to discern the intent behind ERR increases, and executives may have changed the rate for reasons other than to boost earnings. Unless these other rationales for increasing the ERR correlate with my independent variables (e.g., finance CEOs are more optimistic about beating historical returns), finding significant results should be more difficult. It is also reasonable to assume that regardless of motive, executives understand the impact of ERR changes on firm earnings. To address each of these limitations, future research can exploit various methodologies and data to rule out alternative explanations and to better elucidate the mechanisms I hypothesized to be driving these results.

Despite these limitations, this study contributes to current research in several ways. First, I present a fuller picture of the often-competing demands CEOs face and how firms resolve them. Corporate-governance research normally highlights the tensions between shareholders and executives and the various mechanisms used to align their interests. I extend this research by analyzing the impact of stakeholder influence in a context in which the interests of the three key organizational stakeholders—employees, shareholders, and executives—are likely to conflict. Because DB pensions grant poorly defined property rights to workers, these pensions are subject to ex post opportunism by executives (Shleifer and Summers 1988). By treating employees and shareholders as principals, differentiating shareholders by their fiduciary responsibilities, and considering how the functional expertise of CEOs informs how they adjudicate employee and shareholder interests, I attend to the complexity associated with managing multiple stakeholders.

This study also contributes to research on employment relationships and industrial relations by highlighting the effects of labor power and firms’ financial strategies on employment practices. A long research tradition has examined the impact of unions on worker outcomes (e.g., Freeman and Medoff 1984). This study further elucidates the role of organized labor in protecting the welfare of unionized and non-unionized workers alike (see Farber 2005). Specifically, the results indicate that unionization lowers a firm’s probability of using their ERRs as an earnings-management tool, thereby protecting retirement assets for all workers in a firm. Additionally, recent studies have focused on how the rise of financial capitalism has transformed employment relationships in the United States (e.g., Jacoby 2011; Appelbaum and Batt 2014; Cobb 2016; Lin 2016; Cobb and Stevens 2017). I add to this research by demonstrating that equity ownership by institutions with strong fiduciary standards, particularly bank trusts, and CEOs with a finance background lead firms to manage earnings by raising the ERR on pension assets. In so doing, I provide additional evidence that implicit contracts between workers and firms are susceptible to the influence of financial actors and to CEOs who are more likely to be guided by financial logic.

Finally, prior survey evidence has shown that firms’ desire to amend their retirement plans is primarily a response to concerns over the volatility of plan costs (Towers Watson 2010). In this study, I show that declining levels of unionization, equity ownership structures, and CEO backgrounds help to create that uncertainty. My findings suggest that because higher ERRs make future pension costs more uncertain, stakeholder conflict helps to establish the conditions under which firms may shrink the size of their DB plans. This study therefore has implications for policy. Company pension funds hold trillions of dollars in investments, and millions of workers rely on these funds as a key source of postretirement income. Given the increasing problems within the DB pension system in the United States, it is beneficial to know which factors are associated with better or worse management and funding of these plans so that policy interventions can protect workers’ pension benefits (Eaton et al. 2014).

Supplemental Material

DS_10.1177_0019793918789155 – Supplemental material for Managing the Conflicting Interests of Workers and Shareholders: Evidence from Pension-Assumption Manipulations

Supplemental material, DS_10.1177_0019793918789155 for Managing the Conflicting Interests of Workers and Shareholders: Evidence from Pension-Assumption Manipulations by J. Adam Cobb in ILR Review

Footnotes

Acknowledgements

I thank Iwan Barankay, Brian Bushee, Peter Cappelli, Jerry Davis, JR Keller, Martine Haas, Olenka Kascpercyk, John Paul MacDuffie, Anoop Menon, Olivia Mitchell, Mike Useem, and seminar participants at the Wharton Organizational Theory Workshop for thoughtful comments on earlier versions of this article. I also thank Matthew Bidwell and Zeke Hernandez for helpful feedback on multiple versions of this manuscript. Additionally, I thank Ben Davis, John Byon, and Constantin Gumentia for their assistance collecting data and Jiwook Jung for sharing with me data on CEO backgrounds.

