Assuming the Worst: The Shifting Sands of Pension Accounting

Byrne, Alistair; Clacher, Iain; Hillier, David; Hodgson, Allan Article Assuming the Worst: The Shifting Sands of Pension Accounting Journal of Accounting and Management Information Systems (JAMIS) Provided in Cooperation with: The Bucharest University of Economic Studies Suggested Citation: Byrne, Alistair; Clacher, Iain; Hillier, David; Hodgson, Allan (2013) : Assuming the Worst: The Shifting Sands of Pension Accounting, Journal of Accounting and Management Information Systems (JAMIS), ISSN 2559-6004, Bucharest University of Economic Studies, Bucharest, Vol. 12, Iss. 2, pp. 190-212 This Version is available at: https://hdl.handle.net/10419/310522 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/4.0/ Accounting and Management Information Systems Vol. 12, No. 2, pp. 190 –212, 2013 ASSUMING THE WORST: THE SHIFTING SANDS OF PENSION ACCOUNTING Alistair BYRNE Towers Watson Actuarial Consulting, London, United Kingdom Iain CLACHER 1 University of Leeds, United Kingdom David HILLIER University of Strathclyde, United Kingdom Allan HODGSON University of Queensland Business School, Australia ABSTRACT Accounting for defined benefit pension plans is complex, and given the magnitude of many of these schemes relative to their corporate sponsor, understanding whether pension disclosures are value relevant is key to improving the quality of financial reports. The application of fair value accounting for pensions allows for a high level of managerial discretion with respect to ex ante accounting choices. Utilizing a sample of firms that apply FRS-17, we examine the main determinants of the assumptions managers use to arrive at pension scheme valuations. We find significant differences in the stated assumptions across companies, auditors and actuaries. Further, managers display considerable variation in conservatism when implementing fair value accounting, and this variation is related to scheme-specific characteristics, such as asset allocation and pension plan solvency. Crucially, pension disclosures are found to be value relevant, therefore, managers are able to present pension disclosures in a more favorable light, and this is reflected in prices. As a result of the observed inconsistency in reporting across firms, and the value relevance of these disclosures, this brings into question the efficacy of fair value accounting for assessing pension values. 1 Corresponding author: Leeds University Business School, Maurice Keyworth Building, University of Leeds, Clarendon Road, Leeds, United Kingdom, LS2 9JT Tel:(+44) 113 343 6860; E-mail address: [email protected].ac.uk Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 191 Pension Accounting Standards; FRS-17 Pensions Valuation; Managerial Pension Discretion; Pensions and Fair Value Accounting JEL codes M410; M480; G140 INTRODUCTION Proponents of fair value accounting argue that historical cost valuation obscures the “true” underlying economic position of the firm, and that fair value provides a better measure of fundamental value. Conversely, critics argue the transitory nature of fair value injects additional volatility into financial reports that are already difficult to assess. For example, prior research indicates that market participants are unable to reach a consensus on information (accruals) presented in annual reports, particularly when it is complex (Sloan, 1996; Hirst, 1998). Similarly, pension accounting under fair value has the potential to remain opaque and problematic for users of financial accounts. There are two main reasons for this; first, pension valuation is complex. Any assessment of the liabilities in a pension scheme requires detailed mortality calculations and forecasts on future macroeconomic conditions. Second, fair value accounting for pensions provides considerable discretion to management. The accounting assumptions used in pension valuation are ultimately decided upon by management. Although there are a number of factors that guide these assumptions. The accounting standard itself sets some parameters for these estimates, and they are arrived at under the guidance of the firm’s actuary, and are monitored and approved by the firm’s auditor. Despite this, they are potentially open to manipulation within broad confidence intervals. For example, if there is a large variation in pension assumptions across firms, fair value accounting fails in one of its key goals ― namely the provision of transparent, consistent and informative financial statements. In the U.S., SFAS-87 has come under increased criticism and pressure from regulators and industry amid calls for a move towards fair value pension accounting. The CFA Institute stated that the SFAS-87 method of accounting “…imposes a huge and costly burden” on the users of financial accounts. The U.S. Senate Finance Committee also threatened legislation that would remove the complex smoothing mechanism of actuarial gains and losses under SFAS-87. In 2005, the Securities and Exchange Commission (SEC) concluded that balance sheets are “… often not transparent as to the true funded status of pension plans” 1 leading to calls that pension accounting should be reformed by the Financial Accounting Standards Board (FASB). In response, FASB proposed a two-stage process to reform pension accounting, the first part of which was the introduction of FAS-158, which came into effect for fiscal year ends after December 15, 2006.  