Printer Friendly

Further Evidence on the Association between Pension Plan Accounting and Firm Values: The Impact of SFAS No. 158.

I. Introduction

This paper studies whether pension accounting information under the existing accounting standard (SFAS No. 158) is reflected in a firm's stock price. We employ the method used in Coronado and Sharpe (2003) and Coronado, Mitchell, Sharpe and Nesbitt (2008) and extend their studies to subsequent years after the issuance of SFAS No. 158. Overall, we reach the same conclusion that a firm's stock price is overvalued, supporting the opaque view of pension plan accounting in the period from 1993 to 2005. However, after the issuance of SFAS No.158, we find that the association between stock prices and net pension assets significantly increases while the association between stock prices and pension earnings significantly decreases, implying that net pension assets (pension earnings) have become more (less) important in firm valuation. The results indicate a move towards a transparent view of pension plan accounting after the issuance of SFAS No. 158.

Many U.S. companies provide defined benefit plans (DB plans) to their employees. Although the number of defined benefit plans and the number of participants have declined over the years due to the popularity of defined contribution plans, for example, 401 (K), defined benefit plans hold large amounts of assets and play an important role in the U.S. economy. According to the statistics of private pension plans published by the U.S. Department of Labor, private sector defined benefit plans hold approximately three trillion dollars in assets in 2014 (DOL 2016). The accounting of defined benefit plans is complicated. The payment to employees is calculated based on the length of employment, salary history and some other factors. Companies are required to have enough assets to meet pension benefit obligations that are determined by the present values of future cash flow of estimated payments. Thus, it is important for companies and investors to evaluate the plan's funded status accurately. However, it is argued that the accrual accounting and the smoothing techniques used in pension accounting make it difficult for investors to evaluate the true funded status and the health of defined benefit plans (Coronado, Mitchell, Sharpe and Nesbitt 2008).

Coronado and Sharpe (2003) as well as Coronado et al. (2008) test an opaque model versus a transparent model using pension accounting information in the period from 1993 to 2005. According to Gold (2005), an opaque model implies that investors value a firm based on the information conveyed in its pension earnings, which are smoothed and stabilized, instead of net pension assets, which might be volatile but reflect the true economies. The results in Coronado and Sharpe (2003) as well as Coronado et al. (2008) show that defined benefit plans are largely overvalued and pension earnings are inflated in 1990s. Even during the early 2000s, when the market was very unstable and experienced large swings, investors still failed to accurately incorporate information on net pension assets into stock prices. To summarize, their results support the opaque model. One key reason, they show, is that the fair value information on pension assets and liabilities disclosed in the footnotes under SFAS No. 87 have been overlooked. (1)

In 2006, to improve the transparency and relevance of pension accounting, Financial Accounting Standard Board (FASB) issued Statement of Financial Accounting Standards (SFAS) No. 158, Employers' Accounting for Defined Benefit Pension and Other Postretirement Plans. It requires companies to recognize overfunded or underfunded statuses of a defined benefit plan by reporting net pension assets or net pension liabilities on the balance sheets and to recognize any change in the funded status through comprehensive income. However, it does not change the way to measure pension expenses that are recognized in earnings. SFAS No. 158 increases the transparency and visibility of pension assets and liabilities information by moving information from the footnotes to the balance sheets where investors would pay more attention when they process such information into stock prices. Thus, we expect to find a stronger association between stock prices and net pension assets on the balance sheets after the issuance of SFAS No. 158. In other words, we expect the pension accounting after SFAS No. 158 to be consistent with the transparent model.

Following Coronado et al. (2008), we construct a sample of firms that were included in the S&P500 index during the period 1993-2015. We collect the necessary financial statement information, pension information from COMPUSTAT, earnings forecasts and stock prices from l/B/E/S and CRSP. Following their research, we construct two primary measures of pension plan accounting. One is the Pension EPS constructed from the earnings information on the income statements, and the other is the Net Pension Asset Value (Pension NAV) constructed from the footnotes before SFAS No. 158 and recognized on the balance sheets after SFAS No. 158. We first replicate the main results in Coronado et al. (2008). We find that in the period 1993-2005, the Pension EPS is significantly associated with stock prices while the Pension NAV is not. Thus, we draw the same conclusion as Coronado et al. (2008) that investors relied more on the pension earnings information embedded in the income statement to establish firm values in the period of 1993 to 2005, consistent with the opaque model, because the fair value information on pension assets and liabilities disclosed in the footnotes did not draw investors' attention.

Then we investigate the impact of SFAS No. 158 on value relevance of pension earnings and net pension assets. We find that the association between stock prices and net pension assets significantly increases while the association between stock prices and pension earnings significantly decreases, implying that net pension assets (pension earnings) have become more (less) important in firm valuation. The results indicate a move towards a transparent view of pension plan accounting after the issuance of SFAS No. 158, which implies that the evolution of pension accounting standards in the past decade has improved investors' evaluation of defined benefit plans and firm values.

We also explore the changes in value relevance of pension earnings and net pension assets across various firm characteristics and conditions. Specifically, we examine the impact of SFAS No. 158 for different levels of institutional ownership, information environment, business complexity and corporate governance. We find that the association between stock prices and net pension assets significantly increases for firms with larger and more independent boards and firms with higher institutional ownership. We also find that value relevance of net pension assets improves for firms with larger bid-ask spread, more business segments and higher E-index, implying that SFAS No. 158 is especially beneficial for firms with larger information asymmetry, more complex business structure and managerial power. Additional tests show that our results hold when adopting a difference-in-difference approach to control sample selection bias and omitted correlated variables.

Our study contributes to two streams of research. First, this study contributes to research on the value relevance of pension accounting information. There are studies providing evidence that pension expenses were mispriced before SFAS No. 158 (Barth, Beaver and Landsman 1992; Franzoni and Marin 2006 and Picconi 2006). Coronado and Sharpe (2003) and Coronado et al. (2008) provide evidence that investors do not incorporate pension plan information disclosed in the footnotes into stock prices. Second, this study contributes to the research on SFAS No. 158 (Mitra and Hossain 2009, Yu 2013 and Shin and Yu 2016). These studies show that SFAS No. 158 increases value relevance and decreases mispricing and improves investors' ability to process pension accounting information. Our findings are consistent with these studies that SFAS No. 158 increases the visibility and value relevance of net pension asset values by requiring pension assets and liabilities to be recognized on the balance sheets. Further, we find that the impact of SFAS No. 158 on value relevance of net pension assets varies across different firm characteristics and conditions. Specifically, we find that value relevance of net pension assets improves for firms with better monitoring, like firms with higher institutional ownership and more effective board of directors, because such firms are expected to better understand and comply with new accounting rules. We also show that SFAS No. 158 significantly increases the association between stock prices and net pension assets for firms with larger information asymmetry, more complex business and entrenched management since those firms that have worse information environment and complex accounting transactions might benefit most from a new accounting rule that promotes transparency. These results highlight the importance of considering different business conditions in the value relevance research and when evaluating an accounting standard.

The remainder of the paper is organized as follows. Section 2 introduces the background on pension plan accounting and reviews prior studies. Section 3 describes the samples, variables and models. Section 4 discusses the main empirical results. Section 5 discusses additional analyses. Section 6 concludes.

II. Background and Literature Review

Accounting for defined benefit plans in the U.S. is complicated. Before 2006, the accounting practices for pension plans were regulated by the SFAS No. 87 Employers' Accounting for Pensions. SFAS No. 87 requires that certain changes in pension assets and liabilities are not immediately but gradually recognized in earnings. The purpose of smoothing pension-related earnings components is to reduce the uncertainties and risks brought by the volatile markets. In addition, the pension plan's funded status is not fully recognized on the balance sheets. Prior studies show that it is difficult for investors to understand pension plan information and process the information correctly when making investment decisions. Barth et al. (1992) examine the value relevance of different components of pension-related earnings as well as earnings unrelated to pensions. They find that markets assign higher valuation coefficients to pension-related earnings because of the smoothing of pension-related earnings. Franzoni and Marin (2006) show that the market value of a firm does not reflect pension assets and liabilities information correctly. For example, firms with underfunded pension plans are significantly over-valued, and investors could not anticipate the negative effect of pension liabilities on future earnings. Picconi (2006) finds that neither investors nor analysts are able to fully incorporate pension plan information into stock prices and forecasts. The author also finds that the off-balance-sheet information on pension funded status, instead of the on-balance-sheet information on pension funded status, is predictive on future earnings. Hann, Heflin and Subramanayam (2007) show that under SFAS No. 87, fair value accounting could not improve the value relevance of the financial statement information of pension plans. Despite the opaque accounting rules for pension plans, some studies show that the capital market is efficient to reflect the risk of pension plans. Jin, Merton and Bodie (2006) find that firms' pension plan risks are reflected in the systematic equity risk of the market.

To address the concern that accounting standards for pension plans could not communicate plans' funded status to information users in a complete and understandable way, FASB issued SFAS No. 158, Employers' Accounting for Defined Benefit Pension and Other Postretirement Plans in 2006. This standard does not change the way to measure pension expenses that are recognized in earnings. It requires companies to recognize overfunded or underfunded statuses of a defined benefit plan by reporting net pension assets or net pension liabilities on the balance sheets and to recognize any changes in the funded status through comprehensive income. Some studies investigate the impact of SFAS No. 158. Mitra and Hossain (2009) document a negative relationship between stock returns and pension transition adjustments during the initial year of SFAS No. 158. Their study implies that investors process accounting information more effectively when the information is recognized on the financial statements. Yu (2013) finds that the value relevance of the off-balance-sheet pension liabilities increases after SFAS No. 158 for firms with less institutional holding and analyst following. Shin and Yu (2016) find that mispricing of net periodic pension cost reduces after SFAS No. 158, suggesting a positive impact of SFAS No. 158.

This study is an extension on two studies: Coronado and Sharpe (2003) and Coronado et al. (2008). The authors study the association between stock prices and pension earnings as well as net pension asset values disclosed in the footnotes for a sample of S&P 500 companies during 1993-2005. They find that the stocks of these companies are generally overvalued and investors rely on pension plans information embedded in the earnings instead of pension assets and liabilities information disclosed in the footnotes to establish stock prices. Especially during the early 2000s when pension valuations had dramatic swings, firm stocks are significantly overvalued. We extend their studies to a recent period after SFAS No. 158. Since SFAS No. 158 requires companies to recognize the pension plans' funded status by reporting net pension assets and liabilities on the balance sheets, we expect a significant association between stock prices and net pension asset values rather than pension-related earnings. Furthermore, we explore and find that the impact of SFAS No. 158 on the value relevance of net pension assets varies across different firm characteristics and conditions.

III. Data, Variables and Research Methodology

In this section, we discuss the sample, variables and methodology in this research.

