Skip to main content
SpringerLink
Log in

The smoothing of pension expenses: a panel analysis

  • Original Research
  • Published:
Review of Quantitative Finance and Accounting Aims and scope Submit manuscript

Abstract

The main purpose of this paper is to utilize recent developments in panel data techniques to evaluate whether the smoothing of pension expenses is neutral in its long-term effect on reported earnings. Adopting a long-term perspective, the empirical analysis also identifies sources of potential deviations. Results suggest that the current smoothing mechanism tends to induce significant biases in the recognized pension expenses. For a majority of the sample firms, the tendency is to overstate the sponsoring firms’ earnings in the long run. To a large extent, such biases reflect the combination of both ineffective amortization of the deferred gains and losses and questionable latitude in pension rate discretions.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. For instance, see Gold (2005) and Zion and Carcache (2002).

  2. By contrast, investors appear to have better understanding for balance sheet implications of the pension smoothing under SFAS 87 (Gopalakrishnan 1994).

  3. Statement of Financial Accounting Concepts No. 2, Qualitative Characteristics of Accounting Information states that relevance and reliability are the two primary qualities that make accounting information useful. Reliability rests on faithfulness, verifiability, and neutrality, while neutrality interacts with faithfulness and verifiability to affect the usefulness of the information (FASB 1980, p. 6).

  4. For instance, see Zion and Carcache (2002), Gold (2005) and Securities and Exchange Commission (SEC) (2005).

  5. Under SFAS No. 87, the delayed recognition feature also includes the delayed recognition of plan amendments, which results in unrecognized prior service cost. The current study addresses the delayed recognition of actuarial gains and losses exclusively. Unreported large sample analysis reveals that the amount of the unrecognized prior service is much smaller than that of the unrecognized gains or losses. The median ratio of the two is about one-tenth. The median ratio of the magnitude of the change in the prior service cost to the magnitude of the change in the URGL is only about four percent.

  6. Actuarial pension gains and losses are simply referred to as (pension) gains and losses in subsequent text.

  7. At companies’ discretion, this market-related value of plan assets can be a moving average of the fair value of plan assets for up to 5 years. Davis-Friday et al. (2005) show that the difference between the market-related value and the fair value of plan assets results in differences that, on average, amount to 8.5% of the reported expected returns cost component in 1998 and 2.4% in 2001.

  8. Under SFAS 87, the required footnote disclosure reports the net periodic pension cost (and its components) rather than the net pension expense. The two differ by the amount of the net pension cost that is not expensed, but is instead capitalized in inventory or other operating assets (FASB 1985, footnote 2). Pension expenses cannot be inferred under SFAS No. 87 disclosure requirements. Following prior studies (Ali and Kumar 1993; Barth et al. 1992; Coronado and Sharpe 2003), this study treats the net periodic pension cost as pension expenses.

  9. If all or almost all of a plan's participants are inactive, the amortization window is the average remaining life expectancy of the inactive participants (FASB 1985, Para. 32).

  10. Gold (2005) notes that, since the amortization window is adjusted each year to reflect emerging deviations, the remaining number of periods to spread the amount needed to be amortized does not become smaller with time. Also refer to SFAS 87, Appendix B Illustrations 4 (FASB 1985, pp. 78–83), where the active remaining service years were shown as 10 years for all consecutive years.

  11. In his testimony before the Senate Banking Committee on employers’ accounting for defined benefit pensions and other postretirement plans, Robert Herz, the FASB Chairman, explained that the initial hope was that negative and positive results would even out over time. The Chairman went on to say, “[O]f course what we found was that, in a number of companies and industries, they did not even out. They went over a cliff”.

  12. From a corporate finance theory perspective, Gold (2005) contends that the smoothing under SFAS 87 induces an upward bias into accounting earnings because the standard allows the anticipation of an equity premium in setting the ERR without adjusting for the risk of such equity investment. Based on the futility of time diversification in Samuelson (1963), Gold argues that the long-term nature of pension plans does not lead to the conclusion that the bias will cancel out as the horizon is lengthened.

  13. In a longitude study of the rate choices under SFAS 87, Blankley and Swanson (1995) document that changes in the discount rate generally lagged behind the declining yields of high-quality corporate bonds from 1987 to 1993.

