Defined Benefit Pension Plans and Cost of Equity. Abstract

Transcription

1 Defined Benefit Pension Plans and Cost of Equity Abstract This paper extends prior research on defined benefit plans by investigating the effect of pension risk on a firm s cost of equity. In particular, this study evaluates how pension intensity influences a defined benefit plan firm s implied cost of equity. Using proxies to measure pension intensity, and audit quality, we document that a defined benefit plan s cost of equity increases with the scaled magnitude of the firm s pension intensity risk, but that audit quality moderates this relation. We rationalize these findings by asserting that the positive relation between a sponsoring firm s cost of equity and pension intensity reflects increased financial risk associated with higher pension obligations, but that the quality of the auditor reduces this risk. This study also documents a post-sfas 158 decrease in the cost of equity, suggesting that investors require lower returns due to the effect of more transparent accounting for defined benefit plans, including the recognition of defined pension obligations. Keywords: defined benefit pension plans, cost of equity, pension intensity risk, audit quality, SFAS158 1

2 1. Introduction Employer-sponsored retirement plans generally are classified as either defined contribution or defined benefit pension plans (DBP). In the former, contributions are defined and the value of the pension at retirement is determined according to the magnitude of the contributions made and the investment performance of the pension fund. Hence the employee assumes the investme nt risk of the fund. In the latter, the investment risk resides with the employer and the present value of the pension benefits (the projected benefit obligation) is defined ex ante the employee s retirement and the employer assumes the fund s investment risk. In light of their long-term financial and social importance, defined benefit pension plans have continued to attract significant political, financial reporting, and academic scrutiny. Created under the 1974 Employee Retirement Income Security Act (ERISA), the Pension Benefit Guaranty Corporation (PBGC) insures employee pension benefits up to a maximum guaranteed amount. 1 In 2015 the PBGC paid $5.6 billion in claims to failed single employer plans and its deficit increased to $76 billion (PBGC 2015a). Since its enactment, ERISA has been amended several times and the Pension Protection Act of 2006, among other things, increased minimum employer contributions for at risk plans identified according to funded 1 The maximum guaranteed amount is based primarily on the age of the employee on the plan termination date or the date the employer entered bankruptcy. For example at age 65, an employee of a single employer plan terminating in 2015 is guaranteed a maximum monthly of $5, (PBGC 2015b). 2

3 status. The recent 2015 U.S. budget proposal includes a significant increase in the premiums paid by employers to the PBGC. 2 On the financial reporting front in 2006, SFAS 158, Employers Accounting for Defined Benefit Pension and Other Postretirement Plans, an Amendment of FASB Statements Nos. 87, 88, 106, and 132(R) was implemented. SFAS 158 requires that firms with defined benefit plans recognize funded status of plans on the face of the balance sheet (FASB 2006). Prior to SFAS 158 funded status was disclosed in the notes (FASB 1985). Funded status is measured as the difference between the current fair market value of the plan assets and the projected benefit obligation (PBO) with the PBO being equal to the actuarial present value of future expected benefits. An important objective of SFAS 158 was to enhance pension accounting financial reporting transparency. In concert with these political and financial reporting initiatives, academic research has examined financial characteristics unique to DBP firms across related contexts. A unifying theme of these studies is the investigation and assessment of factors related to the financial efficacy and integrity of defined benefit pension obligations as well as how pension obligations affect management decisions. For example, Jones (2013) provides evidence that debt contracting incentives may affect plan assumptions including higher discount rates and lower expected increases in employee salaries to increase funded status after the implementation of accounting 2 The PBGC is funded in part by premiums from employers that sponsor DBP plans. Currently, premiums are assessed according to number of employees and a variable weight component that is based on the percentage of a plan s underfunded liability. 3

4 standard SFAS 158. Rauh (2009) uses a risk management hypothesis to explain results that show financially weak (strong) DBP firms invest plan assets into less (more) risky assets. Sasaki (2015) reports that pension losses decrease the capital expenditures of defined benefit plan firms and Chaudhry et al. (2016) provide evidence that investment into firm projects is associated with the level of underfunding. Cadman and Vincent (2015) report that powerful CEOs receive abnormally high pension benefits. Additional studies also have considered the relation between pension fund characteristics and the value of the sponsoring firm. Alderson and Seitz (2013) document that shareholders lose value to the pension plan when it is undiversified across asset classes and Jin et al. (2006) report that equity betas of DBP firms reflect the betas of their pension assets. Phan et al. (2015) report that earnings management associated with mandatory cash contributions into the plan does not significantly lower the cost of capital. Using a sample of Japanese firms, Nakajima and Sasaki (2010) document that firms with large unrecognized obligations earn lower returns. Jensen and Meckling (1976) assert that corporations may be considered as a nexus of projects whose worth is determined by the difference in the returns earned from those projects less the cost of acquiring the capital used to finance them. A foundational determinant of firm value is, therefore, the return required by equity investors and, ceteris paribus, the lower (higher) the required return, the higher (lower) the residual return on a company s invested capital. Hence, firm value changes not only according to the explicit financial outcome of projects but also in juxtaposition with the cost of the financial capital required to fund those projects. Prior studies document causal factors related to the implied cost of equity (COE), including a firm s level of financial risk (Hail and Leuz 2009). We conjecture that since firm level financial risk generally 4