A. Sample

Table 1 reports the sample construction procedure. To replicate the results in Coronado et al. (2008), we construct a sample of S&P 500 companies from 1993 to 2015 following the same procedure. Our initial sample consists of all firms that were included in the S&P 500 index at any time point from 1993 to 2015. This generates a balanced panel of 24,127 firm-years for 1,049 unique firms. The major data sources are COMPUSTAT, l/B/E/S and CRSP Pension information and other financial information are extracted from COMPUSTAT. Stock prices, long-term growth forecasts, analyst earnings forecasts and analyst following come from l/B/E/S. Bid and ask stock prices are taken from CRSP. Following Coronado et al. (2008), firms with fiscal years ending in October through March in COMPUSTATare matched with price and forecast data in May in the subsequent year in l/B/E/S, and firms with fiscal years ending April through September are matched with price and forecast data in November of the same year. We require a sample firm to have the necessary pension plan information, financial information, analyst earnings forecast information and stock prices information. We exclude financial firms (SIC=6000-6999) and regulated firms (SIC=4000-4999) following prior literature (Beaudoin, Chandar and Werner 2011 and Shin and Yu 2016). (2) Our final sample for the main regressions has 8,080 firm-years for 639 unique firms. (3) Additional analyses require information from Thomson Reuters and Risk Metrics, which further reduces the sample size for these regressions.

B. Variables construction

Following Coronado and Sharpe (2003) and Coronado et al. (2008), we employ a regression model discussed in Feltham and Ohlson (1995) to study the association between stock prices and pension earnings as well as net pension assets. Key variables used in the regression models are listed as below:

Core EPS, Pension EPS

Core EPS has two parts. One is the total expected earnings (EPS) from l/B/E/S. The other one is Pension EPS, which is defined as Net Periodic Pension Cost (NPPC) minus services cost. Assuming a 35 percent tax rate, Pension EPS is expressed in an after-tax per share basis (Equation 1). By excluding service cost, Pension EPS captures the financing accruals of pension-related earnings. Core EPS is defined as the earnings component without pension information (Equation 2).

Pension EPS = 0.65 x (NPPC -service cost). (1)

Core EPS = EPS-Pension EPS. (2)

Pension NAV

Pension NAV equals the market value of pension assets minus projected benefit obligations (PBO). Both the market value of pension assets and projected benefit obligations are taken from COMPUSTAT.

Pension NAV = market value of pension assets--PBO. (3)

Core BV

Core book value is the total book equity value minus Pension NAV.

Core BV = total book equity value--Pension NAV. (4)

Core EPS, Pension EPS, Pension NAV and Core BV are deflated by the number of shares outstanding, so time trends and the associated heteroskedasticity can be removed.

Post

We use an indicator variable Post to identify the post-SFAS 158 period. We code Post as 1 if a firm-year has the fiscal year end after Dec 15, 2006, and 0 otherwise.

EPS Growth Forecast

Growth Forecast is the median of the long-term growth rate in l/B/E/S. By adding this variable, we are able to control for the influence of investors' expectations on a firm's long-term growth.

Pension Indicator

This variable indicates whether a company has a pension plan. If a company has information on the market value of pension assets or projected benefit obligations, we code pension indicator as 1, and 0 otherwise.

C. Empirical Model

Following Feltham and Ohlson (1995) and Coronado et al. (2008), we use the following model to test value relevance of pension earnings and net pension assets as well as the impact of SFAS 158 on their value relevance:

[Price.sub.it] = [[beta].sub.0] + [[beta].sub.1]Core [EPS.sub.it] + [[beta].sub.2]Pension [EPS.sub.it] + [[beta].sub.3]Pension [NAV.sub.it] + [[beta].sub.4]Core [BV.sub.it] + [[beta].sub.5][Post.sub.it] + [[beta].sub.6]Pension [EPS.sub.it] x [Post.sub.it] + [[beta].sub.7]Pension [NAV.sub.it] x[Post.sub.it] + [[beta].sub.8]EPS Growth [Forecast.sub.it] + [[beta].sub.9][Size.sub.it] + [[beta].sub.10][ROA.sub.it] + [[beta].sub.11][Leverage.sub.it] + [[beta].sub.12][Volatility.sub.it] + [[beta].sub.13]Sales [Growth.sub.it] + [[beta].sub.14][Spread.sub.it] + [[beta].sub.15][Segment.sub.it] + [[beta].sub.16]Pension [Indicator.sub.it] + Industry dummies + Year dummies + [[epsilon].sub.it]. (5)

Price is the actual stock price taken from l/B/E/S. Pension variables and forecast variables are as previously defined. We also control for other firm characteristics, such as size, performance, risk, volatility, growth, information asymmetry and business complexity. Size is the natural log of total assets. ROA is return on assets. Leverage is the ratio between total liabilities and total assets. Volatility is the standard deviation of five-year return on assets. Sales growth is the growth rate of sales revenue over the fiscal year. Spread is the bid-ask spread expressed as the percentage of price. Segment is the number of business segments. Finally, we control for year and industry fixed effects (2-digit SIC classifications) and adjust standard errors for firm clustering (Petersen 2009). (4)

Equation 5 is an extension of the model in Coronado et al. (2008), which is an adaption of the residual income valuation model in Feltham and Ohlson (1995). Based on the residual income valuation model, market value is a function of a firm's book value and its expected abnormal earnings. Coronado et al. (2008) further decompose both book values and earnings into core operations and financing activities related to the pension plans. We estimate the Equation 5. The coefficients on Pension EPS ([[beta].sup.2]) and Pension NAV ([[beta].sup.3]) indicate value relevance of pension earnings and net pension assets in the pre-SFAS 158 period. A transparent model of pension accounting implies that $1 Pension NAV contributes to $(1-T) firm value, where T is the effective marginal tax rate faced by the firm. Pension EPS is redundant to the balance sheet pension information and does not contribute to a firm's market value (Coronado et al. 2008). If the estimation results support a transparent model, we expect to find [[beta].sup.2] = 0 and 0.65< [[beta].sup.3] <1 assuming a 35 percent tax rate. On the contrary, an opaque model implies that investors determine firm equity values based on the information embedded in earnings instead of changes in net pension assets; therefore, if the estimation results support an opaque model, we expect [[beta].sup.2] to be positive and [[beta].sup.3] to be zero.

We test the transparent model versus the opaque model in the whole sample period of 1993-2015 to utilize information of all years. We use the interaction between Post and Pension EPS (Pension NAV) to distinguish the value relevance of pension earnings (net pension assets) before and after the enactment of SFAS No. 158. The coefficients on the interactions [[beta].sup.6] and [[beta].sup.7] indicate the differences in value relevance of pension earnings and net pension assets between the periods before and after SFAS No. 158. Post takes one for the post-SFAS 158 period so [[beta].sup.2] + [[beta].sup.6] and [[beta].sup.3] + [[beta].sup.7] test the value relevance of Pension EPS and Pension NAV in the post-SFAS 158 period. Due to the change in the disclosure requirement of SFAS No. 158, investors pay more attention to the net pension assets and liabilities recognized on balance sheets; therefore, we expect [[beta].sup.7] to be positive. That is, the association between net pension assets and stock prices increases in the post-SFAS 158 period. Using the whole sample period also allows us to test the transparent model versus the opaque model in the post-SFAS 158 period. If the estimation results support a transparent model in the post-SFAS 158 period, we expect to find [[beta].sup.2] + [[beta].sup.6] = 0 and 0.65< [[beta].sup.3] + [[beta].sup.7] <1. If the estimation results support an opaque model, we expect [[beta].sup.2] + [[beta].sup.6] to be positive and [[beta].sup.3] + [[beta].sup.7] to be zero.

Table 2 presents the descriptive statistics. Panel A reports the descriptive statistics for the sample in this study. For comparison purpose, we report in Panel B the descriptive statistics of the same variables for all the firms in COMPUSTAT. The average stock price is $36.84 per share, which is higher than the $24.55 per share average price of the COMPUSTAT sample. The mean values of Pension EPS and Pension NAV are $-0.01 and $-0.61 per share, which are lower than the average Pension EPS ($-0.002) and Pension NAV ($-0.27) of the COMPUSTAT sample. The average EPS and book value excluding pension information are $2.09 and $11.86 per share, compared to a lower core EPS ($1.32) and core book value ($10.26) of the COMPUSTAT sample. The average long-term forecast of earnings growth rate is 13.78 percent while the average long-term forecast of earnings growth rate is 16.46 percent for the COMPUSTAT sample. Approximately 65.5 percent of the sample firms have defined-benefit pension plans, which is consistent with Coronado et al. (2008). For the COMPUSTAT sample, only 41 percent of the firms have defined-benefit pension plans. Since the sample firms are from S&P 500 index, they are large, profitable, less risky and complex compared to the COMPUSTAT sample. The mean value of Size is 8.492, and the average ROA is 6.5 percent. The average leverage is 0.553, and ROA volatility is 4.8 percent. The sales growth rate is 10.6 percent on average. Each sample firm has 2.24 business segments on average. The information environment is good, which is reflected by the small bid-ask spread (mean = 0.53 percent) and large analyst following (mean = 3.18).

IV. Empirical Results

A. Correlations between Pension EPS and Pension NAV

Figure 1 illustrates the trend of financing accruals (Pension EPS) and net pension asset values (Pension NAV) over time. Pension EPS is also called financing accruals because they are accruals that are related to financing and managing outstanding pension obligations and assets (Coronado and Sharpe 2003). Both pension earnings and net pension asset values were positive and increased from 1993 to 2000. Around 2001, when the stock market plunged, net pension assets dropped significantly and turned to negative until 2006, meaning the market value of plan assets shrunk because of the economic downturn and companies reported net pension liabilities since 2001. However, companies still reported high pension earnings (financing accruals) in the years 2001 and 2002. Then pension earnings started to decline in 2003. The reflection of market performance in the pension earnings was delayed due to the smoothing practice. If the investors determine the firm values using pension information included in earnings rather than the pension assets information disclosed in the footnotes, the firms would be overvalued. The trend of financing accruals and net pension assets from 1993 to 2005 is consistent with the figure 1 in Coronado et al. (2008).

The financing accruals and net pension assets from 2006 to 2015 also demonstrate the disparity of these two measures. For example, the negative values of net pension assets in 2004 and 2005 were reflected in negative pension earnings in 2006. The increases in net pension assets in 2006 and 2007 accorded with the increases in financing accruals in 2007 and 2008. There were still delays in reflecting true pension values in the earnings because SFAS No. 158 does not change the way to measure pension expenses that are recognized in earnings. However, the requirement of recognizing net pension assets and liabilities on balance sheets would improve investors' ability to incorporate true value information into stock prices. Both financing accruals and net pension asset values stay negative from 2009 to 2015.

Figure 2 plots the correlations between Pension EPS and Pension NAV across years. The plot during 1995-2005 is similar to the Figure 2 in Coronado et al. (2008). The correlations between Pension EPS and Pension NAV were as high as around 0.9 in early years and dramatically reduced to almost zero during the tech bubble period. Then it went back to approximately 0.98 in 2006 and dropped again during the financial crisis. This plot confirms the message in Figure 1 that when the economy is bad, the true pension plan conditions could not be reflected in pension earnings immediately, resulting in low correlations between Pension EPS and Pension NAV. The correlations between the two measures went back to high from 2009 to 2015.