  14. In this study, I use the term “reversing” to refer to the effect of both subsequent offsets and amortizations that reverse the current deferred gains or losses.

  15. The URGL are calculated using the Compustat items POAJU and POAJO. Since POAJU and POAJO give negative numbers for deferred gains and positive numbers for deferred losses, I flip the signs for expositional ease. Therefore, positive URGL it and DeferGL it indicate gains and negative losses.

  16. Following Hunt et al. (1996), I identify the presence of a major merger or acquisition through the coding of Compustat annual sales footnote that contains AB, FD, FE or FF.

  17. I identify Compustat pension firms by non-missing PBO (Compustat items PBPRU and PBPRO).

  18. All continuous variables used in this and subsequent analyses are winsorized at the 1st and 99th percentiles to avoid potential distortions from outliers. Alternatively, trimming the continuous variables reduces the sample sizes but does not change the main results in any qualitative manner.

  19. The example parallels to the illustration of the distortions from cross-sectional analyses in Hsiao (2003, p. 4).

  20. Note that such strategies result in an offsetting impact on the non-reversing deferred gains (or losses) associated with a preference for lower (or higher) earnings. Hence, they understate the biases of the smoothing shown in subsequent firm-specific regressions.

  21. Limitations of panel data include selective problems, distortions of measurement errors and short time-series dimensions (Hsiao 2003).

  22. Unlike panel unit tests proposed by Moon and Perron (2004) and Bai and Ng (2004), which require the number of panels (N) to be small relative to the time dimension (T), i.e., N/T → 0, the Pesaran (2007) test is valid and has satisfactory size and power for panels with N and T of the same orders of magnitudes, even for very small sample sizes (e.g., when N and T equal 10) (Pesaran 2007).

  23. Untabulated results with alternative test specifications, such as including trend and with lag of 1 through 7, are qualitatively similar.

  24. Panel unit root tests using the augmented Dickey-Fuller (ADF) developed by Maddala and Wu (1999) return even weaker rejection results, perhaps because the ADF test does not account for cross-sectional dependency and is hence susceptible to size distortion and low power. Only four out of 48 industries reject the null at the conventional levels.

  25. The change in the URGL for a given year includes deferred gains (or losses), as well as potential amortization of prior URGL. Because H2a concerns the neutrality of the smoothing mechanism, including the effect from both the delayed recognition and the amortization scheme, the definition of DeferGL it here suits the objective to test whether deferred gains (or losses) are offset or amortized in subsequent periods. However, if the change in the URGL consists primarily of the amortization component, alternative explanation for non-zero intercept in (1) may arise as the amortization of URGL would be persistent, depending on subsequent gains and losses (also see Hann et al. 2007, footnote 9). I address the concern with alternative measurement for DeferGL it that excludes the amortization component in Sect. 5.

  26. Mean-reversion of DeferGL it is expected and specifically modeled in Eq. (1) and subsequent models. The question is the value of the mean that DeferGL it reverts to. If DeferGL it reverts to a nonzero mean, the URGL is likely to build up and follow a non-stationary process.

  27. To check for sensitivity of the firm-specific estimates, I use rolling windows with alternate lengths of 8, 10 and 12 years, respectively, to estimating regression (1) and obtain quantitatively similar results.

  28. The year dummies are omitted from Eq. (2) for brevity.

  29. Under the assumptions that there is no serial correlation in ξ it and that the explanatory variables are weakly exogenous (not correlated with future transformed error terms), the Arellano and Bond (1991) two-step GMM estimation procedure first assumes that the error terms are independent and homoskedastic across panels and over time. The second step of the procedure relaxes the assumptions of independence and homoskedasticity through using the residuals obtained in the first steps to construct a consistent estimate of the variance-covariance matrix.

  30. The Sargan (1958) test is inconsistent if non-sphericity is suspected in the errors (Roodman 2006).

  31. If the estimation is exactly identified, detection of an invalid instrument would then be infeasible (Roodman 2006).

  32. For non-pay-related plans, ChgSR it , AdjChgSR it and MDChgSR it are set to zero.

  33. To the extent that AdjChgDR it and AdjChgSR it reflect peculiar structural changes in the benefit plans, AdjChgDR it and AdjChgSR it are proxies for managerial discretion in pension rate choices and thus contain measurement errors.