A supplemental Online Appendix is available at . For information regarding the data and/or computer programs utilized for this study

1

Firms generally determine the interest cost by taking the discount rate multiplied by the beginning of the year pension-benefit obligation adjusted for current-year expected benefit payments. Various laws and statutes have influenced discount rates over time. The discount rate for calculating pension assets is often referred to as a long-term return assumption, although it differs from the assumption on which this article focuses.

2

The Financial Accounting Standards Board (FASB) is responsible for issuing Accounting Standards Codifications. FASB is a private-sector, nonprofit organization that establishes financial and reporting standards for public and private companies.

3

Firms enter deviations between actual and expected returns as an off-balance-sheet unrecognized gain or loss. The rules governing reconciliation between the actual and the expected return rate derive from the corridor approach, whereby the accumulated unexpected gains or losses are compared to the beginning pension benefit-obligation balance and the market value of the plan assets. Firms amortize any amount greater than 10% of the larger of the two over the remaining service lives of the firm’s employees.

4

This is not to say that all union leaders seek to maintain low return assumptions. Anecdotal evidence from the public sector reveals that some union leaders fought against ERR decreases (e.g., ). Additionally, a union official informed me during an informal interview that some union leaders have negotiated ERR increases for higher wages and other benefits, but to his knowledge this practice was not widespread.

5

These holdings represent internally managed investments. I record externally managed holdings, such as when a pension fund invests in a mutual fund, as holdings by investment advisors ().

6

Winsorization involves transforming data by limiting extreme values.

7

See .

8

I also ran models with bank, public pension, company pension, and endowments in separate equations with analysts as the IV in each equation. The results show a positive and significant relationship for bank-share ownership (b = 6.407, p < .05) and public pension fund-share ownership (b = 28.014, p < .05). The results for company pension fund and endowment ownership were insignificant. Additionally, I ran 2SLS models using classifications of institutions based on their investment behavior using analysts and industry dedicated/transient ownership as instruments. I tested whether the proportion of a firm’s equity owned by dedicated and transient investors affects ERRs, as prior research has found that ownership by these two types of investors affects firms’ pension choices (see Cobb 2015). See for further detail on these owner types. In further results, which are available upon request, I found a negative and significant relationship between ERRs and ownership by dedicated institutions (b = −4.227, p < .01), suggesting that increased ownership by institutions with a longer-term investment time horizon motivates firms to lower their ERRs. The results for transient investors, however, were not significant, and the instrumental variables were insignificant in the first stage.

9

A second policy change regarding pension assumptions was enacted in 2003. In that year, FASB issued SFAS 132R, which required companies to disclose their pension investments across four broad investment classes: equities, bonds, real estate, and other. I cannot rule out the possibility that the effects I show for the SEC warning are not attributable to firms anticipating SFAS 132R or firms anticipating the SEC warning in 2001 or in early 2002. There are two potential implications. First, if firms anticipated the SEC warning sometime during 2001 or early 2002, my proxies would show measurement error, thereby reducing the power of my tests. Second, I cannot differentiate the effect of the SEC warning from that of SFAS 132R. However, my goal is not to test the efficacy of a specific intervention but, instead, to examine how my covariates affect firms’ willingness to lower their ERRs in the wake of policy intervention(s) designed to scrutinize firms’ pension accounting decisions more heavily.

10

I do not include results on unionization, as my sample contained too few unionized firms with high ERRs in 2002. The results for unionization showed no significant effects.

11

I also ran models with bank, public pension, company pension, and endowment equity ownership in separate models. The triple difference for bank trusts was marginally significant and positive (b = 0.983, p < .10), and the triple difference for public pension ownership was significant and positive (b = 9.278, p < .05). The results for company pension and endowment ownership were not significant.

References

Aguilera

Ruth V.

Filatotchev

Igor

Gospel

Howard

Jackson

Gregory

. 2008. An organizational approach to comparative corporate governance: Costs, contingencies, and complementarities. Organization Science 19(3): 475–92.

Allen

Steven G.