Table 3 presents Pearson correlations among main variables used in the regressions. Price is positively correlated with Core EPS and Core BV. It is negatively correlated with Pension EPS and Pension NAV. By construction, Core EPS is negatively correlated with Pension EPS, and Pension NAV is negatively correlated with Core BV. The correlation between Core EPS and Core BV reflects the link between income statement information and balance sheet information. The correlation between Pension EPS and Pension NAV confirms Figure 2.

B. The associations between pension earnings, net pension assets and stock prices

In this section, we discuss the multivariate test results. We first replicate the results in Coronado et al. (2008) using the data from 1993 to 2005. Table 4 reports our replication results. Following Coronado et al. (2008), we separate the period 1993-2001 from the period 2002-2005 to compare the value relevance of pension EPS and net pension asset values before and after the stock market plunged, which is in the so called preconscious period in prior studies. Model 1 and Model 4 are expected to be consistent with the transparent view of pension accounting, that is, net pension assets are fully reflected in firm values so Pension EPS (financing accruals) do not contribute any information to firm values and should not appear in the valuation model. In Table 4, Model 1 shows that Core EPS has a coefficient of 9.461, meaning that one-dollar increase in Core EPS leads to 9.461 dollars of increase in market price per share. Pension NAV has a coefficient of 1.497, which shows that Pension NAV has a positive impact on a firm's stock price. The coefficient on Core BV is positive and significant (0.515, p<0.01), as expected. EPS Growth Forecast also positively affects stock prices, suggesting that firms with higher than expected growth rate will be valued higher. The coefficient on Pension Indicator is positive and significant, implying that firms with a defined benefit plan have higher market values. In Model 3, we add two-year lagged value of Pension NAV. Due to the smoothing of pension accounting, changes in pension assets and liabilities are recognized in pension earnings gradually. Therefore, Pension EPS may reflect pension plan conditions for several years. In this case, two-year lagged values of Pension NAV may have an impact on a firm's stock price. Consistent with this prediction, we find that two-year lagged values of Pension NAV positively affect a firm's stock price. Model 4 shows that the coefficient of Pension NAV is 0.851, which is between 0.65 and 1. The estimation results of Core EPS, Core BV, EPS Growth Forecast and Pension Indicator during the period 2002-2005 are similar to the results for the period 1993-2001.

In Models 2 and 5, we include Pension EPS to test the opaque view. After introducing Pension EPS, Pension NAV becomes insignificant and the coefficients on Pension EPS are significantly high (17.836 and 19.969, p<0.01). Turning to the control variables, we find that firm size positively affects stock prices. Bid-ask spreads negatively affect stock prices, consistent with the premise that firms with larger information asymmetry have lower market valuation. During the period of 2002-2005, we find profitability positively affects firm valuation while sales growth rates negatively affect stock prices.

Generally speaking, our estimation results during the period 1993-2005 are consistent with Coronado and Sharpe (2003) and Coronado et al. (2008). We draw the same conclusion that before the issuance of SFAS No. 158, investors rely more on pension plan information embedded in the earnings instead of pension plan assets and liabilities information disclosed in the footnotes to make investment decisions. Therefore, pension earnings are more associated with firm values than net pension assets values.

Then we study the associations between stock prices and pension earnings as well as net pension assets in the whole sample period of 1993-2015. We include the period after the issuance of SFAS No. 158 to investigate its impact on the value relevance of pension earnings and net pension assets. Table 5 presents the estimation results. Models 1 and 3 test the transparent model while Model 2 tests the opaque model by including Pension EPS. Model 4 uses a sample excluding the years 2007 and 2008. In all the four models, we include a dummy variable Post to differentiate the period after SFAS No. 158 from the period before SFAS 158. We use the interactions between Post and Pension EPS as well as Pension NAV to test the differences in the value relevance of pension earnings and net pension assets between the two periods. Both Models 1 and 3 show that the coefficients of net pension assets are positive and significant before SFAS 158. Especially, the coefficient on Pension NAV in Model 3 falls in the theoretical range (0.806). F-tests show that in the post-SFAS 158 period, net pension assets also positively affect market prices.

Models 2 and 4 include Pension EPS to test an opaque model. The R-squares in both models are around 0.77. The variance inflation factors (VIF) in Model 2 range from 1.10 to 5.07, implying that multicollinearity is unlikely to be an issue. In the pre-SFAS 158 period, Pension EPS positively affect stock prices while Pension NAV become insignificant, which confirms the results in Table 4. The interactions between Pension EPS and Post are significantly negative (-10.668 and -11.239, p=0.039 and 0.053) while the interactions between Pension NAV and Post are significantly positive (0.677 and 0.753, p<0.05), suggesting that the value relevance of pension earnings decreases while the value relevance of net pension assets increases in the post SFAS-158 period. The F-tests show that both pension earnings and net pension assets positively affect stock prices in the post-SFAS 158 period, especially if the coefficients on net pension assets are within the theoretical range (0.831 and 0.881). In summary, Table 5 supports the argument that SFAS No. 158 increases the transparency and visibility of pension plan accounting information by requiring pension assets and liabilities to be recognized on the balance sheets. Therefore, investors rely more on net pension asset values to make investment decisions. Although pension earnings still affect stock prices, their value relevance significantly decreases. Instead, the association between stock prices and net pension assets significantly increases. Stock prices better reflect the more accurate net pension assets information compared to the information embedded in earnings.

C. Value relevance of pension earnings and net pension assets across various firm characteristics and conditions

We next explore the impact of SFAS No. 158 on value relevance of pension earnings and net pension assets for various firm characteristics and conditions. Specifically, we test the value relevance for firms with different levels of institutional ownership, information environment, business complexity and corporate governance. Table 6 presents our findings of this exploration.

Table 6 Panel A shows the different impacts of SFAS No. 158 between firms with higher institutional ownership and lower institutional ownership. (5) Institutional is institutional ownership, defined as the percentage of shares held by institutional owners. Table 2 shows that approximately 67.8 percent of the shares are held by institutional owners. lnst_H is a dummy variable equal to 1 if a firm's institutional ownership is higher than the sample median, and 0 otherwise. The two-way interaction between pension earnings (net pension assets) and Post represents the difference in value relevance of pension earnings (net pension assets) between the periods before and after SFAS No. 158 for firms with lower institutional ownership. The three-way interaction represents the differences in value relevance of pension earnings (net pension assets) between the two periods and between firms with higher institutional ownership and lower institutional ownership. F-tests show that the joint effect of Post on the value relevance of net pension assets for firms with higher institutional ownership is positive and significant ([[beta].sub.7]+[[beta].sub.12] =0.974, p<0.01). This is consistent with the finding that SFAS No. 158 increases the value relevance of net pension assets for firms that are owned by institutional owners because institutional owners are more sophisticated and have stronger abilities to process information. In addition, institutional owners are considered to have more resources and incentives to monitor firms (Chung, Firth and Kim 2002). Turning to the pension earnings, value relevance of pension earnings does not significantly change after SFAS No. 158 ([[beta].sub.6]+[[beta].sub.11] =-6.262, p=0.241). We also report the value relevance of pension earnings and net pension assets for firms with higher and lower institutional ownership for both periods. The results are consistent with an opaque model in the pre-SFAS 158 period for firms with lower institutional ownership ([[beta].sub.2]=28.110, p<0.01; [[beta].sub.3] =0.130, p=0.682) and for firms with higher institutional ownership ([[beta].sub.2]+[[beta].sub.9] =17.745, p<0.01; [[beta].sub.3]+[[beta].sub.10] =0.136, p=0.620). And it moves towards a more transparent model in the post-SFAS 158 period for firms with lower institutional ownership ([[beta].sub.2]+[[beta].sub.6] =16.868, p<0.01; [[beta].sub.3]+[[beta].sub.7] =0.719, p<0.1) and firms with higher institutional ownership ([[beta].sub.2]+[[beta].sub.6]+[[beta].sub.11]=21.848, p<0.01; [[beta].sub.3]+[[beta].sub.7]+[[beta].sub.12]=1.104, p<0.01).

Table 6 Panel B reports our investigation on information environment. We use two proxies for information environment. One is analyst following, Analyst, and the other one is information asymmetry, measured by bid-ask spread, Spread. The joint effect of Post on the value relevance of net pension assets for firms with higher bid-ask spread is positive and significant ([[beta].sub.7]+[[beta].sub.12]=0.734, p<0.1) but insignificant for firms with more analyst following ([[beta].sub.7]+[[beta].sub.12]=0.487, p=0.309), implying that SFAS No. 158 significantly increases value relevance of net pension assets for firms with larger information asymmetry. Panel C compares firms with complex business structures to firms with simple business structures. We use the number of business segments to measure business complexity. Similarly, we find that the association between stock prices and pension earnings significantly decreases after SFAS No. 158 for firms with more business segments while the association between stock prices and net pension assets significantly increases. Panels D and E report the effect of corporate governance. (6) We consider two aspects of corporate governance. One aspect is managerial entrenchment, proxied by entrenchment index. The entrenchment index is constructed following Bebchuk, Cohen and Ferrell (2008). They create an index of corporate governance based on six provisions: staggered boards, limits to shareholder bylaw amendments, poison pills, golden parachutes, supermajority requirements for mergers and supermajority requirements for charter amendments. The index adds one for each provision so that it ranges from zero to six. A larger value of E-index indicates larger management power. The average E-index is 2.25, indicating a balance between managerial power and shareholders' power. The other aspect is board effectiveness, measured by board size and board independence. Literature has shown that board size and independence affect the effectiveness in corporate monitoring. Although independent boards are generally considered to be more effective (Klein 2002), there is mixed evidence on the relation between board size and monitoring effectiveness. Yermack (1996) and Eisenberg, Sundgrenb and Wellsc (1998) find a negative association between board size and firm value while Belkhir (2009) finds a positive association between board size and performance in the banking industry. Table 2 shows that on average each board has 10.09 members and 72.9 percent of the board directors are independent directors. Panel D shows that value relevance of net pension assets significantly increases after SFAS No.158 for firms with entrenched management ([[beta].sub.7]+[[beta].sub.12]=0.761, p<0.05). Panel E shows that value relevance of net pension assets significantly increases after SFAS No.158 for firms with larger boards ([[beta].sub.7]+[[beta].sub.12]=1.116, p<0.01) or more independent boards ([[beta].sub.7]+[[beta].sub.12]=0.953, p<0.05). Taken together, Table 6 indicates that SFAS No. 158 significantly increases the association between stock prices and net pension assets for firms with better monitoring. It also shows that SFAS No. 158 is beneficial, especially for firms with larger information asymmetry, more complexity and more entrenched management.