Accounting and Management Information Systems Vol. 12, No. 2 192 Most work on pension accounting is focused on the value and credit relevance of fair value footnote disclosures under SFAS-87. Hann et al. (2007) for example, found that fair value footnote disclosures did not improve the information quality of financial reports and did not allow for an assessment of how management implements fair value when there is a requirement of full recognition in the balance sheet. Although the U.S. has yet to fully adopt a standard that is equivalent to the fair value requirements of IAS-19, the introduction of FAS-158 has moved US standards closer to full fair value pension accounting and there is evidence that the use of aggressive assumptions by management still exist report despite the fact that almost all the fair value disclosures only appear in the footnotes to the accounts (Grant et al., 2007). Based on the above evidence, and the push for full fair value accounting for pensions in the U.S., we analyze how fair value pension accounting has been implemented in practice based on a sample of FRS-17 disclosures from 2001-2004. This allows for us to not only analyze whether fair value pension accounting provides information about the value and risk of pension schemes, but to do so under changing economic circumstances e.g. falling equity values and changing bond yields; both of which are crucial in fair value pension accounting. We make four main contributions to the literature. First, we document the variation in assumptions that management apply when accounting for pensions under fair value. One of the fundamental reasons for adopting fair value is to make the information in financial accounts consistent and representative across firms. If there is significant variation in accounting assumptions across firms, this calls into question the decision usefulness of pension disclosures. Second, we analyze the role and impact of auditors and actuaries on managerial discretion by investigating whether the variation in assumptions across firms can be attributed to either of these external groups. Although audit firms and actuarial firms are likely to have similar technologies, there is scope for different firms to have different ‘house views’ on particular assumptions, and this in turn may influence the choice of the audit or actuarial firm. Conversely, if there is considerable variation in the assumptions used across clients of a particular auditor and/or actuary (i.e. no consistent house view), this suggests that actuaries and auditors adopt the assumption process applied by individual firms. Third, we consider the determinants of both managerial choice and conservatism in pension accounting and their relation with firm characteristics. Prior research has found strong links between the percentage of pension assets held in equity, the expected return on plan assets and corporate events (Bergstresser et al., 2006). Our final contribution is to analyze the value relevance of the assumptions that are used to arrive at the accounting amounts as well as the fair value disclosures presented in the annual report. Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 193 Briefly, our results are as follows. We document that the difference in underlying pension assumptions across firms is substantial. Essentially, there are basic economic reasons why discount rates and expected rates of return on pensions should be similar across firms, but this is not what we observe. Further, the differences are not related to the identity of the firm’s actuary or auditor, suggesting that different ‘house views’ is not an explanation. We also report that management have different valuation objectives depending on the solvency of the pension scheme. Companies with the greatest level of solvency i.e. the ratio of pension assets to pension liabilities, have the highest discount rates and discount rate spread assumptions. In addition, we find that firms with large pension scheme deficits, relative to the size of the firm, tend to choose higher equity return and spread assumptions. Management appear to choose assumptions that maximize the level of reported financial income that can be derived from pension scheme assets. Finally, we show that the assumptions underlying the pension calculation impact prices; with pension funding levels, liabilities and assets all affecting share value. One interpretation is that the external market views both the assets and liabilities of the pension scheme as the assets and liabilities of the firm which is consistent with the corporate view of pensions. The rest of the paper is set out as follows. Section one provides an overview of FRS-17. In section two, we outline our motivation and develop the hypotheses that are tested in the paper. Section three describes the data and the methodology. Section four discusses the empirical results and the last section concludes. 