  34. The system GMM specifications are identical to those outlined in Sect. 3.

  35. Under SFAS 87, the corridor is 10% of the market-related value of plan assets or the beginning PBO, whichever is larger. Davis-Friday et al. (2005) find that a majority of their sample firms use some form of smoothed fair values as the market-related value. However, the market-related value cannot be inferred from sponsoring firms’ footnote disclosure prior to SFAS 132 (effective in 1998). I thus use the fair value as a proxy for the market-related value of plan assets.

  36. The results are available upon request.

References

  • Ali A, Kumar KR (1993) Earnings management under pension accounting standards: SFAS 87 versus APB 8. J Account Audit Finance 8:427–447

    Google Scholar 

  • Amir E, Benartzi S (1998) The expected rate of return on pension funds and asset allocation as predictors of portfolio performance. Account Rev 73:335–352

    Google Scholar 

  • Arellano M, Bond S (1991) Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Rev Econ Stud 58:277–297

    Article  Google Scholar 

  • Arellano M, Bover O (1995) Another look at the instrumental-variable estimation of error components models. J Econom 68:29–52

    Article  Google Scholar 

  • Bai J, Ng S (2004) A PANIC attack on unit roots and cointegration. Econometrica 72:1127–1177

    Article  Google Scholar 

  • Barth M, Beaver W, Landsman W (1992) The market valuation implications of net periodic pension cost components. J Account Econ 15:27–62

    Article  Google Scholar 

  • Bergstresser D, Desai MA, Rauh J (2006) Earnings manipulation, pension assumptions and managerial investment decisions. Q J Econ 121:157–194

    Google Scholar 

  • Blankley AI, Swanson EP (1995) A longitudinal study of SFAS 87 pension rate assumptions. Account Horiz 9:1–21

    Google Scholar 

  • Blundell R, Bond S (1998) Initial conditions and moment restrictions in dynamic panel data models. J Econom 87:115–143

    Article  Google Scholar 

  • Brown S (2004) The impact of pension assumptions on firm value. Working paper, Emory University

  • Chambers D, Linsmeier TJ, Shakespeare C, Sougiannis T (2007) An evaluation of SFAS No. 130 comprehensive income disclosures. Rev Acc Stud 12:557–593

    Article  Google Scholar 

  • Coronado JL, Sharpe SA (2003) Did pension plan accounting contribute to a stock market bubble? Brookings Pap Econ Act 1:323–371

    Article  Google Scholar 

  • Davis-Friday PY, Miller JS, Mitterlstaedt HF (2005) Market-related values and pension accounting. Working paper, University of Nortre Dame

  • Defond ML, Park CW (1997) Smoothing income in anticipation of future earnings. J Account Econ 23:115–139

    Article  Google Scholar 

  • Fama EF, French KR (1997) Industry costs of equity. J Financ Econ 43:153–193

    Article  Google Scholar 

  • Fama EF, MacBeth JD (1973) Risk, return, and equilibrium: empirical tests. J Polit Econ 81:607–636

    Article  Google Scholar 

  • Financial Accounting Standards Board (FASB) (1980) Statement of financial accounting concepts no. 2: qualitative characteristics of accounting information. FASB, Stamford

    Google Scholar 

  • Financial Accounting Standards Board (FASB) (1985) Statement of financial accounting standards no. 87: employers’ accounting for pensions. FASB, Stamford

    Google Scholar 

  • Financial Accounting Standards Board (FASB) (2007) Board meeting. Phase II: postretirement benefit obligations, including pensions—agenda decision. Available at http://www.fasb.org/board_meeting_minutes/08-29-07_pbo.pdf. Accessed 29 Oct 2008

  • Godwin JH, Goldberg SR, Duchac JE (1996) An empirical analysis of factors associated with changes in pension-plan interest-rate assumptions. J Account Audit Finance 11:305–322

    Google Scholar 

  • Goel AM, Thankor AV (2003) Why do firms smooth earnings? J Bus 76:151–192

    Article  Google Scholar 

  • Gold J (2005) Accounting/actuarial bias enables equity investment by defined benefit pension plans. North Am Actuar J 9(3):1–21

    Google Scholar 

  • Gopalakrishnan V (1994) The effect of recognition vs. disclosure on investor valuation: the case of pension accounting. Rev Quant Finance Account 4(4):383–396

    Article  Google Scholar 

  • Hann R, Helflin F, Subramanyam KR (2007) Fair-value based pension accounting. J Account Econ 44:328–358

    Article  Google Scholar 

  • Hansen L (1982) Large sample properties of generalized method of moments estimators. Econometrica 50:1029–1054

    Article  Google Scholar 

  • Hsiao C (2003) Analysis of panel data, 2nd edn. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Hunt A, Moyer SE, Shevlin T (1996) Managing interacting accounting measures to meet multiple objectives: a study of LIFO firms. J Account Econ 21:339–374

    Article  Google Scholar 

  • Maddala GS, Wu S (1999) A comparative study of unit root tests with panel data and a new simple test. Oxford Bull Econ Stat 4(Special Issue):631–652

    Article  Google Scholar 

  • Mitra S, Hossain M (2009) Value-relevance of pension transition adjustments and other comprehensive income components in the adoption year of SFAS No. 158. Rev Quant Finance Account 33(3):279–301

    Article  Google Scholar 

  • Moon HR, Perron B (2004) Testing for a unit root in panels with dynamic factors. J Econom 122:81–126

    Article  Google Scholar 

  • Newey WK, West KD (1987) A simple positive definite heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55:703–708

    Article  Google Scholar 

  • Pesaran HM (2007) A simple panel unit root test in the presence of cross section dependence. J Appl Econom 22:265–312

    Article  Google Scholar 

  • Petersen MA (2008) Estimating standard errors in finance panel data sets: comparing approaches. Rev Financ Stud (forthcoming)

  • Roodman D (2006) How to do xtabond2: an introduction to “difference” and “system” GMM in Stata. Working paper, Center for Global Development

  • Samuelson PA (1963) Risk and uncertainty: a fallacy of large numbers. Scientia 6(April/May):1–6

    Google Scholar 

  • Sargan J (1958) The estimation of economic relationships using instrumental variables. Econometrica 26:393–415

    Article  Google Scholar 

  • Securities and Exchange Commission (SEC) (2005) Report and recommendations pursuant to Section 401(c) of the Sarbanes-Oxley Act of 2002 on arrangements with off-balance sheet implications, special purpose entities, and transparency of filings by issuers, submitted to the President of the United States, the Committee on Banking, Housing, and Urban Affairs of the U.S. Senate, and the Committee of the U.S. House of Representatives in June 2005. Washington, DC

  • Windmeijer F (2005) A finite sample correction for the variance of linear efficient two-step GMM estimators. J Econom 126:25–51

    Article  Google Scholar 

  • Zion D, Carcache B (2002) The magic of pension accounting I. Credit Suisse First Boston (Sep 2002)

  • Zmijewski M, Hagerman R (1981) An income strategy approach to the positive theory of accounting standard setting/choice. J Account Econ 3:129–149

    Article  Google Scholar 

Download references

Acknowledgments

This paper is based on my dissertation at Boston University. I thank members of my dissertation committee, Zvi Bodie, Alison Kirby Jones, and especially Krish Menon (chair) for guidance and discussions. Helpful comments and suggestions were received from workshop participants at the 2009 American Accounting Association Annual Meeting, Boston University, and University of Cincinnati. I would also like to thank Eugene Choo, Alex Maynard, and Bob Reed for helpful consultations. All errors are my own.

Author information

Authors and Affiliations

  1. College of Business, University of Cincinnati, 2925 Campus Green Drive, Cincinnati, OH, 45221-0211, USA

    Xiaowen Jiang

Authors
  1. Xiaowen Jiang
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Xiaowen Jiang.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Jiang, X. The smoothing of pension expenses: a panel analysis. Rev Quant Finan Acc 37, 451–476 (2011). https://doi.org/10.1007/s11156-010-0213-0

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11156-010-0213-0

Keywords

JEL Classification

Search

Navigation