V. Additional Analyses

A. Matched Sample Regressions

The setting in our study mitigates the endogeneity issues because, as stated by Beaudoin, Chandar and Werner (2011), publicly traded firms "did not have discretion on the timing of SFAS 158 adoption. We therefore need not contend with the issue of examining early versus late standard adopters" (102). In additional analyses, we try to address another type of endogeneity. Following Coronado et al. (2008), our full sample uses firms without a defined benefit pension plan as a control sample. In this section, we match firms with a DB pension plan to firms without a DB pension plan and examine the effect of SFAS No. 158 on the value relevance for firms with a DB pension plan. We require the matched sample firms have information in both pre- and post-SFAS 158 periods. In this way, our research naturally adopts a difference-in-difference research design because all firms with a DB pension plan are controlled by similar firms without a DB pension plan and all firms in the post-SFAS 158 period are controlled by themselves in the pre-SFAS 158 period. This further reduces sample selection bias and potential omitted correlated variables. We match firms with a DB pension plan to firms without a DB pension plan by year, industry (2-digit SIC code), total assets and ROA. Then we require each matched pair have data in both periods before and after SFAS No. 158. We identify 756 pairs throughout the whole sample period. Table 7 presents the estimation results based on the matched sample. The coefficient on Pension NAVxPost is 1.062 and significant at 5 percent level (p=0.036). Table 7 suggests that the association between stock prices and net pension assets significantly increases after SFAS No. 158 after controlling for sample selection bias and omitted correlated variables.

B. Implied valuation errors

In this section, we compute the implied valuation errors for the period of 1993-2015. Literature on valuation and mispricing use different models to estimate valuation errors or mispricing in different scenarios, such as tax shields, stock carve-outs, acquisitions and goodwill impairments. (Fernandez 2004, Fernandez 2007, Lamont and Thaler 2003 and Gu and Lev 2011). Follow Coronado et al. (2008), we use a PE ratio model of valuation to estimate the magnitude of valuation errors. The underlying assumption of this model is that Pension EPS and Core EPS are priced equally.

The estimation results of the PE ratio model, together with the standard errors (in parentheses) for the period of 1993 to 2015 are as following:

[mathematical expression not reproducible] (6)

P is the fiscal-end close price from COMPUSTAT. [P/EPS]is PE ratio. EPS is the expected earnings per share from l/B/E/S. div is dividend payout rate, growth is median of long term growth forecast and In (assets) is the natural log of the book value of firm assets. Time and industry fixed effects are included in this model.

We calculate the estimated market price by multiplying Pension EPS with the estimated PE ratio because the model assumes that investors treat pension EPS and operating EPS the same way. The estimated price is based on pension earnings. The estimated firm equity value is the multiplication of the estimated market price and the number of common shares outstanding. Then the implied valuation error is achieved by subtracting the book value of net pension assets from the estimated firm value. A positive value of the implied valuation error suggests that the stock price is overvalued since the estimated market value is based on pension EPS.

Figure 3 plots the means and medians of implied firm valuation errors from 1993 to 2015, expressed as the percentage of market capitalization (price multiplied by common shares outstanding). During the years 1993 to 2005, the means and medians of valuation errors are all above zero, supporting that firm stocks are overvalued, especially during the years 2000-2002 when the market had large swings. We find that the implied valuation errors are close to zero during the years 2006-2015, except for the year 2008. Generally speaking, stock prices are less overvalued during the time period of 2006-2015 and the valuation errors are smaller compared to the period 1993-2005.

VI. Conclusions

In this study, we revisit the research question on the value relevance of pension plan accounting information. We first replicate Coronado et al. (2008) and draw the same conclusion that pension plan accounting in the period 1993-2005 is consistent with an opaque model. In other words, investors rely more on pension plan information embedded in the earnings instead of information of pension assets and liabilities disclosed in the footnotes to establish stock prices.

FASB issued SFAS No. 158 in 2006, which requires companies to recognize funded status of pension plans. SFAS No. 158 migrates information on net pension asset values from footnotes to financial statements, resulting in increased transparency and visibility of pension plan information. We therefore investigate the associations between stock prices and net pension asset values, as well as pension earnings in the period of 2006-2015. We find that the association between stock prices and net pension assets significantly increases after SFAS No. 158 while the association between stock prices and pension earnings significantly decreases. Further, the positive effect of SFAS No. 158 on value relevance is significant for firms with higher institutional ownership, larger bid-ask spread, more business segments, higher E-index and larger and more independent boards. The valuation errors after 2005 indicate that firms are less overvalued. These results together imply a move towards a transparent view of pension plan accounting after the issuance of SFAS No. 158.

In summary, by comparing the value relevance of pension earnings and net pension asset values before and after the issuance of SFAS No. 158, we conclude that investors pay more attention to the information on pension assets and liabilities recognized on balance sheets, and such information is better reflected in stock prices.

References

Barth, M.E., W.H. Beaver, and W.R. Landsman. "The Market Valuation Implications of Net Periodic Pension Cost Components." Journal of Accounting and Economics, vol. 15, no. 1, 1992, pp. 27-62.

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, vol. 27, no. 1, 2011, pp. 99-110.

Bebchuk, L., A. Cohen, and A. Ferrell. "What Matters in Corporate Governance?" The Review of Financial Studies, vol. 22, no. 2, 2008, pp. 783-827.

Belkhir, M. "Board of Directors' Size and Performance in the Banking Industry." International Journal of Managerial Finance, vol. 5, no. 2, 2009, pp. 201-221.

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, vol. 7, no. 3, 2008, pp. 257-276.

Coronado, L.J., and S.A. Sharpe. "Did Pension Plan Accounting Contribute to a Stock Market Bubble?" Brookings Papers on Economic Activity, vol. 34, no. 1, 2003, pp. 323-359.

Chung, R., M. Firth, and J-B. Kim. "Institutional Monitoring and Opportunistic Earnings Management." Journal of Corporate Finance, vol. 8, no. 1, 2002, pp. 29-48.

Eisenberg, T., S. Sundgrenb, and M.T. Wellsc. "Larger Board Size and Decreasing Firm Value in Small Firms." Journal of Financial Economics, vol. 48, no. 1, 1998, pp. 35-54.

Financial Accounting Standards Board (FASB). "Statement of Financial Accounting Standards No. 87: Employers' Accounting for Pensions." Statement of Financial Accounting Standards No. 87. Stamford, Connecticut, 1985.

Financial Accounting Standards Board (FASB). "Statement of Financial Accounting Standards No. 158: Employers' Accounting for Defined Benefit Pensions and Other Postretirement Plans--An Amendment of FASB Statements No. 87, 88, 106, and 132." Statement of Financial Accounting Standards No. 158. Stamford, Connecticut, 2006.

Feltham, G.A., and J.A. Ohlson. "Valuation and Clean Surplus Accounting for Operating and Financial Activities." Contemporary Accounting Research, vol. 11, no. 2, 1995, pp. 689-731.

Fernandez, P. "The Value of Tax Shields Is NOT Equal to the Present Value of Tax Shields." Journal of Financial Economics, vol. 73, no. 1, 2004, pp. 145-165.

Fernandez, P. "Company Valuation Methods. The Most Common Errors in Valuations." 2007. https://s3.amazonaws.com/academia.edu.documents/36234952/COMMON_ERRORS_IN_VALUATION.pdf?AWSAccessKeyld=AKIAIWOWYYGZ2Y53UL3A&Expires=1526359475&Signature=ltS9NJrZMChSfkPeDtE6T7Ji3ok%3D&response-content-disposition=inline%3B%20filename%3DCOMMON_ERRORS_IN_VALUATION.pdf.

Franzoni, F., and J. Marin. "Pension Plan Funding and Stock Market Efficiency." Journal of Finance, vol. 61, no. 2, 2006, pp. 921-956.

Gold, J. "Accounting/Actuarial Bias Enables Equity Investment by Defined Benefit Pension Plans." North American Actuarial Journal, vol. 9, no. 3, 2005, pp. 1-21.

Gu, F., and B. Lev. "Overpriced Shares, lll-Advised Acquisitions, and Goodwill Impairment." The Accounting Review, vol. 86, no. 6, 2011, pp. 1995-2022.

Hann, R.N., F. Heflin, and K.R. Subramanayam. "Fair-Value-Pension Accounting." Journal of Accounting and Economics, vol. 44, no. 3, 2007, pp. 328-358.

Jin, L., R.C. Merton, and Z. Bodie. "Do a Firm's Equity Returns Reflect the Risk of Its Pension Plan?" Journal of Financial Economics, vol. 81, no. 1, 2006, pp. 1-26.

Klein., A. "Audit Committee, Board of Director Characteristics, and Earnings Management." Journal of Accounting and Economics, vol. 33, no. 3, 2002, pp. 375-400.

Lamont, O.A., and R.H. Thaler. "Can the Market Add and Subtract? Mispricing in Tech Stock Carve-outs." Journal of Political Economy, vol. 111, no. 2, 2003, pp. 227-268.

Mitra, S., and M. Hossain. "Value-Relevance of Pension Transition Adjustments and Other Comprehensive Income Components in the Adoption Year of SFAS No. 158." Review of Quantitative Finance and Accounting, vol. 33, no. 3, 2009, pp. 279-301.

Petersen, M. "Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches." Review of Financial Studies, vol. 22, no. 1, 2009, pp. 435-480.

Picconi, M. "The Perils of Pensions: Does Pension Accounting Lead Investors and Analysts Astray." Accounting Review, vol. 81, no. 4, 2006, pp. 925-955.

Shin, Y., and K. Yu. "Do Investors Misprice Components of Net Periodic Pension Cost?" Accounting and Finance, vol. 56, no. 3, 2016, pp. 845-878.

United States Department of Labor (DOL) "Private Pension Plan Bulletin Historical Tables and Graphs 1975-2014." Retirement Bulletins, Washington DC, 2016. https://www.dol.gov/ebsa/pdf/historicaltables.pdf

Yermack, D. "Higher Market Valuation of Companies with a Small Board of Directors." Journal of Financial Economics, vol. 40, no. 2, 1996, pp. 185-211.

Yu, K. "Does Recognition versus Disclosure Affect Value Relevance? Evidence from Pension Accounting." The Accounting Review, vol. 88, no. 3, 2013, pp. 1095-1127.

Xiaolu Xu

University of Massachusetts-Boston

Wen Xiao

University of Massachusetts-Boston

Atreya Chakraborty

University of Massachusetts-Boston

(1) Yu (2013) points out that "[n]et pension assets or liabilities recognized on the balance sheet under SFAS No. 87 are the funded status of a pension plan adjusted for off-balance-sheet pension liabilities" because SFAS No. 87 requires recognizing a minimum liability when the fair value of plan assets is less than the accumulated benefit obligation (1098).

(2) Following Coronado and Sharpe (2003) and Coronado et al. (2008), a firm without a pension plan is kept in the full sample. We also keep financial firms and regulated firms in the sample as a robustness check. The results are qualitatively unchanged.