1. BACKGROUND 1.1. Financial Reporting Standard (FRS) 17 The introduction of FRS-17 fundamentally changed how firms account for defined benefit pensions in the UK. Until 2001, pension accounting was governed by the Statement of Standard Accounting Practice 24 (SSAP-24). This standard, however, was widely criticized as not providing useful or comparable disclosure of the underlying risks of company pension schemes. One of the major criticisms was that the Standard afforded management too much discretion in how they accounted for pensions. After wide consultation, the Accounting Standards Board (ASB) issued FRS-17, which applied to all companies reporting financial statements after June, 2001. The framework for FRS-17 can be split into two broad categories: methodological and disclosure. Unlike previous standards, which allowed the actuary and/or management to select the actuarial method of liability calculation, 2 FRS-17 specified that liabilities must be calculated using the projected unit method 3 ― an accrued benefits valuation model which takes account of the right to benefits earned by scheme members by allowing for future increases in the level of Accounting and Management Information Systems Vol. 12, No. 2 194 pensionable salaries and the value of pensions in payment. The standard also set out the appropriate discount rate to be applied when calculating the present value of the pension liability; namely, the yield on a high quality (AA-rated) bond of equivalent duration to the pension liabilities. This rate was argued reflect both the time value of money and a premium for the risk of the scheme. 4 The second focus of the standard was on disclosure, which was broken down into four categories: valuation assumptions, asset return assumptions, pension costs, and recognition. The required valuation assumptions are the rate of inflation, wage growth, pension growth and discount rate. The pension scheme assets were to be recorded at fair (market) value and their valuation does not require actuarial assumptions. The disclosure of the fair value of pension assets was split into four broad asset classes: equities, bonds, property and ‘other’. ‘Other’ assets are generally cash and annuities, although some firms also hold insurance contracts or small exposures to managed funds. The company was also required to disclose the expected rate of return on the various asset classes. While the expected rates of return assumptions do no affect the stated asset values, they do affect the amount of pension income credited to the profit and loss account. In addition to the different assumptions, a detailed disclosure of the costs of the defined benefit scheme ―including, current service cost, past service cost, actuarial gains and losses (including the difference between the actual and expected return on scheme assets), and any historical adjustments to pension costs as a result of changes in the level of benefit provision. The final disclosure is the difference between the values of assets and liabilities of the pension scheme on the balance sheet of the firm. Where the fair value of assets exceeds the present value of the pension liability, the scheme is in surplus on an FRS-17 basis, and a net asset should appear on the balance sheet. Conversely, where the value of scheme assets are less than the present value of the pension liability, a net liability must appear on the balance sheet. To mitigate the impact of applying market values, the standard separates the normal pension cost and valuation impacts. In so doing, the volatility of market values is lessened as there is a normal pension cost charge against the profit and loss, and any variations that occur year-on-year are included in the Statement of Total Recognised Gains and Loses (STRGL) and taken directly to reserves. Any changes in market values, demographics and other basis measurements are, therefore, accounted for in the STRGL rather than on the face of the profit and loss account. In doing so, the ASB argued that this would allow for a predictable pension charge and that the balance sheet reflects the employer’s liability, while at the same time accounting for the true liability of a scheme based upon market values. Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 195 1.2. Pension Accounting under FRS-17: An example Two stylized examples are presented to illustrate the interactions between different pension components, and to show how management could manipulate the assumptions to impact upon the size of the liability faced by the firm, and the income it can derive from the pension scheme assets. 1.2.1. Liability Manipulations Pension liabilities increase year on year as a function of increases in employee wages, changes to the benefits provided and projections of future mortality and interest rates. If we assume that the present value of this payment is currently valued at £500m 5 based upon a discount rate that is taken from the prevailing AA bond yield (for the example we assume a rate of 5%). The management of the firm may choose to apply a higher discount rate thereby reducing the present value of the reported liability. Consequently, the change in the liability from one year to the next will be underestimated in the annual report. Through the application of an ‘actuarial rule of thumb’ Bozewicz (2004) highlights the impact of small changes to the chosen discount rate and the present value of the pension liability. Where, New Liability = (Old Liability)*(1.06-4∆ ) and, -4∆=(-4)*(Increase/(Decrease) In the Discount Rate) From this, if we assume a an increase in the discount rate 5.00% to 5.50% then New Liability = (£500m)*(1.06(-4)*(0.005)) = £445m The 0.5% change in the discount rate has therefore reduced the present value of the pension liability by 12%. 1.2.2. Profit and Loss Manipulations Firms also have considerable scope to manipulate the pension assets for crediting the profit and loss account under other financial income. If we assume that a firm’s pension assets are 100% invested in equity with a liability of £100m with assets of £80m this means that there is a £20m deficit in the scheme. Further, the discount rate is assumed to be 5% and the expected return on equity 10%. From one year’s unwinding of the scheme, the interest cost will be £5m (5% discount rate multiplied by the £100m pension liability). The expected return on plan assets in the same year will be £8m (10% return on the £80m of plan assets). In reconciling this to the profit and loss statement, there would be an interest charge of (£5m) and a financial income credit of £8m. The profit and loss will therefore be credited with Accounting and Management Information Systems Vol. 12, No. 2 196 a net income of £3m. Consequently, when the difference between the discount rate and the expected return on plan assets is increased, then firms can credit the profit and loss with additional income from the pension assets. 2. RESEARCH QUESTIONS Underpinning pension valuation is the notion that fair value accounting will improve the reporting quality of financial accounts. If this is true, it is to be expected that profit and loss statements will become more meaningful, transparent and comparable. However, the perceived benefits of fair value in theory and its application in practice may differ considerably. We therefore first examine whether the assumptions underlying the calculation of the different pension components are consistent across firms. Where this is so, the application of fair value would address some of the concerns about pension accounting that were raised under SSAP-24. This leads us to the following research question: RQ1: Is the choice of pension accounting assumptions by management consistent and unbiased across firms? Another facet of consistency in reporting is the relationship that different advisors and/or external bodies have with the firm. Firms employ actuaries to advise on their scheme and auditors to provide an objective assessment of the quality of their financial reports. It is possible that any patterns observed in the cross-section of assumptions are a function of who advises the firm, since different advisors may hold ‘house views’ on the various assumptions. Similarly, the auditor may hold a ‘house view’ on what assumptions are acceptable and represent a fair and true view of a firm’s pension liability. Our second research question is therefore: RQ2: Are the pension accounting assumptions of a firm related to the identity of the firm’s actuary and/or the firm’s auditor? Despite the intention for FRS-17 to provide a more prescriptive basis of pension accounting, management still have considerable latitude in the underlying assumptions applied. Bergstresser et al. (2006) focus on the sensitivity of firm earnings to the expected return on pension plan assets. Their results show that where a firm’s income is sensitive to the expected return on pension assets (for example where pension assets are large relative to firm size), management are more likely to choose higher expected return assumptions. They also find that the adoption of such optimistic assumptions is linked to corporate events such as takeovers or the exercise of share options by management. In our sample, equity investments account for the vast majority of pension assets with the average pension portfolio in 2001 consisting of 70% equity, 25% bonds and 5% other assets. This high allocation to equity has implications for the Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 197 potential for management to manipulate earnings via the assumptions for expected returns on pension scheme assets. Consequently, if management wish to boost reported income from pension assets then they would hold larger amounts of equity in the pension portfolio and use a higher expected return figure for those assets. Our third research question is therefore: RQ3: Is the expected return on equity assumption related to the proportion of equity in the pension assets? In the U.S. there has been little scope for management to select a favorable pension discount rate because since 1993 the discount rate has been based upon Moody’s AA interest rate index. 6 As a result, previous studies have not considered the determinants of the choice of discount rate. The example provided in section two highlights how sensitive the pension liability is to small changes in the discount rate; and under FRS-17 management have greater discretion over the discount rate. This therefore raises several questions about the behavior of management in exercising this discretion. The first is there is no consistent relationship between the chosen discount rate and pension scheme characteristics. If this is the case, it implies that management are applying fair value accounting in the spirit of the standard. Under FRS-17 the required discount rate is the yield on an AA-rated corporate bond of equivalent duration to the pension liabilities, which may be a sufficiently prescriptive requirement to limit the exercise of management discretion. However, the above raises a number of other potential mitigating outcomes that warrant investigation. First, when a firm’s pension liability is large relative to the size of the firm then the perception is that the firm is exposed to a significant pension risk. In response, management may elect to apply a higher discount rate to reduce the perceived risk of the firm. Thus research question four is: RQ4a: Do those firms with the largest relative pension liability select a larger discount rate? It may be however that the relative size of the pension liability is not what concerns management. It may be the case that management are concerned with the level of scheme solvency. Solvency, measured as pension assets divided by pension liabilities, is the figure that is regularly quoted in the financial press. Scheme solvency is a function of both the increases in the pension liability and/or a reduction in the value of the assets held to meet the pension liability. Where there are large fluctuations in asset values, a scheme will appear to have a volatile solvency level. 7 Management may therefore choose to apply a higher discount rate to present a ‘stable’ solvency ratio in the annual report. Accounting and Management Information Systems Vol. 12, No. 2 204 Table 3. Sign Test for Distribution of Assumptions across Auditor and Actuary Discount Rate (5.48%) Wage Growth (3.70%) Equity Return (7.10%) Discount Rate Spread (1.78%) Equity Return Spread (1.62%) Mean t-Stat Mean t-Stat Mean t-Stat Mean t-Stat Mean t-Stat Panel A: Auditor comparison Auditor (A) 5.62* 7.04 4.04* 4.65 7.58* 4.50 1.58* -3.48 1.97 1.44 Auditor (B) 5.58* 5.06 4.04* 4.60 7.77* 8.29 1.54* -4.02 2.20* 6.06 Auditor (C) 5.59* 6.43 3.97* 4.24 7.62* 5.75 1.62* -3.28 2.03* 2.94 Auditor (D) 5.60* 8.10 4.07* 6.77 7.80* 14.63 1.51* -6.19 2.21* 10.10 Panel B: Actuary comparison Actuary (A) 5.60* 2.55 3.85 1.39 7.91* 6.40 1.75 -0.75 2.31* 4.63 Actuary (B) 5.58* 4.29 4.02* 3.61 7.84* 9.54 1.55* -3.15 2.26* 7.40 Actuary (C) 5.59* 5.82 4.11* 5.47 7.42 0.60 1.47* -4.95 1.84* -2.32 Actuary (D) 5.63* 5.64 3.96* 4.57 7.60* 25.70 1.66* -3.86 2.00* 19.94 Actuary (E) 5.59* 6.83 4.02* 3.53 8.07* 4.43 1.56* -2.35 2.48 1.83 Note: This table presents a pooled cross-sectional analysis of the pension assumptions across auditors in Panel A, and actuaries in Panel B. We analyze whether the mean assumption passed is significantly different from expected values. The table presents the mean assumption across auditors/actuaries and the associated t-statistic. Our expected values are the median AA bond yield for the discount rate, average wage growth for the private sector in the UK, and the historical average return on equity. Expected values are given in parenthesis under each assumption. * indicates significant at 95%. Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 205 Research questions (RQ’s 4-7) ask whether managers use pension accounting to increase reported income from pension scheme assets. The work of Bergstresser et al. (2006) finds strong results for management deriving income from pension assets based upon assumed returns, pension portfolio composition and corporate events. Consequently, pension derived financial income that credits the profit and loss increases opaqueness as investors now have more difficulty in estimating the ‘fundamental’ profitability of the firm. Table 4. Regression Analysis of Managerial Choice and Conservatism Equity Return Discount Rate Discount Rate Spread Equity Return Spread Intercept -22.51 -9.28 13.75 -18.48 (-1.81)** (-0.47) (1.03) (-1.34) Surplus/Total Assets -36.31 6.27 -3.43 -34.30 (-2.68)* (0.29) (-0.24) (-2.28)* Solvency 20.10 54.76 20.44 2.77 (3.39)* (5.83)* (3.23)* (0.42) Equity Percentage 17.39 4.80 -2.54 14.42 (3.27)* (0.56) (-0.44) (2.45)* Size -0.31 -5.57 -2.80 1.44 (-0.23) (-2.53)* (-1.89)** (0.94) Book-toMarket -5.43 -0.53 -4.20 -2.67 (-2.17)* (-0.13) (-1.55) (-0.96) Debt-to-Equity 2.19 -3.01 1.00 2.01 (1.24) (-1.07) (0.53) (1.03) Note: This table presents the results for fixed effects regressions for the determinants of the pension assumptions and managerial conservatism. The table presents the regression coefficient and immediately below is the corresponding t-statistic. * indicates significant at 99%, ** indicates significance at 95% and *** indicates significance at 90%. The dependent variable for each model is presented at the head of each column and the independent variables are presented in the far left column. Size is measured by the log market value the market-to-book ratio is the market value of equity/book value of equity, debt-to-equity is total debt/market value of equity. Surplus to total assets is the pension surplus (deficit)/ total assets, funding is measured by pension assets/pension liabilities and the equity percentage which is calculated as equity/total pension assets. The standardized assumption is calculated by (Assumptionit -  t)/Standard Error  t. Table 4, column 1, presents our results for the expected return of equity assumption. Results are consistent with Bergstresser et al. (2006) in that, where equity is the largest component of the pension portfolio, the assumed return on equity is higher. We therefore find that portfolio composition and expected return assumptions are related. For the discount rate analysis we find that firms that have Accounting and Management Information Systems Vol. 12, No. 2 206 the highest funding ratio of pension assets to liabilities apply higher discount rates but the size of the pension liability relative to the firm is not a significant determinant of the discount rate. Further the size of the surplus/deficit is insignificant. 15 This suggests that management are concerned only about the perceived solvency of their pension scheme. The solvency variable is commonly reported in the financial press and, as such, will impact the markets perception of firm risk. Another explanation is that well-funded schemes may be concerned about transitory fluctuations in the solvency of the pension scheme. As a result, the selection of a larger discount rate will understate the ‘true’ liability and will reduce the impact of large asset fluctuations. In addition, we examine the level of managerial conservatism that is applied in calculating the present value of the pension liability. The selection of a high discount rate alone is not sufficient to minimize the pension liability. Our discount rate spread variable proxies for the level of conservatism that management are applying in estimating their pension liability. Where the spread is large then management are not applying prudent assumptions. The results in column (3) of Table 4 show that the least conservative estimates are used by firms that have the highest solvency levels. This supports the results in column (2) that managers are concerned about the perceived risk of the firm. As a result, they choose assumptions that understate the liability of the firm. This again supports our assertion that fair value has not improved the transparency of balance sheets as management are systematically understating their pension liability, thus reducing the perceived risk of the firm. Next we consider the ability of management to derive income from the pension assets by employing a large equity spread. In column (1) of Table 4 our analysis on scheme solvency and the surplus/(deficit) to total assets at first appear to be at odds as those schemes that have high solvency and those schemes that have large deficits relative to firm size both adopt higher expected return assumptions. However, when the spread between the expected and return on equity analysis is considered in conjunction with the expected return analysis in column 1, then the result becomes more intuitive. In the final column of Table 4 it can be seen that for the spread variable, solvency becomes insignificant but the higher the equity percentage the higher the net equity return spread. This is important, as we find that across management there are two different objectives. For those schemes that are well funded then management are concerned about the perceived risk and solvency of the pension scheme. Where they adopt a higher discount rate and discount rate spread then they will incur a high interest cost. To offset this interest cost the management then assume a higher net expected return for equity to offset the cost. Conversely, where schemes have significant solvency/funding concerns, management do not elect to manipulate the size of the liability/deficit. Instead they choose to maximize the financial income that can be derived for the profit and loss from the assets of the pension scheme. They Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 207 therefore adopt lower discount rates, higher expected returns on equity, and maximize the equity return spread. 4.5. Value relevance Table 5. Value Relevance Regression Analysis of Pension Assumptions Market Value of Equity Intercept 4.78 1.13 4.44 2.98 (1.62) (0.64) (11.12)* (6.29)* Discount Rate -0.18 - - - -(0.34) Equity Return - 0.34 - - (1.51) Discount Rate Spread - - -0.41 - -(1.73)*** Equity Return Spread - - - 0.38 (1.77)*** Firm Assets 0.06 0.06 0.06 0.06 (4.82)* (4.71)* (4.75)* (4.70)* Firm Debts -0.03 -0.03 -0.03 -0.03 -(0.82) -(0.73) -(0.80) -(0.72) Note: This table presents the results for regressions for the value relevance of the pension assumptions. The table presents the regression coefficient and immediately below is the corresponding t-statistic. * indicates significant at 99%, ** indicates significance at 95%, AND *** indicates significance at 90%. The dependent variable is presented at the head of each column and the independent variables are presented in the far left column. The discount rate and Equity return are taken from the FRS-17 disclosures. Discount rate spread is the difference between the discount rate and wage growth. Equity return spread is the difference between the expected return on equity and the discount rate. Firm assets are the total assets of the firm, firm debts is the total debts of the firm and market value of equity (dependant variable) is the balance sheet market value of equity all scaled by the number of common shares outstanding at the balance sheet date. Table 5 presents value relevance regressions of the disclosed assumptions used for discounting the pension liability and the expected return on scheme assets. Following a similar method to Weidman & Weir (2004) we regress the market value of equity against the different pension assumptions applied. itit eFirmDebtsFirmAssetsAssumptioneofEquityMarketValu  321  (2) where the market value of equity is balance sheet market value scaled by common equity outstanding, assumption is the pension assumption from the annual reports, Accounting and Management Information Systems Vol. 12, No. 2 208 firm assets and firm debts are the total assets and debts of the firm at the balance sheet date, scaled by common shares outstanding. From Table 5 we can see that individually neither the discount rate nor the expected return on equity has a significant effect on the market value of the firm. However, when we consider our spread variables they have a weak association with the value of the firm. The discount rate has a negative relationship with the market value of the firm’s equity. Firms that apply higher discount rates and lower wage growth i.e. understate the magnitude of the pension liability receive a lower market value. This is consistent management manipulation being associated with higher interest charges that reduces the profitability of the firm. Conversely we find that those firms who apply the highest equity return spread have higher market values. This is again consistent with expectations as higher spreads are associated with lower interest costs from one years unwinding of the pension liability and deriving higher levels of income from pension assets. Our final set of tests considers the value relevance of the amounts that