5 increases COE and this risk increases with the scaled magnitude of the firm s pension commitments, the larger the DBP (scaled) the higher the company s COE, ceteris paribus. In particular our study alternatively defines pension intensity risk as (1) the projected benefit obligation of the plan and (2) the fair market value of the plan s assets, both scaled by the total assets of the sponsoring firm. We further proxy for pension intensity risk by computing (3) the yearly employer cash contributions to the plan scaled by operating cash flow. Using extant research, we estimate the implied cost of equity as the average of four well-established earnings growth-residual income models prescribed in cost of equity studies (Claus and Thomas 2001; Gebhardt et al. 2001; Easton 2004; Ohlson and Juettner-Nauroth 2005; Hail and Leuz 2009). Results of our empirical tests support our assertion that pension intensity increases COE. Prior research also provides extensive evidence of the favorable effects of utilizing high quality auditors, including a lower cost of equity (Easton 2004; Khurana and Raman 2004). In the spirit of these studies, we further evaluate how audit quality affects our reported positive relation between a firm s pension intensity risk and COE. Using BIGN auditors and audit specialists as proxies for audit quality, we find that utilizing high quality auditors attenuates the positive relation between pension intensity and COE. 3 This study contributes to the current body of DBP literature in several ways. While prior DBP studies have investigated factors across diverse and relevant valuation contexts (Phan et al. 2015; Alderson and Seitz 2013), our study provides new information pertaining to the relation between 3 Another proxy utilized in audit quality studies is the tenure of the auditor. However, findings of research using this measure are mixed as certain studies document audit quality increasing with the tenure of the auditor (Myers et al. 2003) and others report a negative relation between audit quality and the length of the auditor client relationship (Carcello and Nagy 2004). In light of these conflicting results we restrict our measure of audit quality to BIGN and audit specialist auditors. 5

6 pension intensity and the sponsoring firm s cost of equity. Since COE has direct and explicit valuation implications, insight into this relation is especially relevant for evaluating DBP firms. Finally, this paper informs the debate about the financial reporting efficacy of recognizing items on the face of the financial statements versus disclosing them in the notes by comparing DBP firms cost of equity before and after the implementation of SFAS 158. Prior research documents that investors perceive recognized financial reporting values to be more pertinent than disclosed values (Ahmed et al. 2006). Davis-Friday et al. (2004) provide evidence that shows recognized retiree benefits other than pensions have less of an effect on market valuation than disclosed liabilities. In a more recent study, Coronado et al. (2008) provide findings suggesting that, prior to SFAS 158, markets did a poor job of pricing pension accruals disclosed in the notes. In contrast to these results, however, Beaudoin et al. (2010) utilize value relevant models to provide evidence that post-sfas 158 recognized pension amounts provide no incremental informational value while Mitra and Hossain (2009) show a negative relation between the levels and changes in stock returns and the magnitude of SFAS 158 transition adjustments and use these findings to argue that investors evaluate accounting information more effectively when recognized rather than disclosed. SFAS 158 was the first part of a comprehensive two-phase project in accounting for DBP firms. Preceding the implementation of SFAS 158, the SEC issued a report that emphasized the need for less opaque pension accounting (SEC 2005). Hence a salient objective of this standard was an increase in financial reporting transparency. An additional criterion generally articulated by the FASB for amending and creating standards is that the benefits of implementation should exceed 6

7 its costs. In their summary statement of SFAS 158, the FASB states that The Board recognizes that employers will incur costs to implement this Statement. However, the Board believes that the expected benefits outweigh the costs (FASB 2006). This study looks back over the years before and after the implementation of SFAS 158 to investigate its effectiveness in achieving the FASB s purported goals. In sympathy with our theoretical disposition and the FASB s objective of greater reporting transparency we predict a post-sfas 158 decrease in COE. After controlling for financial performance and other variables associated with COE, results of this study document a post-sfas 158 decrease in the cost of equity. Results further provide marginal support for the notion that SFAS 158 decreases the pension intensity and cost of equity relation. Our paper is organized as follows: section 2 includes relevant literature, along with hypotheses and their motivations; section 3 describes the sample and methodology used to test our hypotheses; section 4 provides results of tests; and section 5 contains concluding remarks. 2. Related Literature, Motivation and Hypotheses Implied Cost of Equity Recent studies utilize earnings growth and residual income models to estimate the ex-ante cost of equity across numerous contexts. For example, Gebhardt et al. (2001) use a discounted residual income model to assert that a firm s implied cost of equity is a function of factors that include industry membership, book-to-market ratio, and the forecasted long-term growth rate. Their study provides results that explain approximately 60% of the cross sectional variation in future years cost of equity. They contend that since risk varies over time, using implied returns in lieu 7

8 of historical returns results in more precise risk estimates. Claus and Thomas (2001) utilize an abnormal earnings (residual income) model to document lower equity risk premiums. Using a sample of ADR 4 firms, Hail and Leuz (2009) employ models to estimate firms implied cost of equity to document that countries with extensive securities regulation and enforcement mechanisms produce lower cost of equity. Using a similar approach, Wang et al. (2016) use implied COE models to document that the government-induced injection of liquidity into the banking system reduces the industry-wide cost of equity. Li et al. (2012) find that XBRL (extensible business reporting language) adoptions reduce the cost of equity and this effect increases for smaller firms, high growth firms, and firms with low analyst coverage. Pension Intensity Risk From a valuation perspective, it may be argued that DBP pension benefits incentivize performance and enhance employee morale. However, compared with other operating liabilities that reflect the acquisition of financial capital for revenue enhancing capital investment, the explicit impact of pension liabilities on a firm s operating capacity and capabilities is relatively subdued. Indeed, Rauh (2006) reports results that suggest sponsors contributions to underfunded plans inhibit investment into capital projects. In addition, financial economic theory suggests that a firm s cost of equity reflects, at least in part, the financial risk of the firm (Hail and Leuz 2009). In a pre-sfas 158 study, Dhaliwal (1986) documents that unfunded pension liabilities are treated as long-term debt. Since the projected benefit obligation (PBO) estimates may be large in magnitude and require forecasts 4 ADR is American depositary receipt. 8

9 over long time horizons, their financial materiality and effects can be significant. In its recently filed K, Abbott Labs reported that in 2016 it will contribute $576 million to its defined benefit plan (Abbott 2015) and General Motors recently announced that it would use $2 billion in proceeds from newly issued debt to shore up the funded status of its plan (Bloomberg 2016). Since increasing pension obligations impose greater financial risk, the higher the PBO (scaled), the greater the pension intensity risk. 5 We, therefore, assert that increases (decreases) in pension intensity risk should increase (decrease) a DBP firm s cost of equity. We test this assertion with the following hypothesis: H1: Pension intensity risk increases a DBP firm s implied cost of equity. Audit Quality Prior studies across numerous contexts document that large audit firms and auditor industry specialists perform higher quality audits (Palmrose 1988; Teoh and Wong 1993; Francis et al. 1999; Menon and Williams 2004; Balsam et al. 2003). Becker et al. (1998) document that clients of large audit firms report lower discretionary accruals. Balsam et al. (2003) show that large auditors and industry specialist auditors report lower accruals and higher earnings response coefficients. Utilizing a price-earnings growth (PEG) ratio model, Easton (2004) and Khurana and Raman (2004) report that Big 4 auditors reduce their client s cost of equity. Choi and Lee (2014) find that the inverse relation between the choice of a Big N auditor and the implied cost of equity is greater for multi-segment firms. Chen et al. (2011) use a sample of Chinese firms to document an inverse relation between the cost of equity alternatively measured using the PEG 5 For the sample of firms in our study, the average PBO scaled by total assets is 16.3%. 9

10 ratio and industry method (Gebhardt et al. 2001) models and audit quality (Big N auditors), but find this result is limited to a small subset of non-state-owned firms. Using large audit firms and industry specialist auditors as alternative proxies for audit quality, we extend prior research by evaluating how audit quality influences the effect of pension intensity risk on COE. Although we expect COE to change in concert with pension intensity risk, we also conjecture that audit quality attenuates this positive relation and test the following hypotheses: H2a: The magnitude of the positive relation between the cost of equity and pension intensity risk decreases for clients of large audit firms. H2b: The magnitude of the positive relation between the cost of equity and pension intensity risk decreases for clients of industry specialists audit firms. SFAS 158 Proponents of market efficiency assert that the position and prominence of accounting values in the financial reports are irrelevant since informed and capable investors are able to quickly impound all publicly available value-relevant information into a firm s stock price regardless of its location. Other studies provide evidence, however, that investors and other users may process information differently according to whether it is recognized or disclosed, and information is more easily reflected in stock price when it is recognized in the statements rather than disclosed in the notes. A recent study of banks shows that the significance of the relation between changes in stock prices and fair values of derivative financial instruments increases when values are recognized instead of disclosed (Ahmed et al. 2006). Using survey results from a sample of

11 commercial lenders, Harper Jr et al. (1991) find that the decisions of lenders are influenced more heavily by liabilities recorded on the balance sheets of loan applicants than by liabilities merely disclosed in the notes. A study by the Divisions of Research & Statistics and Monetary Affairs of the Federal Reserve Board provides empirical evidence that prior to SFAS 158 investors misvalued defined benefit pensions, inducing sizable errors in the value of the sponsoring firm. The authors argue that SFAS 158 should improve the ability of investors to value DBP firms (Coronado et al. 2008). Other research suggests that using the full funded status of defined benefit plans may be useful for valuation (Trivedi and Young 2006). Magni et al. (2007) argue that less opaque and more transparent financial reports have value enhancing effects. Barth et al. (2008) assert that transparency is a desirable characteristic of financial reporting and Barth et al. (2013) provide evidence that firms with more transparent earnings enjoy a lower cost of capital. As stated previously, SFAS 158 requires that DBP firms recognize their pension plan funded status on the face of the balance sheet (rather than disclosing in the notes) and a salient objective of this change was to increase pension accounting financial reporting transparency (FASB 2006). We assert that to the extent that the recognition of pension liabilities on the balance sheet, which previously were disclosed in the notes, enhance the transparency and quality of pension accounting, defined benefit plan firms should enjoy a reduction in their cost of equity. We test this conjecture as follows: H3a: The cost of equity decreases after the implementation of SFAS 158. Similar to the hypothesized transparency effect of SFAS 158 on COE, we also posit a post-sfas 158 decrease in the positive relation between COE and pension intensity. That is: 11

