The effect of SFAS 158 on the mispricing of pension plan funding

ABSTRACT

In this study, we examine whether the stock market pricing of the funding status of pension plans has improved after the adoption of SFAS 158. We find that the overvaluation of firms with severely underfunded pension plans, as documented in the prior literature, is no longer observed in the years after the adoption of SFAS 158. In addition, we find that the positive effect of SFAS 158 adoption in mitigating the mispricing of the funding status of pension plans is more pronounced for firms with a larger amount of off-balance-sheet disclosed pension liabilities. Overall, our results suggest that SFAS 158 has enhanced investors’ valuation of the funding status of pension plans, particularly due to the requirement for previously disclosed pension liabilities to be recognized on the balance sheet.

KEYWORDS:

• SFAS 158
• pension plan funding status
• market efficiency
• investor processing of information
• disclosure versus recognition

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. See for example, (i) Ball and Brown (1968) and Bernard and Thomas (1990) for PEAD, (ii) Sloan (1996), Xie (2001), and Chan et al. (2006) for accruals anomalies, and (iii) Lakonishok, Shleifer, and Vishny (1994) and La Porta et al. (1997).

2. Employer pension programs are broadly classified into two types: defined contribution (DC) and defined benefit (DB) pension plans. A major difference between the two programs is that while in DC plans both employers and employees have the discretion to choose assets for investment contributions, in DB plans only the employers have discretion to choose assets for investment in funds to be used for the employees’ retirement benefits. For more details, see Franzoni and Marin (2006) and Shivdasani and Stefanescu (2010).

3. Underfunding of pension plans means that the pension liabilities exceed the pension assets. Section 3.1.1 defines the variables related to pension plans.

4. As Franzoni and Marin (2006) point out, pension liabilities can affect future earnings and cash flows through two channels: amortization of the losses on a pension plan and mandatory contribution in cases of severe underfunding.

5. The main focus of Campbell, Dhaliwal, and Schwartz (2010) is on the passage of the Pension Protection Act of 2006. The authors find that the market reactions to firms with DB plans during the key dates related to the adoption of the law are significantly negative.

6. In Yu (2013)’s sensitivity test examining the one-year ahead return predictability of pension accounting off-balance-sheet pension liabilities, the control of known risk factors is not sufficient and the pooled cross-sectional regression is subject to the correlation across observations and over time (Fama and MacBeth 1973). Furthermore, the post-SFAS 158 period covers only two years (2006 and 2007), making it hard to make a meaningful inference of any effect on the adoption of SFAS 158 on the mispricing of pension funding status and on whether such an effect, if any, is long-run or only temporary.

7. As pointed out by the FASB, SFAS 158 does not change the basic approach to measuring plan assets, benefit obligations, or annual net periodic benefit costs. Thus, no new information or new computations other than those related to income tax effects are required. The new provision can help users to better understand pension accounting information by eliminating the need to reconcile footnote information with financial statements. This highlights the importance of the potential effect of the recognition versus disclosure of pension funding status (arising from the adoption of SFAS 158) on market mispricing.

8. In addition, the requirement to measure the pension plan status as of the balance sheet date also eliminated the possibility of managers choosing an alternative measurement date, which could have been up to three months earlier.

9. Given that investors did not have such information in the pre-SFAS 158 period and the implications of the new provisions on the investors’ pricing of pension plan funding status is not obvious, we do not conduct a formal test in this direction.

10. Relatedly, a recent study of Bratten, Choudhary, and Schipper (2013) provides evidence suggesting that disclosed items are not processed differently from recognized items when the two items are of similar reliability such as when disclosures are salient, not based on management estimates, and thus amenable to impute as-if recognized amounts. While PBO is still managers’ estimate, when the reliability of PBO does not change after the adoption of SFAS 158, it is possible to observe that the recognized liabilities are processed in the same way as the disclosed liabilities.

11. Although the pension plan related data are available on Compustat from 1980, our sample period begins from 1986 due to few observations of variables in the earlier period.

12. Compustat made two structural changes in the way it reports pension items during our sample period. Before 1987, FVPA equals PBNAA (Compustat data item #245) and PBO equals PBNVV (#243). During this period, PBO corresponds to another measure of pension obligation named the accumulated benefit obligation (ABO) because it does not reflect the projected increases in future salaries. After 1987, the FVPA is the sum of PPLAO (#287) and PPLAU (#296), and PBO is the sum of PBPRO (#286) and PBPRU (#294). Franzoni and Marin (2006) show that such breaks in reporting definitions in the time-series of pension items have little effect on the implications of pricing FR.

13. See the ‘Private Pension Plan Bulletin Abstract of 2012 Form 5500 Annual Reports,’ which is published by the U.S. Department of Labor (http://www.dol.gov/ebsa/pdf/2012pensionplanbulletin.pdf), for more details.

14. Note that the pre-SFAS 158 period contains the years 2003 through 2006, which are not included in Franzoni and Marin (2006).

15. We obtain the data on factor returns from Professor Kenneth R. French’s website: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html. We thank Prof. French for making this data available for research.

16. In an unreported analysis, we use the same period as that used in Franzoni and Marin (2006) in the pre-SFAS period (i.e., from 1980 to 2002) and find that the results are the same as those in the previous study well as the current paper.

17. Interestingly, Portfolio 9 has a significantly negative abnormal monthly returns in the post-SFAS 158 period. While this portfolio is not the main focus of the current study, the new finding of pricing error in the post period warrants a further investigation. The investigation of firm characteristics for Portfolio 9 does not indicate any particular pattern in firm characteristics such as size, B/M, return volatility, accruals and unexpected earnings surprise as well as portfolio (untabulated). However, we find from an unreported analysis that the alpha for Portfolio 9 is not significant when the post-SFAS 158 period began after the financial crisis (i.e., the portfolio returns are calculated from January of 2009). While this suggests that the mispricing of the FR portfolio 9 in the post-SFAS 158 period may be due to the volatile stock market during the financial crisis, we leave it as a future study to examine other aspects of this portfolio.

18. The factor loading on UMD is significantly negative in both the pre- and post-SFAS 158 periods, suggesting that the most underfunded firms behave like momentum losers.

19. One potential concern is that the financial statements information may not be available to investors in the first year of SFAS 158 implementation in our research design. To address this concern, we check the sensitivity of our results by allowing a sufficient time lag in calculating stock returns for the post-SFAS 158 period. In an unreported analysis, the results hold even when we use the monthly stock return data from the month of January of 2008, by which a firm’s financial statement data for the fiscal year of 2006 should be released to the market. Another potential issue is that the recent financial crisis years (2007–2008) are included in the post SFAS 158 period. Excluding the crisis years does not affect the finding that there is no more mispricing of pension information after SFAS 158 is adopted (untabulated).

20. Scaling (PBO – ABO) by the market value of equity enables us to capture the economic significance of IL relative to firm size, and at the same time, to compare IL with FR on the same dimension.

21. As an alternative robustness check, we conduct a DiD (difference-in-differences) test that examines the change in return predictability of the most underfunded firms around the adoption of SFAS 158, compared to that of the least underfunded firms. We run monthly pooled regression of 12-month ahead excess returns on a dummy variable for the most underfunded firms (FRPT1), a dummy variable for the post-SFAS period (POST), and their interaction term (FRPT1× POST), along with control variables included in Table 6. We continue to find that the post-period most underfunded firms are no longer significantly associated with negative future returns, which is consistent with our main finding (untabulated).