(3) Our sample size deviates from Coronado et al. (2008) for the following reasons. First, different from Coronado et al. (2008), we report the number of firm-years used in each regression model, which is the sample after deleting observations with missing information. If we only require that a firm-year included in both COMPUSTAT and l/B/ES, but do not delete any observations with missing information for certain data items, we get 5,715 firm-years in 1993-2001 and 2,270 firm-years in 2002-2005, which is close to Coronado et al. (2008). Second, our models include more control variables compared with Coronado et al. (2008), which reduces our sample sizes further. Third, we delete firm-years in the financial industries (SIC=6,000-6,999) and regulated industries (4,000-4,999) following prior literature (Beaudoin et al. 2011 and Shin and Yu 2016) while there is no mentioning in Coronado et al. (2008) that these industries are excluded.

(4) Coronado et al. (2008) include half-year dummies in their regression models to indicate price and forecast data in May and November, respectively. When we include half-year dummies, the results are qualitatively unchanged.

(5) The institutional ownership data are taken from Thomson Reuters 13F filings. Due to our limited access to the institutional ownership data, the sample period for the test of institutional ownership ranges from 1993 to 2013.

(6) The corporate governance and directors' data are taken from Risk Metrics. Due to our limited access to Risk Metrics, the sample period for the test of managerial entrenchment is 1993-2013 and the sample period for the test of board effectiveness is 1996-2013.
TABLE 1
Sample Selection
Table 1 presents the sample selection procedure. Following Coronado et
al. (2008), our initial sample includes all years for any firms that
had been included in the S&P 500 index at any time point over the
period 1993-2015.

                           Firm-year      Firms
                           Observations

All S&P500 firms           24,127         1,049
from 1993 to 2015
Exclude observations       (9,783)         (157)
with missing necessary
financial information
in COMPUSTAT
Exclude observations       (4,635)         (111)
with missing necessary
price
and forecast
information in
l/B/E/S
Exclude observations         (313)           (9)
with missing bid and ask
stock
price
information
in CRSP
Exclude financial firms    (1,316)         (133)
(SIC = 6000-6999) and
regulated firms
(SIC = 4000 -4999)
Final sample                8,080           639

TABLE 2
Descriptive Statistics
Table 2 presents the descriptive statistics of the variables in
regressions. Panel A reports the descriptive statistics for the sample
in this study. Panel B reports the descriptive statistics for all the
COMPUSTAT firms. Stock price (P) is the actual stock prices taken from
l/B/E/S. Pension EPS is defined as the difference between Net periodic
pension cost (NPPC) and services cost, then multiplied by (1-35
percent). Core EPS is the total expected earnings (EPS) from l/B/E/S
minus Pension EPS. Pension NAV equals the market value of pension
assets minus projected benefit obligations (PBO). Core book value is
the total book equity value minus Pension NAV. Core EPS, Pension EPS,
Pension NAV and Core BV are deflated by number of shares outstanding.
Post equals 1 if a firm-year comes from the post-SFAS 158 period, and 0
otherwise. EPS Growth Forecast is the median of long-term growth rate
in l/B/E/S. Pension Indicator is a dummy variable equal to 1 if a
sample firm has a defined benefit pension plan, and 0 otherwise. Size
is the natural log of total assets. ROA is return on assets. Leverage
is the ratio between total liabilities and total assets. Volatility is
the standard deviation of five-year ROAs. Sales growth is the growth
rate of sales revenue over the fiscal year. Spread is the bid-ask
spread expressed as the percentage of price. Segment is the number of
business segments. Analyst is the number of analysts following a sample
firm in the fiscal year. Institutional is the percentage of shares held
by institutional owners. E-index is the entrenchment index constructed
following Bebchuk et al. (2008). Board Size is the number of directors
in a board. Independence is the percentage of directors in a board who
are outsiders. Alt continuous variables are winsorized by top and
bottom 1 percent.

                Panel A: Final Sample

Variable        N        Mean     Median   Std Dev   Lower      Upper
                                                     Quartile   Quartile

Price            8,080   36.839   29.475   29.582    16.560     48.250
Core EPS         8,080    2.090    1.580    1.969     0.760      2.839
Pension EPS      8,080   -0.006    0.000    0.116    -0.011      0.001
Pension NAV      8,080   -0.611   -0.004    1.909    -0.559      0.000
Core BV          8,080   11.864    8.916   10.726     4.454     16.056
Post             8,080    0.441    0.000    0.496     0.000      1.000
EPS Growth       8,080   13.779   12.500    6.966    10.000     16.000
Forecast (%)
Pension          8,080    0.655    1.000    0.475     0.000      1.000
Indicator
Size             8,080    8.492    8.420    1.330     7.610      9.345
ROA              8,080    0.065    0.067    0.074     0.035      0.103
Leverage         8,080    0.553    0.552    0.203     0.420      0.677
Volatility       8,080    0.048    0.029    0.057     0.016      0.055
Sales Growth     8,080    0.106    0.075    0.217     0.002      0.170
Spread (%)       8,080    0.529    0.098    0.846     0.032      0.743
Segment          8,080    2.240    1.000    1.832     1.000      3.000
Analyst          8,080    3.178    3.258    0.530     2.833      3.555
Institutional    7,336    0.678    0.719    0.204     0.580      0.822
Eindex           6,695    2.252    2.000    1.354     1.000      3.000
Board Size       5,713   10.086   10.000    2.299     8.000     12.000
Independence     5,713    0.729    0.769    0.159     0.636      0.857

                Panel B: The COMPUSTAT Sample

Variable        N        Mean     Median   Std Dev   Lower      Upper
                                                     Quartile   Quartile

Price           40,826   24.546   18.310   21.663     9.750     32.070
Core EPS        40,826    1.323    0.997    1.446     0.420      1.842
Pension EPS     40,826   -0.002    0.000    0.057     0.000      0.000
Pension NAV     40,826   -0.271    0.000    1.025    -0.092      0.000
Core BV         40,826   10.256    7.491    9.612     3.859     13.486
Post            40,826    0.416    0.000    0.493     0.000      1.000
EPS Growth      40,826   16.458   15.000    9.408    10.000     20.000
Forecast (%)
Pension         40,826    0.410    0.000    0.492     0.000      1.000
Indicator
Size            40,826    6.851    6.696    1.858     5.484      8.051
ROA             40,826    0.034    0.048    0.112     0.014      0.086
Leverage        40,826    0.498    0.503    0.225     0.326      0.653
Volatility      40,826    0.073    0.036    0.110     0.017      0.077
Sales Growth    40,826    0.161    0.100    0.306     0.009      0.237
Spread (%)      40,826    1.023    0.268    1.599     0.066      1.333
Segment         40,826    1.933    1.000    1.581     1.000      3.000
Analyst         40,826    2.480    2.485    0.748     1.946      3.045
Institutional   37,025    0.531    0.573    0.281     0.315      0.767
Eindex          17,648    2.227    2.000    1.324     1.000      3.000
Board Size      15,441    9.174    9.000    2.331     7.000     11.000
Independence    15,441    0.704    0.750    0.168     0.600      0.833

TABLE 3
Correlations
Table 3 presents the Pearson correlations among the main variables.
Correlations that are significant at 5 percent are in bold.

                    Price    1        2        3        4        5

Core EPS (1)         0.827
Pension EPS (2)     -0.094   -0.196
Pension NAV (3)     -0.202   -0.309    0.518
Core BV (4)          0.588    0.651   -0.180   -0.373
Post (5)             0.405    0.424   -0.211   -0.228    0.378
EPS Growth          -0.077   -0.219    0.007    0.132   -0.271   -0.203
Forecast (6)
Pension              0.119    0.188   -0.037   -0.232    0.208    0.024
Indicator (7)
Size (8)             0.383    0.426   -0.046   -0.203    0.430    0.359
ROA (9)              0.210    0.252    0.019    0.068   -0.077    0.035
Leverage (10)        0.074    0.141   -0.052   -0.271   -0.070    0.086
Volatility (11)      0.158   -0.171    0.013    0.021   -0.135   -0.030
Sales Growth (12)   -0.005   -0.023    0.019    0.103   -0.111   -0.158
Spread (13)         -0.313   -0.301    0.152    0.181   -0.231   -0.477
Segment (14)        -0.040    0.001    0.096   -0.047    0.036   -0.076

                    Price    7        8        9        10       11

Core EPS (1)         0.827
Pension EPS (2)     -0.094
Pension NAV (3)     -0.202
Core BV (4)          0.588
Post (5)             0.405
EPS Growth          -0.077
Forecast (6)
Pension              0.119
Indicator (7)
Size (8)             0.383    0.276
ROA (9)              0.210   -0.052   -0.013
Leverage (10)        0.074    0.342    0.269   -0.214
Volatility (11)     -0.158   -0.167   -0.209   -0.154   -0.061
Sales Growth (12)   -0.005   -0.228   -0.166    0.186   -0.196    0.109
Spread (13)         -0.313    0.013   -0.337   -0.158    0.048    0.006
Segment (14)        -0.040    0.209    0.175   -0.032    0.096   -0.067

                    6

Core EPS (1)
Pension EPS (2)
Pension NAV (3)
Core BV (4)
Post (5)
EPS Growth
Forecast (6)
Pension             -0.362
Indicator (7)
Size (8)            -0.369
ROA (9)              0.031
Leverage (10)       -0.291
Volatility (11)      0.219
Sales Growth (12)    0.426
Spread (13)          0.041
Segment (14)        -0.123

                    12       13

Core EPS (1)
Pension EPS (2)
Pension NAV (3)
Core BV (4)
Post (5)
EPS Growth
Forecast (6)
Pension
Indicator (7)
Size (8)
ROA (9)
Leverage (10)
Volatility (11)
Sales Growth (12)
Spread (13)         -0.001
Segment (14)        -0.058   -0.013

TABLE 4
Regressions of Stock Prices on Pension Earnings and Net Pension Assets
Value: Pre-SFAS 158 Period.
Table 4 presents the regression results of stock prices on pension
earnings and net pension assets value before SFAS No. 158. Models 1-3
present the results for the sample period of 1993 to 2001, and Models
4-5 present the results for the sample period of 2002 to 2005. Stock
price (P) is the actual stock prices taken from l/B/E/S. Pension EPS is
defined as the difference between Net periodic pension cost (NPPC) and
services cost, then multiplied by (1-35 percent). Core EPS is the total
expected earnings (EPS) from l/B/E/S minus Pension EPS. Pension NAV
equals the market value of pension assets minus projected benefit
obligations (PBO). Core book value is the total book equity value minus
Pension NAV. Core EPS, Pension EPS, Pension NAV and Core BV are
deflated by number of shares outstanding. EPS Growth Forecast is the
median of long-term growth rate in l/B/E/S. Pension Indicator is a
dummy variable equal to 1 if a sample firm has a defined benefit
pension plan, and 0 otherwise. Size is the natural log of total assets.
ROA is return on assets. Leverage is the ratio between total
liabilities and total assets. Volatility is the standard deviation of
five-year ROAs. Sales growth is the growth rate of sales revenue over
the fiscal year. Spread is the bid-ask spread expressed as the
percentage of price. Segment is the number of business segments.
2-digit SIC industry dummies and year dummies are included. Year
dummies are created by fiscal year. Standard errors are adjusted for
firm clusters. P-values are presented in the parentheses below the
estimated coefficients. We use (*), (**), (***) to indicate
significance at 10 percent, 5 percent, 1 percent level, (two-tailed),
respectively.