12 H3b: The magnitude of the positive relation between pension intensity and the cost of equity decreases after the implementation of SFAS Sample and Methodology Our study uses data of DBP firms listed in the CRSP and Compustat files for years 1999 to All continuously measured variables are winsorized at the 1% level and all models include fixed effects for industry and year. After deleting firms with missing data for our models, the total number of firm year observations for the scaled projected benefit and fair market value pension intensity variables is 11,638 firm year observations. The total number of firm year observations for our cash contribution pension intensity variable is 10,221. Dependent Variable: Implied Cost of Equity Following Hail and Leuz (2009), Wang et al. (2016), we measure our dependent variable, implied cost of equity, using the four models prescribed by Claus and Thomas (2001), Gebhardt et al. (2001), Easton (2004), and Ohlson and Juettner-Nauroth (2005). Since these models differ with respect to their applications of certain assumptions including earnings growth expectations, industry membership, forecast horizons, etc., to mitigate the idiosyncratic effect of measurement error on our dependent variable, we follow Hail and Leuz (2009) and measure our dependent 12

13 variable, COEit, for each firm i as the end of fiscal year t mean of the four estimates from each model. Specifications for each model are included in Appendix A. Variables of Interest Pension Intensity The larger the pension plan and its attending commitments relative to the size of the sponsoring company, the greater the pension intensity risk and the greater its potential effect on the financial position of the firm. For each firm i at the end of fiscal year t, we alternatively measure pension intensity risk using the following proxies: the projected benefit obligation and the fair market value of the plan s assets both scaled by total assets (PBOit and FMVit, respectively). In addition we also include the employer s year t cash contribution to the plan scaled by operating cash flow CASHit. 6 A positive sign on the coefficients for these variables would provide support for hypothesis 1. Audit Quality To assess the effect of audit quality on our cost of equity variable, we use an indicator variable, BIGNit, equal to 1 if the audit firm is one of the largest 4, 5, or 6 audit firms (depending on the time period) and 0 otherwise. Similarly, audit industry specialist dummy, SPECit, is coded 1 if the audit firm is the leading year t auditor according to sales and number of clients for each twodigit SIC code industry group and 0 otherwise. 6 Pension plans also may be considered according to their funded status. However, funded status nets pension assets and obligations against each other, which conceals the true risk of the pension obligation relative to the size of the firm. Separating these scaled components provides a more veritable measure of pension risk. 13

14 To evaluate the effect of pension intensity on COE conditional on the quality of the auditor, we interact our BIGNit auditor dummy with each of our pension intensity risk variables (BNxPBOit, BNxFMVit, BNxCASHit). We do the same with our SPECit audit industry specialist dummy (SPECxPBOit, SPECxFMVit, SPECxCASHit). A negative sign on these interaction terms would provide support for hypotheses 2a and 2b. SFAS 158 To test our assertion that the cost of equity decreases after the implementation of SFAS 158, we include an indicator variable equal to 1 (SFAS 158) for the years 2007 to 2014 and 0 otherwise. A positive sign on this indicator variable would support hypothesis 3a. To evaluate the effect of pension intensity on COE conditional on SFAS 158, we interact our SFAS 158 dummy with each of our pension intensity risk variables (SFAS158xPBOit, SFAS158xFMVit, SFAS158xCASHit). Decreases in these post-sfas 158 coefficients would support hypothesis 3b that the effect of pension intensity on a DBP firm s cost of equity decreases in the post-sfas 158 period. Control Variables: Prior studies document causal factors related to the implied cost of equity capital, including size (Banz 1981), price volatility (Wang et al. 2016), dividend policy (Dhaliwal et al. 2007), book-tomarket ratio (Lakonishok et al. (1994), audit quality (Choi and Lee 2014). To control for these and other factors related to our dependent variable, COEit, and pension intensity variables of interest, we include the following controls. 14

15 In accordance with prior cost of equity studies that control for size effects (Dhaliwal et al. 2007; Choi and Lee 2014), for each firm i we include the end of fiscal year t log of total assets (LnASSETit). Firm size proxies for several potential factors including more sophisticated financial reporting, greater investor visibility and scrutiny, reduced information asymmetry. Hence, we expect the sign on LnASSETit to be negative. We also control for the effect of returns variability risk on our COEit variable by including the annualized standard deviation of monthly returns (VOLTit). The expected sign on VOLTit is expected to be positive. Prior studies suggest that since dividends are subject to double taxation, investors may prefer share buybacks as an alternative to dividend payouts. However, dividend-related tax incentives may induce institutional investors to favor cash dividends. Litzenberger and Ramaswamy (1979) report higher returns for higher dividend paying companies and contend that these returns are required by investors since dividends are undesirable. Dhaliwal et al. (2007) document reduced cost of equity around tax cuts. In accordance with these studies, we include the variable DIVit and measure it as the dividend per share of common divided by the end of fiscal year price per share. It is well known that high (low) market-to-book firms earn low (high) stock returns (Lakonishok et al. 1994; Houmes and Skantz 2010). Gebhardt et al. (2001) contend that these returns reflect the respective increases and decreases in risk premiums commanded by high and low valued equity. To control for these results, we include variable MBit and measure it as the firm i end of fiscal year t price per share divided by book value per share. To control for changes in operational factors induced by the size of the workforce and the corresponding effects of these changes on employer pension obligations, we include the variable EMPit and define it as the log 15