are disclosed in the annual report, again following Weidman & Weir (2004) where, itit eControlsSolvencysLiabilitieAssetseofEquityMarketValu  321  (3) The market value of equity is balance sheet market value scaled by common equity outstanding, assets and liabilities are the pension assets and liabilities scaled by common shares outstanding. We also include firm assets and firm debt scaled by common shares outstanding as controls. From Table 6 it can be seen that pension assets and liabilities, solvency levels and asset composition all have some value relevance. This is itself an important factor in examining managerial discretion and the application of fair value accounting. Managers utilize the discretion that fair value affords as it has a tangible effect on the equity value of the firm. From columns 3 and 4 of Table 6 we can see that pension assets have a positive effect on market value while liabilities have no effect. However, from columns 5 and 6 when the solvency level in the scheme is controlled for, this relationship becomes insignificant but solvency becomes significant. Consistent with Barth et al. (1993) and the corporate finance view of pensions, solvency is positively related to the market value of equity, and so the pension assets and liabilities of the firm are perceived as belonging to the firm. This is contrary to the findings of Weidman & Weir (2004) who find evidence of the labour economics view of pension assets and liabilities i.e. surplus assets belong to scheme members. Finally in columns 7 and 8 we examine asset composition. Bergstresser et al. (2006) showed that equity affords management the greatest scope for generating financial income for the profit and loss. If this is the case, then higher equity Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 209 allocations should be associated with higher equity market valuations. Consistent with this we find a significant and positive relationship between the percentage of pension assets invested in equity and the market value of the firm. Interestingly when we control for the solvency level of the scheme we find that the percentage of equity increases in significance and pension liabilities become significantly negative. This suggests that asset allocation and liabilities have a greater bearing on the market value of the firm as opposed to scheme solvency. Table 6. Value Relevance Regression Analysis of Pension Assumptions Market Value of Equity/Ords Intercept 3.74 3.75 3.62 1.89 1.86 2.34 1.88 2.53 (25.84)* (26.09)* (23.82)* (2.61)** (2.56)** (2.94)* (2.59)** (3.51)* Pension Assets/Ords 0.03 - 0.47 0.03 - 0.30 - - (1.55) (2.57)** (1.46) (1.48) Pension Equity/Ords - - - - - - 0.09 1.21 (2.28)* (5.66)* Pension Liabilities/Ords - 0.02 -0.30 - 0.02 -0.19 - -0.36 (1.31) (-2.44)** (1.34) (-1.36) (-5.34)* Solvency - - - 2.27 2.33 1.62 2.25 1.04 (2.60) (4.29)* (4.29)* (2.57)** (1.17) Firm Assets/Ords 0.06 0.06 0.06 0.06 0.06 0.06 0.06 0.06 (4.81)* (4.82)* (4.59)* (4.30)* (4.29)* (4.29)* (4.29)* (4.15)* Firm Debts/Ords -0.03 -0.02 -0.02 -0.02 -0.02 -0.02 -0.02 -0.02 (-0.81) (-0.82) (-0.66) (-0.43) (-0.43) (-0.44) (0.44) (-0.46) Note: This table presents the results for regressions for the value relevance of the pension assumptions. The table presents the regression coefficient and immediately below is the corresponding t-statistic. * indicates significant at 99%, ** indicates significance at 95%. The dependent variable is presented at the head of each column and the independent variables are presented in the far left column. Pension assets are the total assets in the pension portfolio. Pension equity is equities held in the pension assets. Pension liabilities are the gross pension liability owed by the firm. Firm assets are the total assets of the firm, firm debts is the total debt of the firm and market value of equity (dependant variable) is the balance sheet market value of equity all scaled by the number of common shares outstanding at the balance sheet date. Funding is the ratio of pension assets to pension liabilities. Accounting and Management Information Systems Vol. 12, No. 2 210 CONCLUSIONS This paper considers fair value accounting for pensions, managerial discretion, and value relevance. In particular we consider whether the adoption of fair value accounting for pensions addresses the concerns that have been voiced about past methods of pension accounting. Using a sample of fair value pension disclosures from the UK fair value pension accounting standard FRS-17, we analyze the way in which fair value has been applied by firms in practice. Our results are as follows. First, they show that the variation in the underlying pension valuation assumptions across firms is significant. This result brings into question the suitability of managerial discretion under fair value as a method of accounting for pensions as financial accounts will remain opaque if management are not consistently reporting across firms. Second, we find that the variation in the assumptions that are presented in the financial accounts are not explained by the use of different audit or actuarial firms. Consequently, these external bodies do not hold ‘house views’ on what constitutes a reasonable or prudent assumption. This therefore also raises questions as to the efficacy of fair value accounting for pensions as auditors and actuaries do not seem to be applying the standard in a consistent way. Third, we consider the determinants of managerial discretion. Our results show that management adopt different assumptions in response to the solvency of the pension scheme. Where scheme solvency is high, management choose to apply higher discount rates in estimating the pension liability ― that is, systematically understating the liability. Conversely, where schemes have large deficits, firm managers choose to derive a larger amount of financial income from the assets in the pension scheme. Here management apply the least prudent return assumptions and thereby increasing the profit of the firm. Finally, we analyze the value relevance of both the assumptions and accounting amounts that are disclosed in the annual report. We find that these amounts are value relevant. This is important as the variations that have been observed in the assumptions and valuation process in the application of the standard allow for managers to undertake window dressing of the pension scheme, and that this is successful as it has a tangible share price impact. Bringing all of these results together we find that the case for adopting fair value accounting in pension accounting is still unresolved. Advocates of fair value accounting believe that it will make financial accounts more representative of the true economic position of the firm. However, our results show that where management have discretion over how ex ante pension standards applied, financial accounts remain opaque and suitable caveats are required. Assuming the worst: the shifting sands of pension accounting Vol. 12, No. 2 211 REFERENCES Amir, E. & Benartzi, S. (1998) “The Expected Rate of Return on Pension Assets and Asset Allocation as Predictors of Portfolio Performance”, The Accounting Review, vol. 73, no. 3: 335-352 ASB (2000) Financial Reporting Standard 17: Retirement Benefits, London: Accounting Standards Board Bergstresser, D., Desai, M. & Rauh, J. (2006) “Earnings Manipulation, Pension Assumptions and Managerial Investment Decisions”, Quarterly Journal of Economics, vol. 121, no. 1: 157-195 Black, F. (1980) “The Tax Consequences of Long-run Pension Policy”, Financial Analysts Journal, 36: 21-28 Bodie, Z. (1990) “The ABO, the PBO and Pension Investment Policy”, Financial Analysts Journal, 46: 27-34 Cocco, J. & Volpin, P. (2007) “Corporate Governance of Pension Plans: The UK Evidence”, Financial Analysts Journal, vol. 63, no. 1: 70-83 Cooper, S., Sutherland, G., Deng, Z. & Suozzo, P. (2001) “Pension Obligations: Accounting and Valuation Implications”, UBS Equity Research, London: UBS Coronado, J. & Sharpe, S. (2003) “Did Pension Plan Accounting Contribute to a Stock Market Bubble?”, Brookings Papers on Economic Activity, no. 1: 323-371 Gold, J. (2005) “Accounting/Actuarial Bias Enables Equity Investment by Defined Benefit Pension Plans”, North American Actuarial Journal, vol. 9, no. 3: 1-21 Grant, C.T., Grant, G.H. & Ortega, W.R. (2007) “FASB’s Quick Fix for Pension Accounting is Only First Step”, Financial Analyst Journal; March/April vol. 63 no. 2: 21-35 Hann, R., Heflin, F. & Subramanayam, K.R. (2007) “Fair-value Pension Accounting”, Journal of Accounting and Economics, vol. 44, no. 3: 328-358 Healy, P. & Whalen, J. (1999) “A Review of the Earnings Management Literature and its Implications for Standard Setting”, Accounting Horizons vol. 13, no. 4: 365-383 Lane Clark & Peacock (2005) Accounting for Pensions UK and Europe Annual Survey 2005, London: Lane Clark & Peacock Actuaries and Consultants Sloan, R. (1996) “Do Stock Prices Fully Reflect Information in Accruals and Cash Flows About Future Earnings”, The Accounting Review vol. 71, no. 3: 289-315 Tepper, I. (1981) “Taxation and Corporate Pension Policy”, Journal of Finance vol. 36, no. 1: 1-13 1 SEC (2005). 2 SSAP 24 paragraph 18. 3 FRS-17 paragraph 20. Accounting and Management Information Systems Vol. 12, No. 2 212 4 FRS-17 paragraphs 32-33. 5 Although we only focus on wage growth, the increase of the pension liability will be a function of pension growth, inflation wage growth and changes to the level of benefits provided. The assumed increase in wages however, is by far the biggest driver of increases in the liability owed by the firm. 6 In 1993, the SEC’s Chief Accountant ruled that this was the appropriate discount rate for calculating the present value of a firm’s pension liability. 7 This is more likely to be the case where equity is the dominant asset in the pension portfolio. 8 It is also possible that firms will apply a comparable discount rate in their assumptions and then select a much higher expected return on equity to ensure a large spread. 9 U.S. pension schemes are of a similar composition to UK schemes and on average hold large amounts of equity. 10 This, in itself, is large and British Telecom, for example, illustrates the magnitude of the problem faced by some firms. Their ratio of pension liability to total assets was 111% with a 70% funding ratio. The size of this scheme liability was also substantial at approximately £30bn in 2001. 11 This information comes from private correspondence with actuaries and fund managers about the composition of assets in defined benefit pension schemes of large UK corporations. 12 Data is taken from the Office for National Statistics. 13 The 50 year historical average is taken from Barclays Capital Equity Gilt Study 2007. 14 We have only presented a single regression for each assumption. The analysis was carried out over a number of different specifications. The analysis of the size of the pension liability relative to the firm by market value and total assets was insignificant as was the gross surplus/deficit scaled by market value. 15 Results not reported but are available on request.