             1993-2001
             (1)                         (2)
Variables    Coeff.            P-value   Coeff.            P-value

Intercept      -19.126 (***)   0.000       -18.790 (***)   0.000
Core EPS         9.461 (***)   0.000         9.582 (***)   0.000
Pension                                     17.836 (***)   0.000
EPS
Pension          1.497 (***)   0.000         0.133         0.721
NAV
Core BV          0.515 (***)   0.000         0.478 (***)   0.000
EPS              0.580 (***)   0.000         0.581 (***)   0.000
Growth
Forecast
Size             2.544 (***)   0.000         2.541 (***)   0.000
ROA              4.946         0.325         4.195         0.402
Leverage        -1.045         0.607        -2.178         0.290
Volatility       3.389         0.668         2.505         0.751
Sales           -0.636         0.606        -0.588         0.633
Growth
Spread          -0.868 (***)   0.001        -0.859 (***)   0.001
Segment         -0.082         0.680        -0.186         0.333
Pension          2.181 (***)   0.004         2.001 (***)   0.009
Indicator
Pension
NAV, lag 2
Industry     Yes                         Yes
dummies
Year         Yes                         Yes
dummies
N            2,935                       2,935
R-square         0.659                       0.663

             1993-2001
             (3)
Variables    Coeff.            P-value

Intercept      -22.362 (***)   0.000
Core EPS         9.758 (***)   0.000
Pension
EPS
Pension          0.756 (***)   0.008
NAV
Core BV          0.501 (***)   0.000
EPS              0.609 (***)   0.000
Growth
Forecast
Size             2.789 (***)   0.000
ROA              6.502         0.253
Leverage        -1.352         0.537
Volatility       4.574         0.586
Sales           -0.205         0.880
Growth
Spread          -0.837 (***)   0.003
Segment         -0.061         0.772
Pension          2.078 (**)    0.011
Indicator
Pension          0.726 (***)   0.000
NAV, lag 2
Industry     Yes
dummies
Year         Yes
dummies
N            2,537
R-square         0.652

             2002-2005
             (4)                         (5)
Variables    Coeff.            P-value   Coeff.            P-value

Intercept       -9.921 (**)    0.045        -7.196         0.116
Core EPS        10.127 (***)   0.000        10.342 (***)   0.000
Pension                                     19.969 (***)   0.000
EPS
Pension          0.851 (**)    0.012         0.437         0.174
NAV
Core BV          0.368 (***)   0.000         0.310 (***)   0.001
EPS              0.623 (***)   0.000         0.621 (***)   0.000
Growth
Forecast
Size             1.185 (***)   0.002         0.969 (***)   0.008
ROA             27.724 (***)   0.000        23.966 (***)   0.000
Leverage        -0.755         0.767        -1.955         0.441
Volatility      -3.724         0.265        -4.226         0.201
Sales           -5.674 (***)   0.000        -5.589 (***)   0.000
Growth
Spread          -2.478 (***)   0.003        -2.592 (***)   0.002
Segment         -0.083         0.677        -0.085         0.662
Pension          1.263         0.148         1.190         0.165
Indicator
Pension
NAV, lag 2
Industry     Yes                         Yes
dummies
Year         Yes                         Yes
dummies
N            1,517                       1,517
R-square         0.760                       0.769

             2002-2005
             (6)
Variables    Coeff.            P-value

Intercept       -9.661 (**)    0.050
Core EPS        10.307 (***)   0.000
Pension
EPS
Pension          0.570         0.107
NAV
Core BV          0.334 (***)   0.001
EPS              0.633 (***)   0.000
Growth
Forecast
Size             1.158 (***)   0.003
ROA             26.266 (***)   0.000
Leverage        -1.276         0.622
Volatility      -4.340         0.197
Sales           -5.516 (***)   0.001
Growth
Spread          -2.334 (***)   0.008
Segment         -0.047         0.816
Pension          1.306         0.141
Indicator
Pension          0.508 (*)     0.089
NAV, lag 2
Industry     Yes
dummies
Year         Yes
dummies
N            1,481
R-square         0.759

TABLE 5
Regressions of Stock Prices on Pension Earnings and Net Pension Assets
Value: Full Sample Period 1993-2015.
Table 5 presents the regression results of stock prices on pension
earnings and net pension assets value for the whole sample period from
1993 to 2015. Model 4 excludes the Years 2007 and 2008. Stock price (P)
is the actual stock prices taken from l/B/E/S. Pension EPS is defined
as the difference between Net periodic pension cost (NPPC) and services
cost, then multiplied by (1-35 percent). Core EPS is the total expected
earnings (EPS) from l/B/E/S minus Pension EPS. Pension NAV equals the
market value of pension assets minus projected benefit obligations
(PBO). Core book value is the total book equity value minus Pension
NAV. Core EPS, Pension EPS, Pension NAV and Core BV are deflated by
number of shares outstanding. Post equals 1 if a firm-year comes from
the post-SFAS 158 period, and 0 otherwise. EPS Growth Forecast is the
median of long-term growth rate in l/B/E/S. Pension Indicator is a
dummy variable equal to 1 if a sample firm has a defined benefit
pension plan, and 0 otherwise. Size is the natural log of total assets.
ROA is return on assets. Leverage is the ratio between total
liabilities and total assets. Volatility is the standard deviation of
five-year ROAs. Sales growth is the growth rate of sales revenue over
the fiscal year. Spread is the bid-ask spread expressed as the
percentage of price. Segment is the number of business segments.
2-digit SIC industry dummies and year dummies are included. Year
dummies are created by fiscal year. Standard errors are adjusted for
firm clusters. P-values are presented in the parentheses below the
estimated coefficients. We use (*), (**), (***) to indicate
significance at 10 percent, 5 percent, 1 percent level, (two-tailed),
respectively.

                     (1)                       (2)
Variables            Coeff.           P-value  Coeff.           P-value

Intercept                3.740        0.544        6.094        0.280
Core EPS                10.331 (***)  0.000       10.460 (***)  0.000
Pension EPS                                       24.506 (***)  0.000
Pension NAV              1.239 (***)  0.000        0.154        0.533
Core BV                  0.463 (***)  0.000        0.426 (***)  0.000
Post                    -0.146        0.905       -0.080        0.947
Pension EPS (*)Post                              -10.668 (**)   0.039
Pension NAV (*)Post      0.074        0.824        0.677 (**)   0.032
EPS Growth Forecast      0.646 (***)  0.000        0.657 (***)  0.000
Size                     0.576        0.152        0.454        0.238
ROA                     10.382 (**)   0.020        8.515 (*)    0.054
Leverage                 6.966 (***)  0.007        5.395 (**)   0.038
Volatility             -16.390 (***)  0.002      -17.447 (***)  0.001
Sales Growth            -2.726 (**)   0.017       -2.725 (**)   0.016
Spread                  -1.293 (***)  0.000       -1.289 (***)  0.000
Segment                 -0.004        0.987       -0.082        0.585
Pension Indicator       -0.848        0.422       -0.992        0.345
Pension NAV, lag 2
Pension NAV,
lag 2 (*)Post
F test:
Pension EPS+Pension                               13.838 (***)  0.001
EPS (*)Post
Pension NAV+Pension      1.313 (***)  0.000        0.831 (**)   0.010
NAV (*)Post
Industry dummies     Yes                       Yes
Year dummies         Yes                       Yes
N                    8,080                     8,080
R-square                 0.764                     0.768

                      (3)                       (4)
Variables             Coeff.           P-value  Coeff.           P-value

Intercept                 2.929        0.647        6.368        0.271
Core EPS                 10.367 (***)  0.000       10.712 (***)  0.000
Pension EPS                                        24.590 (***)  0.000
Pension NAV               0.806 (***)  0.005        0.128        0.612
Core BV                   0.469 (***)  0.000        0.405 (***)  0.000
Post                     -0.356        0.773       -0.192        0.875
Pension EPS (*)Post                               -11.239 (*)    0.053
Pension NAV (*)Post       0.285        0.494        0.753 (**)   0.041
EPS Growth Forecast       0.685 (***)  0.000        0.665 (***)  0.000
Size                      0.582        0.163        0.418        0.289
ROA                      11.526 (**)   0.015        8.904 (*)    0.063
Leverage                  7.120 (***)  0.009        4.923 (*)    0.070
Volatility              -16.213 (***)  0.003      -20.184 (***)  0.000
Sales Growth             -2.574 (**)   0.039       -2.963 (**)   0.010
Spread                   -1.371 (***)  0.000       -1.229 (***)  0.000
Segment                  -0.020        0.903       -0.010        0.957
Pension Indicator        -0.804        0.462       -0.947        0.384
Pension NAV, lag 2        0.667 (***)  0.003
Pension NAV, lag         -0.496        0.148
2 (*)Post
F-test:
Pension EPS+Pension                                13.351 (***)  0.006
EPS (*)Post
Pension NAV+Pension       1.091 (***)  0.001        0.881 (**)   0.018
NAV (*)Post
Industry dummies      Yes                       Yes
Year dummies          Yes                       Yes
N                     7,542                     7,365
R-square                  0.764                     0.770

TABLE 6
Regressions of Stock Prices on Pension Earnings and Net Pension Assets
Value across Various Firm Characteristics and Conditions.
Table 6 reports the impact of SFAS No. 158 on value relevance of
pension earnings and net pension assets for various firm
characteristics and conditions. Panel A reports the results for
different levels of institutional ownership. Inst H is a dummy variable
equal to 1 if a sample firm's institutional ownership is higher than
the sample median, and 0 otherwise. Panel B reports the results for
different levels of analysts following and bid-ask spread. Analyst H is
a dummy variable equal to 1 if the number of analysts following a
sample firm is more than the sample median, and 0 otherwise. Spread_H
is a dummy variable equal to 1 if the bid-ask spread of a sample firm
is larger than the sample median, and 0 otherwise. Panel C reports the
results for different levels of business complexity. Segment.H is a
dummy variable equal to 1 if the number of business segments of a
sample firm is more than the sample median, and 0 otherwise. Panel D
reports the results for different levels of managerial entrenchment.
Eindex_H is a dummy variable equal to 1 if the E-index of a sample firm
is higher than the sample median, and 0 otherwise. Panel E reports the
results for different levels of board size and board independence.
Boardsize_H is a dummy variable equal to 1 if the board size of a
sample firm is larger than the sample median, and 0 otherwise.
Independence. H is a dummy variable equal to 1 if percentage of
independent directors of a sample board is higher than the sample
median, and 0 otherwise. All other variables are previously defined.
2-digit SIC industry dummies and year dummies are included. Year
dummies are created by fiscal year. Standard errors are adjusted for
firm clusters. P-values are presented in the parentheses below the
estimated coefficients. We use (*), (**), (***) to indicate
significance at 10 percent, 5 percent, 1 precent level, (two-tailed),
respectively.