16 of the total number of employees. Although our study makes no explicit theoretical predictions regarding the effect of employee size on our dependent variable, cost of equity, we conjecture that since the effect of pension intensity on cost of equity should increase with the size of the workforce, the expected sign on this variable is positive. Since profitability enhances firm value, the cost of equity should be reduced (increased) for more (less) profitable firms. We control for this inverse relation by including profitability measure ROEit equal to firm i s end of fiscal year t net income divided by stockholders equity. We also control for financial stress by including the firm s Z score (Zit) (Altman 1968). Higher Z scores denote a lower likelihood of bankruptcy. We therefore expect the coefficient on this variable to be negative. In addition, we also include control for the potential effect of a firm s cost structure on our results. We measure this variable, LEVit, as the ratio of a DBP firm s total long-term debt scaled by its total assets. According to Modigliani and Miller (1958), firm value is indifferent to financial leverage and the expected return in levered firms will be a linear increasing function of leverage. We predict a positive relation between leverage and COE. Our general models are specified as follows: COEit = α0 + α1contlit + α2piit + Industry Effects + Year Effects + εit (1) COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5piit + Industry Effects + Year Effects + εit (2) 16

17 COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5piit + α6bnitxpiit + Industry Effects + Year Effects + εit (3) COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5piit + α6specitxpiit + Industry Effects + Year Effects + εit (4) COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5piit + α6sfas158xpiit + Industry Effects + Year Effects + εit (5) Where COEit is our above described cost of equity measure, CONTLit represents controls, BIGNit and SPECit denote audit quality variables, SFAS158 is our SFAS 158 indicator variable, and PIit represents the three pension intensity proxies. 4. Results Table 1 provides definitions for model variables. //Insert Table 1 About Here// Descriptive Statistics Table 2 shows the sample distribution. The distribution of firm year observations is relatively uniform over the 1999 to 2014 period of our study with the greatest number of observations (7.20%) occurring in year 2013 and the least (4.95%) in the first year, The table also reports the number of sample observations for each of the Fama French 12 industry groups 17

18 represented in our sample. Manufacturing firms represent the highest percentage of firms (22.36%), followed by business equipment at 12.2 %. The lowest industry group percent is finance, at 1.86%. //Insert Table 2 About Here// Table 3, Panels A and B, reports means and medians for model variables. Panel A reports both the mean cost of equity capital for each of the four cost of equity models Claus and Thomas (2001), Gebhardt et al. (2001), Easton (2004), and Ohlson and Juettner-Nauroth (2005) and the overall mean cost of equity from these models, which we use to measure our dependent variable, COEit. The overall mean (median) cost of equity capital is 8.6% (7.7%). Panel B reports descriptive statistics for our variables of interest and control variables. Regarding our pension intensity variables of interest, the average projected benefit obligation as a percentage of assets is 16.3 percent. Employers contribute an average 9.9% of their yearly operating cash flow to their plans and the mean fair market plan value as a percentage of the firm s total assets is 13.8%. Regarding our audit quality variables, 95.8% of our sample of firms use BIGN auditors and 27.1% use industry specialists. The mean (median) dividend yield for our sample of DBP firms is 1.6% (1.1%). The mean (median) market to book ratio is 3.01 (2.14). The mean (median) Z score for our sample of firms is 2.51 (2.40). 7 The average total long-term debt as a percentage of assets is 23.1%. //Insert Table 3 About Here// 7 Altman Z scores may be ranked according to likelihood of bankruptcy; i.e., high (< 1.81), medium (1.81 < Z < 2.99) and low (>2.99) probabilities of bankruptcy. For more details see Appendix B. 18

19 Table 4, Panel A, provides Pearson correlations for the COE models estimates and all estimates are statistically significantly correlated. Table 4, Panel B, includes univariate correlations for other model variables. Firm size (LnASSETit), market to book (MBit) return on equity (ROEit), and our financial distress variable (Zit) are negatively correlated with the cost of equity. Volatility (VOLTit), dividend yield (DIVit), and leverage (LEVit) are positively correlated with COEit. All pension intensity variables, PBOit, FMVit, and CASHit, are positively correlated with the cost of equity and both audit quality proxies are negatively correlated with COEit. //Insert Table 4, Panels A and B About Here// Regression Results for the Effect of Pension Intensity on a DBP Firm s Implied Cost of Equity Pension Intensity For our sample of DBP firms, Table 5, columns 1, 2, and 3, provides results for controls and the main effects of pension intensity and SFAS 158 on the cost of equity without and with audit quality variables (equations 1 and 2). Audit quality variable BIGNit is significantly negative and SPECit is not significant. Regarding our pension intensity variables of interest, the cost of equity increases across all of our measures of pension intensity, PBOit, FMVit, and CASHit, with p values <.001. Hence after controls, the cost of equity capital increases with the magnitude of a firm s scaled level of projected benefit obligation, fair market value of pension plan assets, and employer cash contributions. These results provide support for hypothesis 1. 19

20 SFAS 158 As stated previously, SFAS 158 requires that DBP firms recognize pension plan funded status on the face of the balance sheet (rather than disclosing in the notes). A purported objective of the standard was to increase pension accounting financial reporting transparency. We conjecture that the increased visibility from the inclusion of pension liabilities on the face of the balance sheet that previously were relegated to the notes facilitated this objective, resulting in a post-sfas 158 reduction in the cost of equity. Results support this assertion. The SFAS 158 variable of interest shows a post-sfas 158 decrease in the cost of equity capital and these findings are robust across all models (p <.001.) Results provide support for hypothesis 3a. Controls Variables As per prior studies, controls variables generally are significant in the expected directions and conform with previously reported univariate results. For our sample of DBP firms, the implied cost of equity capital decreases with firm size, market to book, and return on equity, and increases with volatility, dividend yield, financial distress, and leverage. Hence investors require lower costs of equity capital for larger firms with less information asymmetry and more sophisticated financial reporting infrastructures. In addition investors require lower returns for profitable companies (as measured by their return on equity) and for highly valued, high marketto-book growth companies. The significantly positive coefficient on our variable, VOLT it, reflects how COE increases with a DBP firm s dispersion of returns. Also, as per Dhaliwal et al. (2007), COE increases for high dividend paying companies, which may, in part, reflect the effect of the tax penalized portion of the dividend yield on equity value. The significantly positive (negative) coefficients for LEV it (Z it ) reflect the impact of cost structure and financial stress on COE. 20

21 Further, our findings also show a statistically significantly positive relation between number of employees and the cost of equity capital, suggesting that pension obligations associated with the size of the workforce affect COE. //Insert Table 5 About Here// Regression Results for Effect of Pension Intensity on the Cost of Equity Conditional on Audit Quality Tables 6 and 7 provide results for our tests of pension intensity on the cost of equity conditional on the BIGN and audit specialist audit quality effects, respectively (equations 3 and 4). For our BIGN models (Table 6), all controls generally are significant in the expected direction and all pension intensity variables remain significantly positive. Further, all BIGN auditor-pension intensity interaction terms, BNxPBOit, BNxFMVit, and BNxCASHit, are significantly negative with p values of p <.001, p <.010, and p <.001, respectively (Hypothesis2a). //Insert Table 6 About Here// Although our audit specialist interaction terms are similarly negative, variables SPECxPBOit and SPECxFMVit are significant at the p <.10 level, while SPECxCASHit is not significant. Furthermore, Tables 5, 6, and 8 generally report that auditor specialist variable, SPECit, is not significant. Fernando et al. (2010) document that auditor size and specialization are inversely related to the client firm s implied cost of equity, but that this relation is confined to smaller 21

22 firms. Univariate results in Panels A of Table 6 and 7 report significantly lower cost of equity for both BIGN and specialists auditors. However, 96% (n = 11,149) of the firms in our sample are audited by BIGN auditors and 27% (n = 3,159) are audited by specialists auditors. Hence, our BIGN audit quality variable may be subsuming the relation between COE and our audit specialist dummies. Taken together, however, these overall findings suggest that the enhanced audit quality associated with large audit firms and specialists auditors attenuates the negative effect of pension intensity risk on the cost of equity capital (H2b). //Insert Table 7 About Here// In addition to our primary pension intensity tests, we also further examine the pension intensity effect of SFAS 158 on DBP firms cost of equity. Consistent with the results in Tables 5 7, Panel B of Table 8 reports a post-sfas 158 decrease in the cost of equity and provides support for hypothesis 3a. We further hypothesize that the previously documented increase in the cost of equity induced by higher pension intensity decreases after SFAS 158 (H3b). Our results provide marginal support for this assertion as SPECxPBOit and SFAS158xFMVit, are negative and SFAS158xCASHit is positive. As previously reported, our pension intensity cash contribution variable, CASHit, is significantly negative across all of our main tests. Our cash flow-related variable reflects cash contributions that are not explicitly related to the balance sheet. While we believe that our CASHit variable is an appropriate proxy for pension intensity, it may be worth noting in this regard that SPECxPBOit and SFAS158xFMVit better capture the hypothesized balance sheet related transparency effect of SFAS

23 A logical extension of our analysis is a comparison of the pre- and post-sfas 158 audit quality- pension intensity interaction terms. Since our findings show no significant difference between these two periods, to avoid unnecessary clutter, we do not formally report these results. //Insert Table 8 About Here// Robustness To assess the robustness of our findings we conduct the following tests. Although our models control for financial distress, the effect of SFAS 158 as well as other related factors we test for in our models may differ when the firm is financially constrained. We therefore investigate the potential effects of financial distress on our results by separating our overall sample into high Z score (firms with low financial distress) and low Z score (firms with high financial distress) portfolios. We then run separate regressions of COE on each of our pension intensity variables using the following models: COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5pboit + α6bnitxpboit + α7specitxpboit + α8sfas158xpboit + Industry Effects + Year Effects + εit (6) COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5fmvit + α6bnitxfmvit + α7specitxfmvit + α8sfas158xfmvit + Industry Effects + Year Effects + εit (7) COEit = α0 + α1contlit + α2bignit + α3specit + α4sfas158 + α5cashit + α6bnitxcashit + α7specitxcashit + α8sfas158xcashit + Industry Effects + Year Effects + εit (8) 23

24 We report these findings in Table 9. Results confirm our earlier findings as SFAS158 remains negative and all pension intensity variables are positive across both portfolios. In addition, our audit quality-pension intensity interaction terms generally are similarly negative. 8 //Insert Table 9 About Here// Other studies use market model betas to control for market systematic risk. While we believe our proxy for total returns variability is theoretically preferable, in sympathy with other cost of equity studies we replace our variable VOLT with BETA and rerun our models. The direction and significances of our variable of interests are unchanged. Implemented in 2006, SFAS 158 mandates that DBP firms recognize pension liabilities (funded status) on the face of the balance sheet (FASB 2006). Prior to this standard the funded status was reported in the notes. Jones (2013) reports a post-sfas 158 improvement in the funded status and explains this finding by asserting that DBP firms with contracting incentives modify assumptions used to estimate retirement obligations. In addition there has been a post-sfas 158 decrease in the number of firms offering DBPs. To assess the sensitivity of the results of our main pension intensity and audit quality tests to these recent changes we rerun all our models for the period after the implementation of SFAS 158; i.e., 2007 to Findings over this alternative period do not change. 8 To avoid clutter we do not formally report results of the following robustness tests. Results will be provided upon request. 24

25 As stated previously, to mitigate potential idiosyncratic measurement errors that may be unique to each COE model, we use the average of these models as our dependent variable, COEit. To assess the sensitivity of our results across all models, we rerun our pension intensity and audit quality tests using the specific estimates from each model. Results of these individual tests do not alter our findings. In addition, to reduce the effect that the financial crisis may have had on our findings we eliminate firm year observations from our sample and rerun our main tests. Once again this change does not affect our main results. 5. Conclusion This study extends the current body of literature on defined benefit plans by investigating the effect of pension intensity risk on a DBP firm s implied cost of equity capital. In addition, this study includes new evidence on the favorable role that high quality auditors may play in reducing COE for high pension intensity DBP firms. Using proxies for pension intensity, tests of this study provide evidence that the cost of equity increases with a firm s pension intensity, but that audit quality attenuates this relation. Our research additionally investigates the transparency effect of SFAS 158 on COE. After controlling for financial performance and other variables associated with cost of equity, results document a post-sfas 158 decrease in the cost of equity. Results further provide marginal support for the notion that the cost of equity decreases the pre- SFAS 158 positive relation between pension intensity and cost of equity. A limitation of this research is that the sample is limited to DBP companies only. Although results of our tests show similar results for controls that are used in other COE related studies 25

26 that include DBP and non-dbp firms, the generalizability of our findings may be limited to the extent that the financial characteristics of firms with defined contribution plans differ from our sample. Nevertheless, the results of this study provide evidence that suggests pension intensity risk affects a sponsoring firm s cost of equity capital. Hence, initiatives to modify current plans or to eliminate them entirely may have valuation implications. The number of firms offering DBP has decreased over recent years. Potential explanations for this decrease include cost savings from reduced employee compensation (Rauh 2006), the investment risk of the plan (Munnell 2006), and the recognition of funded status of the plan on the balance sheet (Levine and Golmbic 2006). While the efficacy of these findings is beyond the empirical scope of our paper, it may nevertheless be conjectured that to the extent a DBP s pension intensity risk affects financial capital costs, analyses of both their risk and magnitude should provide additional insight into companies motivations to terminate. A potential outcome of discontinuing employer sponsored defined benefit plans could be a change in the DBP firm s cost of equity capital. Further studies that investigate changes in COE over periods before and after the plans terminations could provide additional relevant insight. 26

27 References Abbott Annual Report: Abbott. Ahmed, A. S., E. Kilic, and G. J. Lobo Does recognition versus disclosure matter? Evidence from value-relevance of banks' recognized and disclosed derivative financial instruments. The Accounting Review 81 (3): Alderson, M. J., and N. L. Seitz Pension Policy and the Value of Corporate Level Investment. Financial Management 42 (2): Altman, E. I Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. The Journal of Finance 23 (4): Balsam, S., J. Krishnan, and J. S. Yang Auditor industry specialization and earnings quality. Auditing: A Journal of Practice & Theory 22 (2): Banz, R. W The relationship between return and market value of common stocks. Journal of Financial Economics 9 (1):3-18. Barth, M. E., Y. Konchitchki, and W. R. Landsman Cost of capital and earnings transparency. Journal of Accounting and Economics 55 (2-3): Barth, M. E., W. R. Landsman, and M. H. Lang International accounting standards and accounting quality. Journal of Accounting Research 46 (3): Beaudoin, C., N. Chandar, and E. M. Werner Good disclosure doesn't cure bad accounting Or does it?: Evaluating the case for SFAS 158. Advances in Accounting 27 (1): Becker, C. L., M. L. DeFond, J. Jiambalvo, and K. Subramanyam The effect of audit quality on earnings management. Contemporary Accounting Research 15 (1):1-24. Bloomberg GM to issue debt to shore up U.S. hourly pension plan. Bloomberg website: Bloomberg. Cadman, B., and L. Vincent The role of defined benefit pension plans in executive compensation. European Accounting Review 24 (4): Carcello, J. V., and A. L. Nagy Audit firm tenure and fraudulent financial reporting. Auditing: A Journal of Practice & Theory 23 (2): Chaudhry, N., H. H. Au Yong, and C. Veld How does the Funding Status of Defined Benefit Pension Plans Affect Investment Decisions of Firms in the United States? Journal of Business Finance & Accounting. Chen, H., J. Z. Chen, G. J. Lobo, and Y. Wang Effects of Audit Quality on Earnings Management and Cost of Equity Capital: Evidence from China. Contemporary Accounting Research 28 (3): Choi, J. H., and W. J. Lee Association between Big 4 auditor choice and cost of equity capital for multiple segment firms. Accounting & Finance 54 (1): Claus, J., and J. Thomas Equity premia as low as three percent? Evidence from analysts' earnings forecasts for domestic and international stock markets. The Journal of Finance 56 (5): Coronado, J., O. S. Mitchell, S. A. Sharpe, and S. B. Nesbitt Footnotes aren't enough: The impact of pension accounting on stock values. Journal of pension economics and finance 7 (03): Davis-Friday, P., C.-S. Liu, and H. Mittelstaedt Recognition and disclosure reliability: Evidence from SFAS No Contemporary Accounting Research 21 (2):

28 Dhaliwal, D. S Measurement of financial leverage in the presence of unfunded pension obligations. Accounting Review 61 (4): Dhaliwal, D. S., L. Krull, and O. Z. Li Did the 2003 Tax Act reduce the cost of equity capital? Journal of Accounting and Economics 43 (1): Easton, P. D PE ratios, PEG ratios, and estimating the implied expected rate of return on equity capital. The Accounting Review 79 (1): FASB Statement of Financial Accounting Standards No. 87: Employers' Accounting for Pensions. Financial Accounting Standards Board (FASB) Statement of Financial Accounting Standards No. 158: Employers' Accounting for Definded Pension and Other Postretirement Plans. Financial Accounting Standards Board (FASB). Fernando, G. D., A. M. Abdel-Meguid, and R. J. Elder Audit quality attributes, client size and cost of equity capital. Review of Accounting and Finance 9 (4): Francis, J. R., E. L. Maydew, and H. C. Sparks The role of Big 6 auditors in the credible reporting of accruals. Auditing: A Journal of Practice & Theory 18 (2): Gebhardt, W. R., C. Lee, and B. Swaminathan Toward an implied cost of capital. Journal of Accounting Research 39 (1): Hail, L., and C. Leuz Cost of capital effects and changes in growth expectations around US cross-listings. Journal of Financial Economics 93 (3): Harper Jr, R. M., W. G. Mister, and J. R. Strawser The Effect of Recognition Versus Disclosure of Unfunded Postretiremen Benefifs on Lenders' Perceptions of Debt. Accounting Horizons 5 (3):50. Houmes, R. E., and T. R. Skantz Highly valued equity and discretionary accruals. Journal of Business Finance & Accounting 37 (1 2): Jensen, M. C., and W. H. Meckling Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3 (4): Jones, D. A Changes in the Funded Status of Retirement Plans after the Adoption of SFAS No. 158: Economic Improvement or Balance Sheet Management? Contemporary Accounting Research 30 (3): Khurana, I. K., and K. Raman Litigation risk and the financial reporting credibility of Big 4 versus non-big 4 audits: Evidence from Anglo-American countries. The Accounting Review 79 (2): Lakonishok, J., A. Shleifer, and R. W. Vishny Contrarian investment, extrapolation, and risk. The Journal of Finance 49 (5): Levine, D. N., and L. C. Golmbic Freezing Defined Benefit Pension Plans The Process & Post-Freeze Issues. Groom Law Group, Chartered 2006 [cited June ]. Available from Final.pdf. Li, O. Z., C. Ni, and Y. Lin Does XBRL adoption reduce the cost of equity capital? Available at SSRN Litzenberger, R. H., and K. Ramaswamy The effect of personal taxes and dividends on capital asset prices: Theory and empirical evidence. Journal of Financial Economics 7 (2): Magni, C. A., S. Malagoli, and A. Damodaran Information transparency and valuation: Can you value what you cannot see? Managerial Finance 33 (11):

29 Menon, K., and D. D. Williams Former audit partners and abnormal accruals. The Accounting Review 79 (4): Mitra, S., and M. Hossain Value-relevance of pension transition adjustments and other comprehensive income components in the adoption year of SFAS No Review of Quantitative Finance and Accounting 33 (3): Modigliani, F., and M. H. Miller The cost of capital, corporation finance and the theory of investment. The American Economic Review 48 (3): Munnell, A. H Employer-sponsored plans: The shift from defined benefit to defined: The Oxford Handbook of Pensions and Retirement Income. Myers, J. N., L. A. Myers, and T. C. Omer Exploring the term of the auditor-client relationship and the quality of earnings: A case for mandatory auditor rotation? The Accounting Review 78 (3): Nakajima, K., and T. Sasaki Unfunded pension liabilities and stock returns. Pacific-Basin Finance Journal 18 (1): Ohlson, J. A., and B. E. Juettner-Nauroth Expected EPS and EPS growth as determinantsof value. Review of accounting studies 10 (2-3): Palmrose, Z.-V Competitive Manuscript Co-Winner: An analysis of auditor litigation and audit service quality. Accounting Review 63 (1): PBGC. 2015a. Annual Report, edited by P. B. G. Corporation: Pension Benefit Guaranty Corporation. 2015b. Maximum Monthly Guarantee Tables, edited by P. B. G. Corporation: Pension Benefit Guaranty Corporation. Phan, H. V., H. Khieu, and J. H. Golec Does Earnings Management Relieve the Negative Effects of Mandatory Pension Contributions? Available at SSRN Rauh, J. D Investment and financing constraints: Evidence from the funding of corporate pension plans. The Journal of Finance 61 (1): Risk shifting versus risk management: Investment policy in corporate pension plans. Review of Financial Studies 22 (7): Sasaki, T The effects of liquidity shocks on corporate investments and cash holdings: Evidence from actuarial pension gains/losses. Financial Management 44 (3): SEC SEC Staff Report on Off-Balance Sheet Arrangements, Special Purpose Entities and Related Issues, edited by U. S. Securities and Exchange Commission. Teoh, S. H., and T. Wong Perceived auditor quality and the earnings response coefficient. Accounting Review: Trivedi, K., and G. Young Corporate Cross Holdings of Equity, Leverage and Pensions: Simulation and Empirical Evidence from the UK. The Economic Journal 116 (510):C190-C208. Wang, D., T. Ngo, O. A. Esqueda, and R. Houmes Impact of Government Bailout on Bank's Cost of Equity: Evidence from the Financial Bailout of (Working paper). 29