Clark

Robert L.

1986. Unions, pension wealth, and age-compensation profiles. Industrial and Labor Relations Review 39(4): 502–17.

Heng

Lee

Yul W.

Zhang

Ting

. 2014. Do corporations manage earnings to meet/exceed analyst forecasts? Evidence from pension plan assumption changes. Review of Accounting Studies 19(2): 698–735.

Aoki

Masahiko

. 2001. Toward a Comparative Institutional Analysis. Cambridge, MA: MIT Press.

Appelbaum

Eileen

Batt

Rosemary

. 2014. Private Equity at Work: When Wall Street Manages Main Street. New York: Russell Sage Foundation.

Appelbaum

Eileen

Batt

Rose

Clark

Ian

. 2013. Implications of financial capitalism for employment relations research: Evidence from breach of trust and implicit contracts in private equity buyouts. British Journal of Industrial Relations 51(3): 498–518.

Asher

Cheryl C.

Mahoney

James M.

Mahoney

Joseph T.

2005. Towards a property rights foundation for a stakeholder theory of the firm. Journal of Management and Governance 9(1): 5–32.

Badrinath

S. G.

Gay

Gerald D.

Kale

Jayant R.

1989. Patterns of institutional investment, prudence, and the managerial safety-net hypothesis. Journal of Risk and Insurance 56(4): 605–29.

Bascle

Guilhem

. 2008. Controlling for endogeneity with instrumental variables in strategic management research. Strategic Organization 6(3): 285–327.

10.

Becker

Brian E.

Olson

Craig A.

1989. Unionization and shareholder interests. Industrial and Labor Relations Review 42(2): 246–61.

11.

Benmelech

Efraim

Bergman

Nittai

Enriquez

Ricardo

. 2012. Negotiating with labor unions under financial distress. Review of Corporate Financial Studies 1(1): 28–67.

12.

Bergstresser

Daniel

Desai

Mihir A.

Rauh

Joshua

. 2006. Earnings manipulation, pension assumptions, and managerial investment decisions. Quarterly Journal of Economics 121(1): 157–95.

13.

Bidwell

Matthew J.

2013. What happened to long-term employment? The role of worker power and environmental turbulence in explaining declines in worker tenure. Organization Science 24(4): 1061–82.

14.

Blouin

Jennifer

Core

John E.

Guay

Wayne

. 2010. Have the tax benefits of debt been overestimated?Journal of Financial Economics 98(2): 195–213.

15.

Buffett

Warren

Loomis

Carol

. 2001. Warren Buffett on the stock market [online version]. Fortune, December10. Accessed at http://archive.fortune.com/magazines/fortune/fortune_archive/2001/12/10/314691/index.htm (June 2015).

16.

Bundy

Jonathan

Shropshire

Christine

Buchholtz

Ann K.

2013. Strategic cognition and issue salience: Toward an explanation of firm responsiveness to stakeholder concerns. Academy of Management Review 38(3): 352–76.

17.

Bushee

Brian J.

1998. The influence of institutional investors on myopic R&D investment behavior. Accounting Review 73(3): 305–33.

18.

Bushee

Brian J.

2001. Do institutional investors prefer near-term earnings over long-run value?Contemporary Accounting Research 18(2): 207–46.

19.

Bushee

Brian J.

Carter

Mary Ellen

Gerakos

Joseph

. 2014. Institutional investor preferences for corporate governance mechanisms. Journal of Management Accounting Research 26(2): 123–49.

20.

Cappelli

Peter H.

1999. The New Deal at Work: Managing the Market-Driven Workforce. Boston: Harvard University Press.

21.

Choi

Nicole

Sias

Richard W.

2009. Institutional industry herding. Journal of Financial Economics 94(3): 469–91.

22.

Cobb

J. Adam

. 2015. Risky business: Firms’ shifting of retirement risk and the decline of defined benefit pension plans. Organization Science 26(5): 1332–50.

23.

Cobb

J. Adam

2016. How firms shape income inequality: Stakeholder power, executive decision-making, and the structuring of employment relationships. Academy of Management Review 41(2): 324–48.

24.

Cobb

J. Adam

Lin

Ken-Hou

. 2017. Growing apart: The changing firm-size wage premium and its inequality consequences. Organization Science 28(3): 429–46.

25.

Cobb

J. Adam

Stevens

Flannery G.

2017. These unequal states: Corporate organization and income inequality in the United States. Administrative Science Quarterly 62(2): 304–40.

26.

Comprix

Joseph

Muller

Karl A.

2011. Pension plan accounting estimates and the freezing of defined benefit pension plans. Journal of Accounting & Economics 51(1-2): 115–33.

27.

Cornett

Marcia M.

Marcus

Alan J.

Saunders

Anthony

Tehranian

Hassan

. 2007. The impact of institutional ownership on corporate operating performance. Journal of Banking & Finance 31(6): 1771–94.

28.

Coronado

Julia

Sharpe

Steven A.

2003. Did pension plan accounting contribute to a stock market bubble?Brookings Papers on Economic Activity 2003(1): 323–71.

29.

Coronado

Julia

Mitchell

Olivia S.

Sharpe

Steven A.

Nesbitt

S. Blake

. 2008. Footnotes aren’t enough: The impact of pension accounting on stock values. Journal of Pension Economics & Finance 7(3): 257–76.

30.

Custódio

Cláudia

Metzger

Daniel

. 2014. Financial expert CEOs: CEO’s work experience and firm’s financial policies. Journal of Financial Economics 114(1): 125–54.

31.

Cyert

Richard M.

March

James G.

1963. A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall.

32.

Davis

Gerald F.

2005. New directions in corporate governance. Annual Review of Sociology 31: 143–62.

33.

Dearborn

DeWitt C.

Simon

Herbert A.

1958. Selective perception: A note on the departmental identifications of executives. Sociometry 21(2): 140–44.

34.

Del Guercio

Diane

. 1996. The distorting effect of the prudent-man laws on institutional equity investments. Journal of Financial Economics 40(1): 31–62.

35.

Eaton

Tim V.

Nofsinger

John R.

Varma

Abishek

. 2014. Institutional investor ownership and corporate pension transparency. Financial Management 43(3): 603–30.

36.

Farber

Henry S.

2005. Nonunion wage rates and the threat of unionization. Industrial and Labor Relations Review 58(3): 335–52.

37.

Fiss

Peer C.

Zajac

Edward J.

2004. The diffusion of ideas over contested terrain: The (non)adoption of a shareholder value orientation among German firms. Administrative Science Quarterly 49(4): 501–34.

38.

Fligstein

Neil

Markowitz

Linda

. 1993. The finance conception of the corporation and the causes of the financial reorganization of large corporations, 1979–1989. In Wilson

William J.

(Ed.), Sociology and Social Policy, pp. 185–206. Newbury Park, CA: Sage Publications.

39.

Freeman

Richard B.

Medoff

James L.

1984. What Do Unions Do?New York: Basic Books.

40.

Gillan

Stuart L.

Starks

Laura T.

2007. The evolution of shareholder activism in the United States. Journal of Applied Corporate Finance 19(1): 55–73.

41.

Gompers

Paul A.

Ishii

Joy

Metrick

Andrew

. 2003. Corporate governance and equity prices. Quarterly Journal of Economics 118(1): 107–55.

42.

Gramm

Cynthia L.

Schnell

John F.

2001. The use of flexible staffing arrangements in core production jobs. Industrial and Labor Relations Review 54(2): 245–58.

43.

Gustman

Alan L.

Mitchell

Olivia S.

Steinmeier

Thomas L.

1994. The role of pensions in the labor market: A survey of the literature. Industrial and Labor Relations Review 47(3): 417–38.

44.

Hawley

James

Johnson

Keith

Waitzer

. 2011. Reclaiming fiduciary duty balance. Rotman International Journal of Pension Management 4(2): 4–16.

45.

Jacoby

Sanford M.

1985. Employing Bureaucracy: Managers, Unions, and the Transformation of Work in American Industry, 1900–1945. New York: Columbia University Press.

46.

Jacoby

Sanford M.

2005. The Embedded Corporation: Corporate Governance and Employment Relations in Japan and the United States. Princeton, NJ: Princeton University Press.

47.

Jacoby

Sanford M.

2011. Labor and finance in the United States. In Williams

Cynthia

Zumbanse

Peer

(Eds.), The Embedded Firm: Corporate Governance, Labor, and Finance Capitalism, pp. 277–317. Cambridge, UK: Cambridge University Press.

48.

Jensen

Michael C.

Meckling

William H.

1976. Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3(4): 305–60.

49.

Jung

Jiwook

. 2015. Shareholder value and workforce downsizing, 1981–2006. Social Forces 93(4): 1335–68.

50.

Kochan

Thomas

. 1980. Collective Bargaining and Industrial Relations: From Theory to Policy and Practice. Homewood, IL: Richard D. Irwin.

51.

Kruse

Douglas L.

1995. Pension substitution in the 1980s: Why the shift toward defined contribution?Industrial Relations 34(2): 218–41.

52.

Lazear

Edward P.

1986. Retirement from the labor force. In Ashenfelter

Orley C.

Layard

Richard

(Eds.), Handbook in Economics, Vol. 1, pp. 305–55. New York: Elsevier Science Publishers.

53.

Lin

Ken-Hou

. 2016. The rise of finance and firm employment dynamics. Organization Science 27(4): 972–88.

54.

Mitchell

Ronald K.

Agle

Bradley R.

Wood

Donna J.

1997. Toward a theory of stakeholder identification and salience: Defining the principle of who and what really counts. Academy of Management Review 22(4): 853–86.

55.

Ocasio

William

Kim

Hyosun

. 1999. The circulation of corporate control: Selection of functional backgrounds of new CEOs in large US manufacturing firms, 1981–1992. Administrative Science Quarterly 44(3): 532–62.

56.

Pfeffer

Jeffrey

Salancik

Gerald R.

1978. The External Control of Organizations: A Resource Dependence Perspective. New York: Harper & Row.

57.

Picconi

Marc

. 2006. The perils of pensions: Does pension accounting lead investors and analysts astray?Accounting Review 81(4): 925–55.

58.

Rauh

Joshua D.

Stefanescu

Irina

Zeldes

Stephen P.

2017. Cost saving and the freezing of corporate pension plans. Stanford University Graduate School of Business Research Paper No. 16-4.

59.

Shapiro

Susan P.

2005. Agency theory. Annual Review of Sociology 31: 263–84.

60.

Shin

Jae Yong

Seo

Jeongil

. 2011. Less pay and more sensitivity? Institutional investor heterogeneity and CEO pay. Journal of Management 37(6): 1719–46.

61.

Shleifer

Andrei

Summers

Lawrence H.

1988. Breach of trust in hostile takeovers. In Auerbach

Alan J.

(Ed.), Corporate Takeovers: Causes and Consequences, pp. 33–68. Chicago: University of Chicago Press.

62.

Solomon

Deborah

Hawkins Jr.

Lee

2005. Pension inquiry shines spotlight on assumptions [online version]. Wall Street Journal, November9, p. C1. Accessed at http://www.wsj.com/articles/SB113150828703592018 (June 2015).

63.

Towers Watson. 2010. Towers Watson–Forbes insights: 2010 pension risk survey. New York: Towers Watson.

64.

Useem

Michael

. 1996. Investor Capitalism: How Money Managers Are Changing the Face of Corporate America. New York: Basic Books.

65.

Walsh

Mary Williams

Hakim

Danny

. 2012. Public pensions faulted for bets on rosy returns [online version]. New York Times, May27. Accessed at https://www.nytimes.com/2012/05/28/nyregion/fragile-calculus-in-plans-to-fix-pension-systems.html (June 2015).

66.

Waring

M. Barton

. 2012. Pension Finance: Putting the Risks and Costs of Defined Benefit Plans Back under Your Control. Hoboken, NJ: John Wiley & Sons.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.13 MB