Panel A. Effect of Institutional Ownership
Variables                                     Coeff.          P-value

Intercept                                      -6.709         0.153
Core EPS                    [[beta].sub.1]     10.408 (***)   0.000
Pension EPS                 [[beta].sub.2]     28.110 (***)   0.000
Pension NAV                 [[beta].sub.3]      0.130         0.682
Core BV                     [[beta].sub.4]      0.367 (***)   0.000
Post                        [[beta].sub.5]      8.885 (***)   0.000
Pension EPS (*)Post         [[beta].sub.6]    -11.242         0.125
Pension NAV (*)Post         [[beta].sub.7]      0.589         0.206
Institutional               [[beta].sub.8]      0.761         0.612
Pension EPS (*)lnst. H      [[beta].sub.9]    -10.365 (*)     0.081
Pension NAV (*)Inst.H       [[beta].sub.10]     0.006         0.987
Pension EPS (*)Post         [[beta].sub.11]     4.980         0.567
(*)lnst._H
Pension NAV                 [[beta].sub.12]     0.385         0.439
(*)Post (*)lnst_H
F-test:
Joint effect of                                -6.262         0.241
Post on the value
relevance of
Pension EPS for
firms with higher
institutional ownership
([[beta].sub.6]+
[[beta].sub.11])
Joint effect of                                 0.974 (***)   0.001
Post on the value
relevance of
Pension NAV for
firms with
higher institutional
ownership ([[beta].sub.7]
+[[beta].sub.12])
Value relevance                                17.745 (***)   0.000
of Pension EPS
for firms with
higher institutional
ownership pre-SFAS
158 ([[beta].sub.2]
+[[beta].sub.9])
Value relevance                                 0.136         0.620
of Pension NAV for
firms with higher
institutional ownership
pre-SFAS 158
([[beta].sub.3]
+[[beta].sub.10])
Value relevance of                             16.868 (***)   0.010
Pension EPS for
firms with lower
institutional
ownership
post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6])
Value relevance                                 0.719 (*)     0.088
of Pension NAV
for firms with
lower institutional
ownership post-SFAS
158 ([[beta].sub.3]
+[[beta].sub.7])

Panel A. Effect of Institutional Ownership
Variables           Coeff.            p-value

Value relevance        21.848 (***)   0.003
of Pension
EPS for firms
wtih higher
institutional
ownership
post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6]
+[[beta].sub.11])
Value relevance         1.104 (**)    0.007
of Pension
NAV for firms
with higher
institutional
ownership
post-SFAS 158
([[beta].sub.3]
+[[beta].sub.7]
+[[beta].sub.12]
Control variables   Yes
Industry dummies    Yes
Year dummies        Yes
N                   7,336
R-square                0.749

Panel B. Effect of Information Environment
                                          Analyst Foliowing
                                          Coeff.          P-value

Intercept                                 -11.418 (**)    0.027
Core EPS                 [[beta].sub.1]    10.366 (***)   0.000
Pension EPS              [[beta].sub.2]    21.033 (***)   0.000
Pension NAV              [[beta].sub.3]     0.109         0.671
Core BV                  [[beta].sub.4]     0.498 (***)   0.000
Post                     [[beta].sub.5]     9.281 (***)   0.000
Pension EPS (*)Post      [[beta].sub.6]    -0.456         0.943
Pension NAV (*)Post      [[beta].sub.7]     0.499         0.179
Analyst                  [[beta].sub.8]     6.180 (***)   0.000
Spread
Pension EPS              [[beta].sub.9]    14.142 (*)     0.056
(*)Analyst_H
Pension NAV              [[beta].sub.10]                  0.524
(*)Analyst_H
Pension EPS
(*)Spread_H
Pension NAV
(*)Spread_H
Pension EPS (*)post      [[beta].sub.11]  -27.087 (***)   0.001
(*)Analyst_H
Pension NAV (*)post      [[beta].sub.12]   -0.012         0.982
(*)Analyst_H
Pension EPS
(*)post (*)Spread_H
Pension NAV
(*)post (*)Spread_H
F-tests:
Joint effect of Post                      -27.543 (***)   0.000
on the value relevance
of Pension EPS for
firms with more
analyst
following/higher
bid-ask spread
([[beta].sub.6]
+[[beta].sub.11])
loint effect of                             0.487         0.309
Post on the
value relevance
of Pension NAV
for firms
with more analyst
following/higher
bid-ask
spread
([[beta].sub.7]
+[[beta].sub.12)

Panel B. Effect of Information Environment
                                          Bid-Ask Spread
                                          Coeff.          P-value

Intercept                                  -6.098         0.250
Core EPS                 [[beta].sub.1]    10.488 (***)   0.000
Pension EPS              [[beta].sub.2]    25.739 (***)   0.003
Pension NAV              [[beta].sub.3]     0.431         0.136
Core BV                  [[beta].sub.4]     0.430 (***)   0.000
Post                     [[beta].sub.5]     9.902 (***)   0.000
Pension EPS (*)Post      [[beta].sub.6]   -13.618         0.138
Pension NAV (*)Post      [[beta].sub.7]     0.438         0.238
Analyst
Spread                   [[beta].sub.8]    -1.327 (***)   0.000
Pension EPS
(*)Analyst_H
Pension NAV
(*)Analyst_H
Pension EPS              [[beta].sub.9]    -1.350         0.886
(*)Spread_H
Pension NAV              [[beta].sub.10]    0.367         0.235
(*)Spread_H
Pension EPS (*)post
(*)Analyst_H
Pension NAV (*)post
(*)Analyst_H
Pension EPS                                11.585         0.308
(*)post (*)Spread_H
Pension NAV              [[beta].sub.11]    0.296         0.558
(*)post (*)Spread_H
F-tests:                 [[beta].sub.12
Joint effect of Post                       -2.033         0.792
on the value relevance
of Pension EPS for
firms with more
analyst
following/higher
bid-ask spread
([[beta].sub.6]
+[[beta].sub.11])
loint effect of                             0.734 (*)     0.075
Post on the
value relevance
of Pension NAV
for firms
with more analyst
following/higher
bid-ask
spread
([[beta].sub.7]
+[[beta].sub.12)

Panel B. Effect of lnformation Environment

                     Analyst Following          Bid-Ask Spread
                     Coeff.           p-value   Coeff.        P-value

Value relevance         35.175 (***)  0.000        24.389 (***)  0.000
of Pension
EPS for firms with
more analyst
following/higher
bid-ask spread
pre-SFAS 158
([[beta].sub.2]
+[[beta].sub.9])
Value relevance          0.348        0.356         0.064        0.833
of Pension
NAV for firms
with more
analyst
following/higher
bid-ask spread
pre-SFAS 158
([[beta].sub.3]
+[[beta].sub.10])
Value relevance         20.577 (***)  0.000        12.121 (**)   0.012
of Pension
EPS for firms
with less analyst
following/lower
bid-ask spread
post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6])
Value relevance          0.608        0.106         0.869 (**)   0.020
of Pension
NAV for firms with
less analyst
following/lower
bid-ask spread
post-SFAS 158
([[beta].sub.3]
+[[beta].sub.7])
Value relevance         -6.51         0.339        23.706 (**)   0.029
of Pension
EPS for firms with
more analyst
following/higher
bid-ask spread
post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6]
+[[beta].sub.11])
Value relevance          0.596        0.260         1.165 (**)   0.010
of Pension
NAV for firms with
more analyst
following/higher
bid-ask spread
post-SFAS 158
([[beta].sub.3]
+[[beta].sub.7]
+[[beta].sub.12])
Control variables    Yes                        Yes
Industry dummies     Yes                        Yes
Year dummies         Yes                        Yes
N                    8,080                      8,080
R-square                 0.774                      0.767

Panel C. Effect of Business Complexty

Variables                                    Coeff.         P-value

Intercept                                    -6.085         0.255
Core EPS                   [[beta].sub.1]    10.481 (***)   0.000
Pension EPS                [[beta].sub.2]    24.357***      0.000
Pension NAV                [[beta].sub.3]     0.228         0.519
Core BV                    [[beta].sub.4]     0.430 (***)   0.000
Post                       [[beta].sub.5]     9.991 (***)   0.000
Pension EPS (*)Post        [[beta].sub.6]   -10.731         0.212
Pension NAV (*)Post        [[beta].sub.7]     0.499         0.274
Segment                    [[beta].sub.8]    -0.109         0.468
Pension EPS (*)Segment H   [[beta].sub.9]     0.418         0.954
Pension NAV (*)Segment_H   [[beta].sub.10]   -0.031         0.930
Pension EPS (*)post        [[beta].sub.11]    2.045         0.827
(*)Segment H
Pension NAV (*)post        [[beta].sub.12]    0.287         0.562
(*)Segment_H
F-tests:

Panel C. Effect of Business Complexity
Variables                Coeff.            P-value

Joint effect of             -8.686 (*)     0.092
Post on the value
relevance of
Pension EPS for
firms with more
business segments
([[beta].sub.6]
+[[beta].sub.11])
Joint effect of              0.786 (**)    0.015
Post on the value
relevance of Pension
NAV for firms with
more business segments
([[beta].sub.7]
+[[beta].sub.12])
Value relevance             23.939 (***)   0.000
of Pension EPS
for firms with
more business
segments pre-SFAS
158
([[beta].sub.2]
+[[beta].sub.9])
Value relevance              0.197         0.469
of Pension NAV
for firms with
business segments
pre-SFAS "158
([[beta].sub.3]
+[[beta].sub.10])
Value relevance             13.626 (**)    0.017
of Pension EPS
for firms with
less business
segments
post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6])
Value relevance              0.727 (*)     0.089
of Pension NAV
for firms with
less business
segments
post-SFAS 158
([[beta].sub.3]
+[[beta].sub.7])
Value relevance             15.671 (*)     0.055
of Pension EPS
for firms with
more business
segments post-SFAS
158
([[beta].sub.2]
+[[beta].sub.6]
+[[beta].sub.11])
Value relevance              1.014 (**)    0.025
of Pension NAV
for firms with
more business
segments
post-SFAS 158
([[beta].sub.3]
+[[beta].sub.7]
+[[beta].sub.12])
Control variables        Yes
Industry dummies         Yes
Year dummies             Yes
N                        8,080
R-square                     0.767

Panel D. Effect of Managerial Entrenchment
Variables                                  Coeff.         P-value

Intercept                                  -6.249         0.207
Core EPS               [[beta].sub.1]      10.495 (***)   0.000
Pension EPS            [[beta].sub.2]      32.043 (***)   0.000
Pension NAV            [[beta].sub.3]       0.234         0.609
Core BV                [[beta].sub.4]       0.391 (***)   0.000
Post                   [[beta].sub.5]       8.441 (***)   0.000
Pension EPS (*)Post    [[beta].sub.6]     -24.128 (***)   0.003
Pension NAV (*)Post    [[beta].sub.7]       0.666         0.362
E-index                [[beta].sub.8]      -0.299         0.260
Pension EPS            [[beta].sub.9]     -12.090         0.134
(*)Eindex H
Pension NAV            [[beta].sub.10]      0.052         0.913
(*)Eindex_H
Pension EPS            [[beta].sub.11]     23.960 (***)   0.003
(*)post (*)Eindex_H
Pension NAV            [[beta].sub.12]      0.095         0.897
(*)post (*)Eindex_H
F-tests:
Joint effect of                            -0.168         0.973
Post on the value
relevance of Pension
EPS for firms
with higher
E-index
([[beta].sub.6]
+[[beta].sub.11])

Panel D. Effect of Managerial Entrenchment
Variables                                Coeff.            P-value

Joint effect of Post on the                  0.761 (**)    0.013
value relevance of Pension NAV
for firms with higher
E-index
([[beta].sub.7]+[[beta].sub.12])
Value relevance of Pension EPS for          19.953 (***)   0.000
firms with higher E-index pre-SFAS 158
([[beta].sub.2]+[[beta].sub.9])
Value relevance of Pension                   0.286         0.223
NAV for firms with higher E-index
pre-SFAS 158
Value relevance of Pension EPS               7.915         0.363
for firms with lower E-index
post-SFAS-158
([[beta].sub.2]+[[beta].sub.6])
Value relevance of Pension NAV               0.900         0.133
for firms with lower E-index
post-SFAS 158
([[beta].sub.3.]+[[beta].sub.7])
Value relevance of Pension                  31.875 (***)   0.000
EPS for firms with higher E-index
post-SFAS 158
([[beta].sub.2]+[[beta].sub.6]
+[[beta].sub.11])
Value relevance of Pension NAV               0.995 (*)     0.095
for firms with higher E-index
post-SFAS 158
([[beta].sub.3]+[[beta].sub.7]
+[[beta].sub.12])
Control variables                        Yes
Industry dummies                         Yes
Year dummies                             Yes
N                                        6,695
R-square                                     0.758

Panel E. Effect of Board Effectiveness
                                        Board Size
Variables                               Coeff.         P-value

Intercept                               -9.270 (*)     0.096
Core EPS              [[beta].sub.1]    10.615 (***)   0.000
Pension EPS           [[beta].sub.2]    14.703 (***)   0.001
Pension NAV           [[beta].sub.3]     0.560         0.100
Core BV               [[beta].sub.4]     0.377 (***)   0.000
Post                  [[beta].sub.5]     8.342 (***)   0.000
Pension EPS (*)Post   [[beta].sub.6]     7.189         0.470
Pension NAV (*)Post   [[beta].sub.7]    -0.326         0.602
Board size            [[beta].sub.8]     0.035         0.845
Independence
Pension EPS           [[beta].sub.9]    14.462 (**)    0.029
(*)Boardsize_H
Pension NAV           [[beta].sub.10]   -0.495         0.191
(*)Boardsize_H
Pension EPS
(*)lndependence_H
Pension NAV
(*)lndependence_H
Pension EPS (*)post   [[beta].sub.11]   -22.851 (**)   0.028
(*)Boardsize_H
Pension NAV (*)post   [[beta].sub.12]     1.442 (**)   0.034
(*)Boardsize_H

                                        Board Independence
Variables                               Coeff.          P-value

Intercept                               -10.619 (*)     0.085
Core EPS              [[beta].sub.1]     10.624 (***)   0.000
Pension EPS           [[beta].sub.2]     31.087 (***)   0.000
Pension NAV           [[beta].sub.3]      0.174         0.624
Core BV               [[beta].sub.4]      0.370 (***)   0.000
Post                  [[beta].sub.5]      8.284 (***)   0.000
Pension EPS (*)Post   [[beta].sub.6]     -5.912         0.597
Pension NAV (*)Post   [[beta].sub.7]     -0.031         0.967
Board size
Independence          [[beta].sub.8]      2.437         0.285
Pension EPS
(*)Boardsize_H
Pension NAV
(*)Boardsize_H
Pension EPS           [[beta].sub.9]    -10.290         0.178
(*)lndependence_H
Pension NAV           [[beta].sub.10]    -0.065         0.868
(*)lndependence_H
Pension EPS (*)post
(*)Boardsize_H
Pension NAV (*)post
(*)Boardsize_H

Panel E. Effect of Board Effectiveness
                      Board Size                  Board Independence
Variables             Coeff.            P-value

Pension EPS (*)post                               [[beta].sub.11]
(*)lndependence H
Pension NAV (*)post                               [[beta].sub.12]
(*)lndependence H
F-tests:
Joint effect            -15.662 (***)   0.002
of Post on the
value relevance
of Pension
EPS for firms
with larger/more
independent board
([[beta].sub.6]
+[[beta].sub.11])
Joint effect              1.116 (***)   0.003
of Post on the
value relevance
of Pension
NAV for firms
with larger/more
independent board
([[beta].sub.7]
+[[beta].sub.12])
Value relevance          29.165 (***)   0.000
of Pension
EPS for firms with
larger/more
independent
board pre-SFAS 158
[[beta].sub.2]
+[[beta].sub.9]
Value relevance          -0.065         0.823
of Pension
NAV for firms with
larger/more
independent
board pre-SFAS 158
[[beta].sub.3]
+[[beta].sub.10]
Value relevance          21.892 (**)    0.013
of Pension
EPS for firms with
smaller/less
independent
board post-SFAS 158
Value relevance           0.234         0.664
of Pension
NAV for firms with
smaller/less
independent
board post-SFAS 158
[[beta].sub.3]
+[[beta].sub.7]
Value relevance          -0.959         0.882
of Pension
EPS for firms with
larger/more
independent
board post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6]
+[[beta].sub.11])
Value relevance           1.458 (***)   0.003
of Pension
NAV for firms with
Larger/more
independent board
post-SFAS 158
Control variables     Yes
Industry dummies      Yes
Year dummies          Yes
N                     5,713
R-square                  0.749

                      Board Independence
Variables             Coeff.            P-value

Pension EPS (*)post      -1.020         0.934
(*)lndependence H
Pension NAV (*)post       0.984         0.218
(*)lndependence H
F-tests:
Joint effect             -6.932         0.258
of Post on the
value relevance
of Pension
EPS for firms
with larger/more
independent board
([[beta].sub.6]
+[[beta].sub.11])
Joint effect              0.953 (**)    0.015
of Post on the
value relevance
of Pension
NAV for firms
with larger/more
independent board
([[beta].sub.7]
+[[beta].sub.12])
Value relevance          20.797 (***)   0.000
of Pension
EPS for firms with
larger/more
independent
board pre-SFAS 158
[[beta].sub.2]
+[[beta].sub.9]
Value relevance           0.109         0.705
of Pension
NAV for firms with
larger/more
independent
board pre-SFAS 158
[[beta].sub.3]
+[[beta].sub.10]
Value relevance          20.797 (**)    0.022
of Pension
EPS for firms with
smaller/less
independent
board post-SFAS 158
Value relevance           0.143         0.847
of Pension
NAV for firms with
smaller/less
independent
board post-SFAS 158
[[beta].sub.3]
+[[beta].sub.7]
Value relevance          19.777 (***)   0.007
of Pension
EPS for firms with
larger/more
independent
board post-SFAS 158
([[beta].sub.2]
+[[beta].sub.6]
+[[beta].sub.11])
Value relevance           1.127 (**)    0.040
of Pension
NAV for firms with
Larger/more
independent board
post-SFAS 158
Control variables     Yes
Industry dummies      Yes
Year dummies          Yes
N                     5,713
R-square                  0.749

TABLE 7
Regression of Stock Prices on Pension Earnings and Net Pension Assets
Value: Difference-in-Difference Approach.
Table 7 reports the regression results using a difference-in-difference
approach, comparing firms with a DB pension plan to firms without a DB
pension plan by year, industry (2-digit SIC code), total assets and
ROA. Then we require each matched pair have data in both periods before
and after SFAS No. 158. We identify 756 pairs throughout the whole
sample period. Stock price (P) is the actual stock prices taken from
l/B/E/S. Pension EPS is defined as the difference between Net periodic
pension cost (NPPC) and services cost, then multiplied by (1-35
percent). Core EPS is the total expected earnings (EPS) from l/B/E/S
minus Pension EPS. Pension NAV equals the market value of pension
assets minus projected benefit obligations (PBO). Core book value is
the total book equity value minus Pension NAV. Core EPS, Pension EPS,
Pension NAV and Core BV are deflated by number of shares outstanding.
Post equals 1 if a firm-year comes from the post-SFAS 158 period, and 0
otherwise. EPS Growth Forecast is the median of long-term growth rate
in l/B/E/S. Pension Indicator is a dummy variable equal to 1 if a
sample firm has a defined benefit pension plan, and 0 otherwise. Size
is the natural log of total assets. ROA is return on assets. Leverage
is the ratio between total liabilities and total assets. Volatility is
the standard deviation of five-year ROAs. Sales growth is the growth
rate of sales revenue over the fiscal year. Spread is the bid-ask
spread expressed as the percentage of price. Segment is the number of
business segments. 2-digit SIC industry dummies and year dummies are
included. Year dummies are created by fiscal year. Standard errors are
adjusted for firm clusters. P-values are presented in the parentheses
below the estimated coefficients. We use (*), (**), (***) to indicate
significance at 10 percent, 5 percent, 1 percent level, (two-tailed),
respectively.

Variables             Coeff.            P-value

Intercept                18.501 (**)    0.056
Core EPS                 10.838 (***)   0.000
Pension EPS              17.872 (**)    0.020
Pension NAV              -0.718         0.255
Core BV                   0.385 (**)    0.015
Post                      1.627         0.525
Pension EPS (*)Post       5.190         0.602
Pension NAV (*)Post       1.062 (**)    0.036
EPS Growth Forecast       0.654 (***)   0.000
Size                     -0.090         0.915
ROA                       8.640         0.177
Leverage                  0.283         0.949
Volatility              -12.189         0.172
Sales Growth             -4.150 (**)    0.020
Spread                   -1.301 (*)     0.099
Segment                  -0.528 (*)     0.058
Pension Indicator        -0.056         0.966
Industry dummies        Yes
Year dummies            Yes
N                     1,512
R-square                  0.814
COPYRIGHT 2019 Creighton University
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Xu, Xiaolu; Xiao, Wen; Chakraborty, Atreya
Publication:Quarterly Journal of Finance and Accounting
Article Type:Report
Date:Jun 22, 2019
Words:14735
Previous Article:The Role of Founder-CEOs and Founder-Families in IPOs.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters