Value relevance; A comparative analyzes of the revised
pension accounting standards in the Netherlands
Student name: Zoureena Vrede
Thesis Title: Value relevance; A comparative analyzes of the revised pension accounting standards in the Netherlands Student Number: 10326065
Date final version: June 22nd, 2015 Word Count: 11151
MSc Accountancy & Control, variant Accountancy Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam Supervisor: dr. A. Sikalidis
Statement of Originality This document is written by student Zoureena Vrede who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Employee benefits (IAS19), enabled companies to recognize their actuarial gains and losses through, either profit and loss, equity or the corridor method. The main criticism of this standard was the deferred recognition of actuarial gains and losses (corridor method). Starting January 2013, European companies were no longer allowed to defer actuarial gains and losses. These had to be recognized immediately in equity. The amendment to IAS19 lead to billion Euros of actuarial losses, which companies had to recognize immediately, where early application was permitted. The main reason for this amendment was to switch to a fair-value accounting approach, where the balance sheet portrays the true fundamental value of the firm and to move to more transparent and understandable financial statements.
Recent empirical researches of Coronado and Sharpe (2003), Picconi (2006) Franzoni and Marín (2006) show evidence that stock markets are not fully efficient with regard to pension accounting information. The reason for this is that mostly analysts and investors have difficulties in interpreting pension disclosures. Another reason is because firms with DB pensions are often misvalued by the market. However, on the other hand Hann et al. (2007), Kiosse et al. (2007) and Werner (2011), concluded that stock markets are fully efficient with regard to pension accounting information.
Based on these conclusions, one can state that researchers on value relevance of pension accounting arrive at conflicting and inconclusive results. However most research was conducted on US company data, which prepare their financial statements on the basis of US GAAP. Therefore I investigated to what extend share prices reflect corporate pension accounting information under IAS19 versus IAS19R in the Netherlands. The main research question that is answered is: “Do equity values of firms with defined benefit plans reflect the true economic value of pension assets and liabilities?” The research sample was over the period 2010 to 2014. The companies used in the analyses are listed on the AEX and AMX. In order to investigate the main research question the Ohslon (1995) model is used and two hypotheses were tested.
The results led to the conclusion that equity values do not reflect true economic value of pension assets and liabilities. Even though the sample size is rather small, the results are in line with Vuijk (2010) and Nandram (2011), whom also use EU company data.
List of Abbreviations
AEX: Amsterdam Exchange Index AMX: Amsterdam Midkap Index BVE: Book Value of Equity Coef: Coefficient
COREEARN: Core Earnings DB: Defined Benefit
DBO: Defined Benefit Obligation DC: Defined Contribution
df: Degrees of freedom
EU: European Union
FS: Funded Status
GAAP: General Accepted Accounting Standards Board IAS: International Accounting Standard
IASB: International Accounting Standards Board IFRS: International Financial Reporting Standards LIAB: Liability
MAX: Maximum
MIN: Minimum
MVE: Market value of equity
NIBX: Net Income Before Extraordinary Items NPA: Net pension assets
OBS: Observations
OCI: Other Comprehensive Income
P: P-Value
PENSEARN: Pension Earnings Std.Dev: Standard Deviation Std.Err: Standard Error
T: T-Statistic
List of Equations
Equation 1a The Ohslon model (Wiedman and Wier, 2004) p. 20
Equation 1b The Ohlson model, extended version (Wiedman and Wier, 2004) p. 28
Equation 2 The Ohslon model to test value relevance (Coronado et al., 2008) p. 21
List of Exhibits
Exhibit 1 Structure of Employee Benefits p. 11
List of Tables
Table 1 Final sample selection – Hypothesis 1 p. 22
Table 2 Final sample selection – Hypothesis 2 p. 23
Table 3 Descriptive statistics hypothesis 1 (corridor method) p. 24
Table 4 Descriptive statistics hypothesis 1 (equity method) p. 24
Table 5 Descriptive statistics hypothesis 2 (corridor method) p. 25
Table 6 Descriptive statistics hypothesis 2 (equity method) p. 25
Table 7 Results hypothesis 1 (corridor method) p. 26
Table 8 Results hypothesis 1 (equity method) p. 27
Table 9 Additional analysis results hypothesis 1 (corridor method) p. 28
Table 10 Additional analysis results hypothesis 1 (equity method) p. 29
Table 11 Result hypothesis 2 (corridor method) p. 31
Abstract ... 3 List of Abbreviations ... 4 List of Equations ... 5 List of Exhibits ... 5 List of Tables ... 5 1 Introduction ... 7 1.1 Background ... 8 1.2 Research question ... 8 1.3 Motivation ... 9
2 Background to pension accounting ... 10
2.1 Scope IAS 19 ... 10
2.1.1 Post-employment benefits ... 11
2.1.1.1 Defined contribution plan ... 12
2.1.1.2 Defined benefit plan ... 12
2.1.2 Recognition and measurement of defined benefit plan ... 12
2.2 IAS 19R ... 14
2.2.1 Scope IAS19R ... 14
2.2.2 Changes IAS19R ... 14
2.2.3 Choices of accounting methods for recognizing actuarial gains and losses under IAS 19R ... 15
3 Literature review on value relevance of pension accounting ... 15
3.1 Value relevance of pension accounting ... 15
3.1.1 Proponents of fair-value pension accounting ... 16
3.1.2 Opponents of fair-value pension accounting ... 17
3.2 Conclusion ... 18
4 Research design ... 18
4.1 Hypotheses development ... 19
4.2 Model development ... 19
4.3 Data and sample selection ... 21
5 Empirical results ... 23
5.1 Descriptive statistics ... 23
5.1.1 Descriptive statistics hypothesis 1 ... 24
5.1.2 Descriptive statistics hypothesis 2 ... 25
5.2 Test results and finding hypothesis 1 ... 25
5.2.1 Test results and findings for companies, which used the Corridor method .... 26
5.2.2 Test results and findings for companies, which used the Equity method ... 26
5.2.3 Additional analysis ... 27
5.2.3.1 Results additional analysis results hypothesis 1 (corridor method) ... 28
5.2.3.2 Results additional analysis results hypothesis 1 (equity method) ... 29
5.2.4 Conclusion ... 30
5.3 Test results and finding hypothesis 2 ... 30
5.3.1 Test results and findings for companies, which used the Corridor method .... 30
5.3.2 Test results and findings for companies, which used the Equity method ... 31
5.3.3 Conclusion ... 32
6 Conclusion ... 33
1 Introduction
The European Union agreed that for years beginning on or after January 1, 2005 the application of International Financial Reporting Standards (IFRS) would be mandatory for the consolidated financial statements of European (EU) listed companies. According to Jermakowicz and Gornik-Tomaszewski (2006) the International Accounting Standards (IAS) regulation is expected to reach global accounting convergence, help eliminate barriers to cross-border trading in securities, increase market efficiency and reduce the cost of raising capital for EU listed companies.
The general purpose of financial reporting is to provide financial information about the reporting entity that is useful in the decision making for equity investors, lenders and other creditors (Vehmanen, 2009). “For financial information to be useful, it must be relevant and faithfully represent what it purports to represent. The usefulness of financial information is enhanced if it is comparable, verifiable, timely and understandable”
(Conceptual Framework, 2010: A33). “Relevant financial information is capable of making a difference in the decisions made by users. Information may be capable of making a difference in a decision even if some users choose not to take advantage of it or are already aware of it from other sources” (Conceptual Framework, 2010: A33). “To be useful, financial information must not only represent relevant phenomena, but it must also faithfully represent the phenomena that it purports to represent. To be a perfectly faithful representation, a depiction would have three characteristics. It would be complete, neutral and free from error” (Conceptual Framework, 2010: A34).
According to Glaum (2009) one of the most challenging areas in IFRS is pension accounting. Under the previous pension accounting standard (IAS 19), pension assets and liabilities were amortized over the expected employees remaining service, leading to accumulated pension expenses. This choice of accounting led companies to avoid reporting losses in current and future financial statements, and being a burden on corporate income. This also led to managers’ the ability to influence the financial performance and treat the actuarial gains and losses in their beneficiary. Based on the mentioned reasons above, the standard was heavily criticized. As a result the FASB revised IAS19. Under the current pension accounting standard (IAS19R) the balance sheet reflects fair value of net pension assets and all changes in the fair value of net pension assets flow through income (Hann et al, 2007). The most important amendment to IAS19R is the elimination of deferred recognition of actuarial gains and losses
(corridor method). According to Sun (2011) the impact of these revisions (IAS19R) could range from material to immaterial.
1.1 Background
Severinson (2008) states that the recent changes to pension accounting standards tend to have greater transparency and comparability between companies and countries, but they also lead to greater volatility in a company’s balance sheet and earnings. Research by Fasshauer et al. (2008) showed that for companies with accumulated unrecognized actuarial gains and losses the effect on the P&L would be immaterial. However, they state that one should keep in mind that these companies are first-time IFRS adopters and thus may have minimal unrecognized actuarial gains and losses in 2005. Based on this conclusion one might expect that companies will accumulate gains and losses over the next years using the corridor method. Full recognition of actuarial gains or losses in fluctuating markets can cause substantial volatility in equity (Fasshaeur et al., 2008). Reporting under the proposed new amendments of IAS19, Hann et al. (2007), finds evidence that fair-value accounting would lead to a greater volatility in net income, reducing its persistence. They conclude that due to its lower persistence, fair-value accounting is less value relevant. Morais (2008) concluded that the corridor method (IAS19) provided less value relevant financial information than the equity or the profit and loss method for her sample of 91 company year observations. Based on the results of Werner (2011), pension information recognized under a fair‐value‐based accounting model is no more or less value relevant than pension information recognized under the SFAS 87 model. Results of Vuijk, S. (2010) show empirical evidence indicating that there is no significant difference in providing value relevant financial information between the equity and corridor method, but both providing slightly more value relevant financial information than the profit and loss method.
1.2 Research question
According to these contradictory results, I shall examine whether pension accounting information under IAS 19R, recognized under a fair-value approach, is more highly associated with firm equity values than pension accounting information reported under IAS19. The main research question is:
“Do equity values of firms with defined benefit plans reflect the true economic value of pension assets and liabilities?”
1.3 Motivation
A significant amount of previous literature have pointed out that financial statements lost their value relevance because of a shift from traditional capital intensive economy into a service-oriented economy (Dontoh et al., 2007).
Due to the new amendments made to IAS19, companies were compulsory to
derecognize their assets and liabilities, which were realized through pension expense and therefore led to a gain or loss for that specific year (Hartwell, 2012).
It is argued by the European commission that IAS19R meets the criteria of understandability, relevance, reliability and comparability required of the financial information needed for making economic decisions and assessing the stewardship of management.
According to Schoonderbeek (2010) the elimination of the use of the corridor method has the most impact, because it has a dampening effect on the result of the financial statements. The abolishment of this means that the results in the financial statements of many (listed) companies in the Netherlands are of the dependent pension deficits or pension surpluses in a defined benefit plan. According to a survey held by KPGM approximately 70% of the companies in the Netherlands applied the corridor method. They concluded that the elimination of the corridor method resulted in a total obligation around 22.3 billion euros on the balance sheet of AEX listed companies. This is due to the fact that the total actuarial gains at 31 December 2012 amounted to around € 1.3 billion and the total actuarial losses around 23.6 billion euros. Based on this
assumption the focus of this thesis lies in the changes in recognition of actuarial gains and losses under IAS19R.
While there have been an extensive amount of value relevance studies, most of these studies have focused on the economic consequences of pension accounting in the US, which prepare their financial statements under US GAAP. Meanwhile there has been a little amount of research in the Netherlands, which report under IFRS. In 2006, Laning stated in his article that there is little academic research, which is based on the Dutch situation. The main difference between US GAAP and IFRS is that US GAAP is “rules based” and IFRS is “principles based”. This may lead to different results compared to studies, which were mostly conducted in the US. The reason for the particular interest in pension accounting value relevance is that for those companies that sponsor defined benefit (DB) pension schemes for their employees there can be major amounts relating to pensions that have a very significant impact on the results in the published financial statements.
The first contribution of my study is that I directly address the currently improved standard, which may lead standard setters to critically think whether this change leads to more value relevance. The second contribution, to my knowledge, is that this is one of the few studies to examine value-relevance of pension accounting in the Netherlands.
2 Background to pension accounting
IAS 19, Employee benefits, applied to all employee benefits except those to which IFRS 2 Share-based payment applies. Employee benefits arise from formal agreements, which are often referred to as workplace agreements, between an entity and its individual employees. Accounting for employee benefits is complicated because some benefits may be provided many years after employees have provided services. Employee benefits are defined as: “all forms of consideration given by an entity in exchange for service rendered by employees
or for the termination of employment.” (IAS 19.8). According to Carpenter & Mahoney (2004)
pension obligations were not recorded as they accrued, but were instead recorded when paid (i.e. corridor method). This approach constituted a major violation of the tenets of accrual basis accounting and was considered conceptually inappropriate. The economic and political consequences were far too significant and standard setters saw a need to depart from this approach to a more appropriate full accrual of pension costs (Carpenter & Mahoney, 2004). Based on this assumption the focus of this thesis lies in the changes in recognition of actuarial gains and losses under IAS19R.
In this chapter an overview will be given of IAS19. First, there will be looked at the two types of post-employment benefits. Second, the three choices for recognizing actuarial gains and losses will be highlighted. Finally, the new standard (IAS19R) will be discussed.
2.1 Scope IAS 19
IAS 19 regulated the accounting for employee benefits. Employee benefits comprise all forms of consideration given by an entity in exchange for service rendered by employees (IAS 19.7). IAS19 also describes how listed companies should assume their pension liabilities on the balance sheet. Under IAS19 the following four categories of employee benefits are identified (IAS 19.4):
1) Short-term employee benefits include wages, salaries, bonuses and profit-sharing arrangements;
2) Post-employment benefits such as pension plans & employee retirement plans. Post- employments benefits can be divided into two types: defined benefit plans and defined contribution plans;
3) Other long-term employee benefits such as long-service or sabbatical leave which do not fall within twelve months or more after the end of the period in which the employee render the related service and
4) Termination benefits.
Figure 1: Structure of Employee benefits
For the purpose of this thesis only the accounting of pensions, which forms part of the post-employment benefits (category 2) is described.
2.1.1 Post-employment benefits
Post-employment benefit plans are informal or formal arrangements under which an entity provides post-employment benefits to one or more employees, e.g. retirement benefits (pensions or lump sun payments), life insurance and medical care. Post-employment benefit plans are payable after the completion of Post-employment and can be qualified either as defined contribution plan (DC) or defined benefit plan (DB). The qualification for a post-employment benefit plan depends on the economic substance of the plan as derived from its principal terms and conditions. Both have different
recognition and measurement rules. The main difference between these two types of plans is that with DC, the risks remain with the employee and DB, the risks remain with the employer. EMPLOYEE BENEFITS Short-term benefits Post-employment benefits EMPLOYEE Other long-term benefits Termination benefits Defined contribution plans Defined benefit plans
2.1.1.1 Defined contribution plan
A pension plan qualifies as a defined contribution when the following criteria are met (IAS19.25):
The entity’s legal or constructive obligation is limited. Fixed contributions are paid by the entity whereas it is deposited into a fund. If the fund does not hold sufficient assets to pay all employee benefits related to employee services in current and prior periods, the entity has no legal or constructive obligation to pay further contributions.
In consequence, actuarial risk (that benefits will be less than expected) and investment risk (that assets invested will be insufficient to meet expected benefits) fall on the employee.
2.1.1.2 Defined benefit plan
A pension plan qualifies as a defined benefit when the following criteria are met (IAS19.27):
The entity’s legal or constructive obligation is to provide the agreed benefits to current and former employees;
In consequence, actuarial risk and investment risk fall on the entity.
The focus of this thesis will be on DB plans. Since this is the most common used scheme in the Netherlands.
2.1.2 Recognition and measurement of defined benefit plan
The accounting for defined benefit plans is considered to complex. This is mainly because actuarial assumptions are required to measure the pension obligation and the pension expense, which leads to actuarial gains and losses. Actuarial assumptions comprise assumptions relating to demographic- and financial parameters. Moreover, companies with defined benefit pension plans were able to choose the following reporting options for the recognition of actuarial gains and losses (IAS 19.54), namely:
1. Profit and loss method (IAS 19.93 and 19.95): all actuarial gains and losses are immediately recognized in the period in which they occur through profit and loss;
2. Equity method (IAS 19.93A-19.93D): if all actuarial gains and losses are recognized in the period in which they occur, they may be recognized in other comprehensive income;
3. Corridor method (IAS 19.92 – 19.93): in which a portion of the actuarial gains and losses are recognized as income or expense if the net cumulative
unrecognized actuarial gains and losses at the end of the previous reporting period exceeded the greater of: (1) 10% of the present value of the defined benefit obligation at that date (before deducting plan assets); and (2) 10% of the fair value of any plan assets.
Ad 1. Profit and loss method for recognizing actuarial gains and losses
Paragraph 93 of IAS 19 states that all actuarial gains and losses under the profit or loss method, are recognized immediately in the profit and loss statement in the period in which they occur. However, IASB members concluded that immediate recognition of actuarial gains and losses can lead to fluctuations in the balance sheet and income
statement. These fluctuations may result in an unfaithful representation of changes in the defined benefit asset or liability, but on the other hand they may simply reflect an
unavoidable inability to predict accurately the future events that are anticipated in making period-to-period measurements (IAS 19. BC48C).
Ad 2. Equity method for recognizing actuarial gains and losses
The equity method was introduced by the IASB as an option for recognizing actuarial gains and losses. Leading to actuarial gains and losses being recognized immediately outside the profit and loss statement in the period in which they occur. Actuarial gains and losses were now presented as OCI in the ‘Statement of Recognized Income and Expense’ ('SoRIE') (IAS 19.93). Immediate recognition intends to provide more transparency for users than deferred recognition (IAS 19. BC48G).
Ad 3. Corridor method for recognizing actuarial gains and losses
Under this method, the company recognizes a portion (only the accumulated actuarial gains and losses that exceed a predetermined level) of its actuarial gains and losses in profit and loss. So basically there is no recognition. The difference between pension obligations and pension assets are offset by unrecognized actuarial gains and losses (Jullens, 2011). Therefore the following limits were set, calculated and applied separately if the portion of net cumulative actuarial gains and losses exceeded the greater of 10% of the present value of the defined benefit obligation or 10% of the fair value of any plan assets.
Based on research of Morais (2008), the corridor method was the most used method in European companies. This is mainly because, in short term, this method led
to less volatility in the balance sheet and income statement. However, the IASB’s main arguments about the corridor method were that it causes high volatility in long term on actuarial gains and losses in the balance sheet and income statement (IAS 19. BC39).
2.2 IAS 19R
June 2011 the IASB revised IAS19 “Employee Benefits”. The amendments in pension standard (IAS19R) were endorsed by the European Union on 5th June 2012 and effective as from 1st January 2013. The objective of IAS19R (employee benefits) is to prescribe the accounting and disclosure for employee benefits, requiring an entity to recognize a liability where an employee has provided service and an expense when the entity consumes the economic benefits of employee service.
2.2.1 Scope IAS19R
The scope of the standard remains unchanged. It is applied by all entities in accounting for all employee benefits other than those to which IFRS 2 Share-based payments applies (IAS 19R.2). Employee benefits continue to be classified into four separate categories. However, the boundaries between these categories have changed (IAS 19R.5).
2.2.2 Changes IAS19R
In June 2011 the IASB proposed amendments to be made to IAS19. These amendments included changes to the recognition, presentation and disclosure of defined benefits plans by:
Eliminating the option to defer the recognition of actuarial gains and losses (i.e., the 'corridor method') in order to improve comparability and faithfulness presentation.
Plan cost should now be disaggregated into three components (service costs, finance costs and re-measurement). Service and finance costs should be
recognized in profit and loss. Whereas, re-measurement should be recognized in other comprehensive income.
Enhancing the disclosure requirements for defined benefit plans, providing better information about the characteristics of defined benefit plans and the risks that entities are exposed to through participation in those plans.
The accounting for post-employment benefits and, in particular, defined benefits plan was most significantly impacted by IAS 19R.
2.2.3 Choices of accounting methods for recognizing actuarial gains and losses under IAS 19R
Under IAS19R accounting for actuarial gains and losses is now recognized in OCI. Resulting in a permanent bypass profit or loss for entities that recognized actuarial gains and losses, leading to a more volatile financial position. It is expected that IFRS users and analysts will place greater scrutiny or importance on amounts recognized in OCI, actuarial estimates and the disclosure of historical experience gains or losses. The IASB explains in the Basis for Conclusions to the amended standard that it believes that immediate recognition provides more relevant information to users of financial
statements and provides a more faithful representation of the financial effect of defined benefit plans. It also improves comparability by eliminating the options allowed under the previous IAS19.
3 Literature review on value relevance of pension accounting
Deegan & Unerman (2006) argued that accounting regulations have real social and economic consequences for many organizations and people. De Jong et al. (2006) argue that these accounting standards affect a company’s financial statements.As mentioned by Glaum (2009) pension accounting is one of the most complex standards. Therefore, it is obvious that researchers of value relevance pension accounting information came to conflicting and inconclusive results (Glaum, 2009, p. 306). Some argue that the fair value approach will lead to an increase in income volatility and will be more susceptible to manipulation. On the other hand some argue that the fair value model improves the transparency, comparability and even improves the relevance of asset and liability measurement. There are an extensive amount of value relevance studies conflicting with each other even though they undertake research on very similar topics or use similar or even the same data sources (Kirkpatrick, 2012)).
In paragraph 3.1 the relevance of value relevance of pension accounting will be highlighted. Followed by a summary of researches conducted on the value relevance of pension accounting and the conflicting and inconclusive results.
3.1 Value relevance of pension accounting
According to Kirkpatrick (2012), value relevance refers to the relationship between published accounting data and the market value of the reporting business entity. The IASB’s intention to shift from IAS 19 to IAS19R was mainly to shift to a balance sheet
approach, where the balance sheet should portray the true fundamental value of the firm. Moreover, assets and liabilities are now measured based on fair value.
In this paragraph the different views of researchers will be discussed.
3.1.1 Proponents of fair-value pension accounting
The use of a balance sheet model enables you to split a company’s total assets into pension assets and non-pension assets, while liabilities are split into pension liabilities and non-pension liabilities. Landsman (1986) was one of the first who used a balance sheet model to examine the value relevance of pension assets and liabilities. While using a sample of US companies in the period 1979-1981, he found that pension assets and liabilities are value relevant and treated similarly to corporate assets and liabilities.
The risk analysis approach was introduced by Dhaliwal (1986). He investigated the impact of unfunded vested pension obligations on market perceived risk. The sample period consisted of 55 firms in the period 1976-1979. Their findings implicate that the effect of unfunded vested pension liabilities is not significant or statistically different from that of debt or other liabilities. Their research was heavily criticized because the sample size was rather small and that there was a weak association between the accounting numbers and market values revealed by an R2 in the range of 0.20 to 0.27.
A more recent study by Jin et al. (2006) continued the risk analysis approach. They conducted their study on US firm data for the period 1993-1998. Their findings implicate that equity betas appear accurately to reflect the betas of their pension assets and liabilities, “despite the practical difficulties of deciphering corporate pension accounts” (Jin et al., 2006, p.22).
Another research by Hann et al. (2007), investigated whether the fair value model is more relevant than the ‘smoothing model’. They do this by combining value relevance using both income statement and balance sheet information over the period 1991-2002. The results they find for the smoothing model lead to an adjusted R2 of 0,573 compared to an adjusted R2 of 0,551 using the fair value model. Surprisingly, they concluded that the difference is significant. However, if we look at the adjusted R2,it would be more appropriate to conclude that the fair value model is neither more nor less, value relevant than the smoothing model.
Werner (2011) also investigated whether pension accounting information under the ‘smoothing model’ is more or less value relevant than information under the fair value model. Their sample consisted of Fortune 200 firms, for 1998-2005. Under the ‘smoothing model’ they found an adjusted R2 of 0,343 and under the fair value model an
adjusted R2 of 0.342. These results are largely consistent with the results of the earlier study by Hann et al (2007).
Research by Kiosse et al. (2007) compared the value relevance of pension
accounting information prepared on fair value and ‘smoothing basis’. They used a sample of US companies over the period 1998-2005. Under the ‘smoothing model’ they found an adjusted R2 of 0.5118 and under the fair value model an adjusted R2 of 0.5517. Implying that pension costs reported under the ‘smoothing model’ are more closely correlated with stock market values than fair value pension cost measures.
Morais (2008) conducted a value relevance study for the three different methods of accounting for actuarial gains and losses. Her sample was based on 91 European companies in the period 2005-2007. The results suggest that the recognition of all actuarial gains and losses provides more value relevant information than the deferred recognition of actuarial gains and losses. She found that the equity method provides more value relevant than the profit or loss method or the corridor method.
Summarizing the above mentioned studies provide strong evidence that stock markets are fully efficient with regard to pension accounting information.
3.1.2 Opponents of fair-value pension accounting
Franzoni and Martin (2006) investigated the value relevance of pension accounting information. Their findings implicate that pension accounting information is value relevant with an adjusted R2 in the range of 0.52 to 0.96 (Franzoni and Martin 2006, p. 953). They concluded that there is a tendency for investors to overvalue firms that have large pension scheme exposures, in particular, where pension schemes are underfunded.
Coronado and Sharpe (2003) investigated the value relevance of recognized and disclosed pension accounting measures. They found that pension income and expenses are value relevant but balance sheet pension information is not. They argue that investors are often misled by the smoothing model of pension accruals, which leads to companies being overvalued (Coronado et al., 2008, p. 263). Another study by Coronado et al. (2008) investigated the value relevance of recognized and disclosed pension accounting measures. They used the Ohslon based model and found an adjusted R2 in the range of 0,922 to 0,934. Again they find that pension income and expenses are value relevant but balance sheet pension information is not. Both studies by Coronado are in conflict of earlier research of Barth et al. (1993).
Barth et al. (1993) investigated the value relevance of pension assets and liabilities and pension cost components using the Ohslon model. Their sample consisted of 300
US firms in the period 1987-1990. They found that pension assets and liabilities are value relevant, because they are significantly correlated with share price valuations. Whereas, pension cost components are not value relevant because they are not needed in explaining share prices (Barth et al., 2003, p.25).
Picconi (2006) investigated the value relevance of pension accounting information. The study is based on US companies in the period 1988-2001. He
concluded that there is evidence of value relevance of pension accounting information but it is also concluded that neither prices nor forecast fully reflect the quantifiable future earnings effects of changes in pension accounting information at the time it is published (Picconi, 2006, p. 951). These findings are supported by the study conducted by Franzoni and Marin (2006).
Summarizing, the mentioned studies above, they provide strong evidence that stock markets are not fully efficient with regard to pension accounting information. Especially investors and analysts encounter difficulties in processing information about entities’ pension positions that are only disclosed in the notes.
3.2 Conclusion
From the above literature review there can be concluded that empirical studies on the value relevance of pension accounting are sometimes based on data from different time periods and accounting regimes, but also data from exactly the same regimes. Still these studies arrive at conflicting and inconclusive results. Previous studies leave unanswered questions concerning the share price implications of pension cost disclosures that are crucial in today’s society. Such mixed results may be due to correlated omitted variables. Many of these prior studies choose either a balance sheet or an income statement focus, though both may be necessary to explain valuation assessments in the presence of measurement error such issues (Ohlson, 1995). While most value relevant studies are performed on US data which prepare their financial statements under US GAAP, there are not a lot of studies performed based on EU data which prepare their financial statements under IFRS. Based on the latter I shall preform a value relevance study on pension accounting information in the Netherlands.
4 Research design
Based on the previous chapter we can conclude that prior research on the value
relevance of pension accounting arrive at conflicting and inconclusive results. Therefore there is further research needed in order to arrive at a more in-depth conclusion. Most of these studies were conducted on US company data, and as mentioned by Glaum, there is
too little focus on European companies. This research is an attempt to extend value relevance literature in the Netherlands. In this chapter the hypotheses development will be set out, followed by the regression models which are useful for this research. Finally, the data and sample selection will be discussed.
4.1 Hypotheses development
In chapter 2.2 the new standard (IAS19R) is discussed. The main difference between IAS19 and IAS19R is the abolishment of the corridor method and profit and loss method. Under IAS19R accounting for actuarial gains and losses is now recognized in other comprehensive income. Therefore it is no longer allowed to defer actuarial gains and losses. The main aim of revising the standard was to enhance transparency and comparability. Many reasons can be cited for why pension risk might not be reflected in equity returns. Jin et al. (2006) argue that pension plan assets and liabilities are off-balance sheet and are often viewed as segregated from the rest of the firm. Moreover, pension accounting rules are complicated. Due to the fact that many studies with regard to pension accounting, arrived at inconclusive results I find it interesting to know whether investors fully understand and incorporate published pension accounting information in their decisions. The hypotheses that will be tested are as follows:
Hypothesis 1: The funded status of a company's pension plan is more strongly associated
with its stock price
Hypothesis 2: Companies using the equity method will lead to more value relevant
information compared to the corridor method
4.2 Model development
Various valuation models have been used over time to investigate value relevance. In general, value relevance is defined as the ability of financial information to explain stock prices and returns (Jianwei and Chunjiao, 2007). The Ohlson model is the most used model by many researchers (Glaum, 2009, p.283). To test my hypotheses, I will be also using the Ohlson model, because this model provides a direct link between accounting measures and firm value (Kothari, 2001; Beaver, 2002).
The first hypothesis was previously tested by Wiedman and Wier (2004) with data of Canadian firms in the period 2000-2001. So, by testing this hypothesis for Dutch
companies in the period 2010-2014, this study is extended. In order to assess the value relevance of pension assets and liabilities the first hypothesis is tested by a regression of market value of equity on pension information. The hypothesis will be investigated for companies that applied the corridor method in the period 2010-2012, and companies that applied the equity method in the period 2013-2014. The funded status (FS) will be calculated as pension assets minus pension liabilities, and will be treated as the variable of interest (Wiedman and Wier, 2004). However, for over- and under-funded plans to be treated equally, the coefficient on FS should be positive and significant. The following regression model will be used to test the first hypothesis.
Equation 1a – The Ohslon model (Wiedman and Wier, 2004)
MVEit= α _+ β1*ASSETit + β2*LIABit + β3*NIBXit + β4*FSit + εit
MVEit = the year end market value of firm i in period t - computed as the number of common shares outstanding times the share price of the company’s common stock at the end of the quarter following the year-end
ASSETit = total assets of firm i at the end of year t LIABit = total liabilities of firm i at the end of year t
NIBXit = net income before extraordinary items of firm i for year t
FSit= funding status of the DB pension plan of firm i at the end of year t = pension assets -/- pension liabilities
εit= error term
The second hypothesis states that companies that use the equity method will give a more appropriate reflection of corporate pension accounting information than companies that use the corridor method. The hypothesis states that R2
equity method > R2corridor method and will be investigated for companies that originally adopted the corridor method (2010-2012) and companies which were required to switch to the equity method (2013-2014). Therefore R2 is the measure of value relevance, it is the variance in Y that is explained by the model ad a percentage of the total variance in Y. It is expected that the R² for companies applying the equity method will be higher than for companies applying the corridor method, since this is also the intention of standard setters. By testing this hypothesis, we can evaluate whether it was necessary to abolish the corridor method. Since the main reason for revising the standard was to improve transparency, comparability and
faithfully represent what it purports to present. The following regression model will be used to test the second hypothesis.
Equation 2 – The Ohslon model (Coronado et al., 2008)
MVEit = α + β1*BVCit + β2*EMPit + β3*R&Dit +β4*Core EARNit + β5*NPAit + β6*Pension EARNit + εit
MVEit= the year end market value of firm i in period t = firms’ non-financial operating activities + value of unrelated financial activities
BVEit= Book Value of Equity of firm i at end of year t = Total BV -/- prepaid pension cost (net actuarial gains or losses in the case of corridor method)
Core EARNit= pre-tax profit -/- pension earnings NPAit = prepaid pension cost
Pension EARNit= net financing accruals = service cost + interest cost + other cost -/- expected return on plan assets
εit= error term
The reasons for the preference for the Opaque Model is that the qualitative research suggests that analysts use both balance sheet and income information and the
quantitative results using the Opaque Model are also stronger than for the Transparent Model which is consistent with Coronado and Sharpe (2003).
4.3 Data and sample selection
The samples will consist of firms that are listed on the Dutch stock market indexes Amsterdam Exchange Index (‘AEX’) and the Amsterdam Midkap Index (‘AMX’). The AEX is a weighted index covering 25 companies, which have the largest market capitalization and trading volume. The AMX is also a weighted index covering 25 companies based on size and trading volume criteria of the Euronext. In the analysis I will be comparing data from different periods, because IAS19R was effective since January 2013. For IAS19, I will be analyzing the data from 2010 up to and including 2012. And for IAS19R, I will be analyzing data from 2013 up to and including 2014. Only companies, which prepare their consolidated financial statements under IFRS will be selected.
The data are extracted from the DataStream database, however the data on the defined benefit plans which are not included in DataStream will be manually collected by the use of the concerning annual reports of the companies. The variables of interest are Common shareholders’ equity (WC03501), Net Income (WC01751), Common shares outstanding (WC05301), Total assets (WC02999), Total Liabilities (WC03351), Fair value of pension plan assets (WC18807), Projected benefit pension obligation (WC18809), Net pension asset/liability (WC18821), Unrealized net actuarial gains and losses under the corridor method (WC18819), Net income before preferred dividends (WC01551), Market Capitalization (WC08001). The variables manually collected from the annual reports are the Service cost, Interest cost, Other cost, Expected return on plan assets, Pre-tax profit and Recognized actuarial gains and losses. For the second hypothesis, only the companies, which originally adopted the corridor method were included in the sample selection. This is done because I wanted to see whether there is a significant change in the R2 when implementing the new standard.
For hypothesis 1 all variables which were used in the model are divided by the number of shares outstanding (WC05301) at year end. For hypothesis 2 all variables are divided by total assets. This is done in order to minimize scale effects.
Furthermore, firms with missing financial statement data, firms with no defined benefit pension plans and firms that did not have December as financial year end were excluded from the sample.
In order to make sure that the data from DataStream is retrieved properly, forty at random selected variables for a certain firm-year are manually reconciled to the relevant annual report and no significant expectations are noted. Table 1 provides a summary of the final sample observation for hypothesis 1 and table 2 provides a summary of the final sample observation for hypothesis 2.
Table 1: Final sample selection – Hypothesis 1
2010 2011 2012 2013 2014 Total
Initial sample observation 50 50 50 50 50 250
Reason for exclusion:
No December 31th financial year end 3 3 3 3 3 15
No defined benefit pension plan 11 11 11 11 11 55
For hypotheses 2, only companies that originally adopted the corridor method were chosen because I want to see whether the implementing the new standard really increases the value relevance for those companies.
Table 2: Final sample selection – Hypothesis 2
2010 2011 2012 2013 2014 Total
Initial sample observation 50 50 50 50 50 250
Reason for exclusion:
No initial application corridor method 11 11 11 11 11 55
No December 31th financial year end 3 3 3 3 3 15
No defined benefit pension plan 11 11 11 11 11 55
Final sample selection 25 25 25 25 25 125
5
Empirical results
In the previous chapter the hypothesis, methodology, data and sample selection were outlined. In this chapter the results of the empirical research will be discussed per hypothesis.
For the statistical tests of both hypothesis a significance level of 5% is used. This entails having a probability of 5% for a Type 1 error. A Type 1 error is (falsely) rejecting the null hypothesis that is actually true (Keller, 2002). Therefore, the probability values (p-values) in all the tables need to be lower than 0.05 in order meet the significance criteria of rejecting a true null hypothesis.
Furthermore, for each regression there will be looked at the adjusted R2, which is the proportion of variability in a data set that is accounted for by the statistical model. The adjusted R2 gives information about the goodness of fit of the model and is a statistical measure of how well the regression line approximates the real data. If the R2 is equal to one, then there is a perfect fit between the regression line and the data (Keller, 2002).
Before discussing the results for each hypothesis, an overview will be provided of the used samples in the first paragraph of this chapter. In the second paragraph the test results and findings will be outlined, followed by an overall conclusion.
5.1 Descriptive statistics
Before discussing the results and findings of every hypothesis it is important to describe the used samples. This will be done per hypothesis.
5.1.1 Descriptive statistics hypothesis 1
The initial sample observations for the companies that applied the corridor method consisted of 250 firm-years. However each hypothesis had its own sample of observations. The descriptive statistics of each hypothesis 1, corridor- and equity method, will be provided below.
Table 3 – Descriptive statistics hypothesis 1 (corridor method)
Variable Obs Mean Std.Dev. Min Max
MVE 75 25.4935 33.4260 2.1934 218.3193 ASSET 75 47.7326 61.7056 6.0137 336.9348 LIAB 75 29.4287 46.5605 1.5334 288.9023 NIBX 75 0.76501 3.7135 -20.1070 13.1503 FS 75 1.37027 7.4204 -24.5138 41.68204 Hypothesis 1:
The final sample observations for hypothesis 1 (corridor method), consists of 75 firm-years. Table 3 provides the minimum, maximum, mean, standard deviation, and pooled standard error for each variable. The descriptive statistics show that the range between minimum and maximum amount for each variable is large. Also, the standard deviation is very high, which is an indication that the data is spread out over a wide range of values. This could be explained by the fact that the companies included in the sample have different sizes in line with their quotes on the AEX and AMX index. Furthermore, the standard errors are given and these reflect the level of uncertainty in the mean.
Table 4 – Descriptive statistics hypothesis 1 (equity method)
Variable Obs Mean Std.Dev. Min Max
MVE 105 33.7692 36.2126 2.343 212.8508
ASSET 105 45.8708 63.6783 4.0471 362.6007
LIAB 105 24.2813 32.4041 1.6101 189.9152
NIBX 105 1.7184 5.1643 -30.3524 23.8441
FS 105 6.5630 26.5491 -2.3135 161.2359
The final observation for hypothesis 1 (equity method), consists of 105 firm years. Table 4 provides the minimum, maximum, means, standard deviation and standard error for each variable. Again we see that the descriptive statistics show that the range between
minimum and maximum amount for each variable is large, and the standard deviation is large.
5.1.2 Descriptive statistics hypothesis 2
Hypothesis 2:
The final sample observations for hypothesis 2 (corridor method) consisted of 75 firm-years. The table below provides the minimum, maximum, mean and standard deviation for each variable. The descriptive statistics show that the range between minimum and maximum does not differ much, except for COREEARN. The standard deviation is low, which is an indication that the data is not spread out over a wide range of values, again except for COREEARN. Furthermore, the standard errors are given and these reflect the level of uncertainty in the mean.
Table 5 – Descriptive statistics hypothesis 2 (corridor method)
Variable Obs Mean Std.Dev. Min Max
MVE 75 0.6995 0.5891 0.0162 3.7529
BVE 75 0.3961 0.1757 0.0353 0.7804
PENSEARN 75 0.0388 0.1431 -0.1757 0.6624
NPA 75 0.0462 0.6984 -2.3856 3.4564
COREEARN 75 0.9403 5.6511 -0.8746 35.4309
The final sample for hypothesis 2 (equity method) consisted of 50 firm-years. The table below provides the minim, maximum, mean and standard deviation for each variable. Table 6 – Descriptive statistics hypothesis 2 (equity method)
Variable Obs Mean Std.Dev. Min Max
MVE 50 0.7593 0.54320 0.3653 2.5789
BVE 50 0.3960 0.2478 -0.2791 0.9338
PENSEARN 50 0.0059 0.0845 -0.2525 0.3226
NPA 50 -0.3454 0.2702 -1.5981 0.2222
COREEARN 50 1.4660 6.9945 -0.2337 37.0566
5.2 Test results and finding hypothesis 1
5.2.1 Test results and findings for companies, which used the Corridor method Equation 1 (see subparagraph 4.2.2) is used to test for the role of funding status in explaining market values. In table 7 the pooled regression results for companies that applied the corridor method are shown.
Table 7 – Results hypothesis 1 (corridor method)
Source SS df MS Number of obs
F( 4 , 70 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 75 100.60 0.000 0.8518 0.8434 13.229 Model 70429.1726 4 17607.2932 Residual 12251.0294 70 175.014706 Total 82680.2021 74 1117.30003
MVE Coef. Std.Err. t P>|t| [95% Conf. Interval]
ASSET 0.2279 0.1137 2.00 0.049 0.0010 0.4547
LIAB 0.3340 0.1525 2.19 0.032 0.0300 0.6380
NIBX 2.2730 0.4432 5.13 0.000 1.3890 3.1570
FS 0.176857 0.2222 0.80 0.429 -0.2663 0.6200
_cons 2.8072 2.0940 1.34 0.184 -1.3692 6.9835
It is expected that the coefficient on FS should be positive and significant if over- and underfunded plans are treated equally (Wiedman and Wier, 2004, p. 235). However the results are not in line with the expectation. The coefficient of interest is FS and the regression results of H1 reveal that the coefficient on FS is positive but not significant at a 5% significance level. So, a higher FS will lead to a higher MVE and a lower FS will lead to a lower MVE. Based on the results, it can be concluded that over- and
underfunded plans are not treated equally under the corridor method. The output reveals an adjusted R2 of 0.8434. This implies that the model has an explanatory power of
approximately 84%.
5.2.2 Test results and findings for companies, which used the Equity method The same equation is also used to perform the regression analysis for the equity method. In table 8 the pooled regression results are shown.
Table 8 – Results hypothesis 1 (equity method)
Source SS df MS Number of obs
F( 4 , 45 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 105 171.95 0.0000 0.8731 0.8680 13.157 Model 119068.968 4 29767.2421 Residual 17311.6218 100 173.1162 Total 136380.59 104 1311.3518
MVE Coef. Std.Err. t P>|t| [95% Conf. Interval]
ASSET 0.7804 0.14511 5.38 0.000 0.4925 1.0683
LIAB -0.6129 0.2848 -2.15 0.034 -1.1778 -0.0479
NIBX 0.0416 0.3353 0.12 0.901 -0.6235 0.7068
FS 0.1431 0.1328 1.08 0.284 -0.1203 0.4066
_cons 11.8423 2.2353 5.30 0.000 7.4076 16.2770
The test results of hypothesis 1, companies that use the equity method, again reveal a positive coefficient on FS and not significant. So, a higher FS will lead to a higher MVE and a lower FS will lead to a lower MVE. Again suggesting that under the equity method over- and underfunded plans are not treated equally. Furthermore, there is an adjusted R2 of 0.8680. This implies that the model has an explanatory power of approximately 87%. 5.2.3 Additional analysis
In the additional analysis a dummy variable was created. The dummy variable equals 1 for every absolute value of FS and equals 0 for every absolute negative value.
First, to see if funded and unfunded pension plans are treated equally, I would like the coefficient on FS to be positive and significant. Second, I would like to see if a portion of the pension surplus belongs to employees. In order for that to be so, I would like the coefficient on FSOVER to be negative. This is consistent with the labor economics perspective, which states that the pension fund is a separate legal entity from the employer sponsor. Under this perspective the net FS of a pension plan is only shown on the balance sheet when there is a deficit. So, a pension surplus is not viewed as an economic resource to the sponsoring firm according to this perspective (Wiedman and Wier, 2004, p. 231).
Equation 1b - The Ohlson model, extended version (Wiedman and Wier, 2004) MVE
it= α _+ β1*ASSETit + β2*LIABit + β3*NIBXit + β4*FSit + β5*OVERit + β6*F MVE
it = the year end market value of firm i in period t - computed as the number of common shares outstanding times the share price of the company’s common stock at the end of the quarter following the year-end
ASSETit = total assets of firm i at the end of year t LIABit = total liabilities of firm i at the end of year t
NIBXit = net income before extraordinary items of firm i for year t
FSit= funding status of the DB pension plan of firm i at the end of year t = pension assets -/- pension liabilities
OVER= Dummy variable. Equal to one for overfunded plans and equal to zero for underfunded plans
εit= error term
5.2.3.1 Results additional analysis results hypothesis 1 (corridor method)
Table 9 – Additional analysis results hypothesis 1 (corridor method)
Source SS df MS Number of obs
F( 6 , 68 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 75 68.22 0.0000 0.8575 0.8450 13.161 Model 70901.4433 6 11816.9072 Residual 11778.7587 68 173.2170 Total 82680.2021 74 1117.3000
MVE Coef. Std.Err. T P>|t| [95% Conf. Interval]
ASSET 0.2351 0.1133 2.07 0.042 0.0089 0.4613 LIAB 0.3181 0.1522 2.09 0.040 0.0144 0.6218 NIBX 2.2817 0.4414 5.17 0.000 1.4010 3.1625 FS -0.4759 0.4541 -1.05 0.298 -1.3821 0.4302 OVER 2.7939 4.4380 0.63 0.531 -6.0619 11.6498 FSOVER 0.8553 0.5210 1.64 0.105 -0.1842 1.8949 _cons -0.2466 4.2776 -0.06 0.954 -8.7824 8.2893
The results presented in table 9 under H1 dummy reveal that FS has a negative
coefficient, but not significant. From this it can be concluded that pension plans are not treated equally. The interaction term FSOVER reveals a positive coefficient, implying that under the corridor method pension surpluses do not belong to employees. The explanatory power, as reflected in adjusted R2,decreases from 84.34% to approximately 84.50%, which is minim. However, since the coefficient for OVER and FSOVER are both insignificant at a 5% level it can be concluded that these variables have no importance in the regression.
So, the test results of hypothesis 1 with the interaction term do not provide evidence that the funded status of a company’s pension plan is more strongly associated with its stock price for under funded plans than for overfunded plans under the corridor method.
5.2.3.2 Results additional analysis results hypothesis 1 (equity method)
Table 10 – Additional analysis results hypothesis 1 (equity method)
Source SS df MS Number ofobs
F( 6 , 43 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 105 115.20 0.0000 0.8758 0.8682 13.145 Model 119445.803 6 19907.6338 Residual 16934.7871 98 172.8040 Total 136380.95 104 1311.3518
MVE Coef. Std.Err. T P>|t| [95% Conf. Interval]
ASSET 0.8041 0.1618 4.97 0.000 0.4829 1.1253 LIAB -0.6420 0.3217 -2.00 0.049 -1.2803 -0.0036 NIBX 0.0686 0.3400 0.20 0.841 -0.6061 0.7432 FS -0.3102 6.3946 -0.05 0.961 -13.0002 12.3797 OVER 7.2326 6.7536 -1.07 0.287 -6.1697 20.6350 FSOVER 0.4242 6.3870 0.07 0.947 -12.2506 13.0991 _cons 4.9638 7.1667 0.69 0.490 -9.2583 19.1859
The results presented in table 10 under H1 dummy again revealing that FS has a negative coefficient, but not significant. From this it can be concluded that pension plans are not treated equally. The interaction term FSOVER reveals a positive coefficient, again
implying that under the equity method pension surpluses do not belong to employees. The explanatory power, as reflected in adjusted R2, increases from 86.80% to
approximately 86.82%, which also is minim.
So again, the test results of hypothesis 1 with the interaction term do not provide evidence that the funded status of a company’s pension plan is more strongly associated with its stock price for under funded plans than for overfunded plans under the equity method.
5.2.4 Conclusion
The hypothesis states that the funded status of a company's pension plan is more strongly associated with its stock price. The results for both the corridor- and equity method reveal that over- and underfunded plans are not treated equally. So, the funded status of a company’s pension plan, are not associated with market values. However, if we compare the R2 we can see that the adjusted R2 increases from 84%, for the corridor method, to 87% for the equity method.
5.3 Test results and finding hypothesis 2
The third hypothesis states that R2
equity method > R 2
corridor method. According to the corridor method only the accumulated actuarial gains and losses that exceed a predetermined level (10%) have to be recognized in profit and loss. By testing this hypothesis, we can
evaluate whether it was necessary to abolish the corridor method. And whether the corridor method really is less value relevant than the equity method.
5.3.1 Test results and findings for companies, which used the Corridor method Because actuarial gains and losses are offset, pension obligations and pension assets were not recognized on the balance sheet (Jullens, 2011, p. 3). Due to this it was expected that companies using the corridor method would give a less appropriate reflection of
corporate pension accounting information, than companies that are using the equity method to account for their DB plans. The regression results are given in table 11.
Table 11 – Result hypothesis 2 (corridor method)
Source SS df MS Number of obs
F( 4 , 70 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 75 29.30 0.000 0.6261 0.6047 0.3704 Model 16.0802 4 4.0200 Residual 9.6036 70 0.1372 Total 25.6838 74 0.3471 MVE Coef. Std.Err. t P>|t| [95% Conf. Interval]
BVE 0.8695 0.2754 3.16 0.002 0.3203 1.4187
PENSEARN 1.8301 0.3526 5.19 0.000 1.1270 2.5333
NPA 0.2910 0.0652 4.46 0.000 0.1609 0.4210
COREEARN -0.0121 0.0077 -1.57 0.122 -0.0275 0.0033
_cons 0.2821 0.1116 2.53 0.014 0.0594 0.5047
The results show that the coefficients on book value of equity, pension earnings and net pension assets are positive and significant. COREEARN has a negative coefficient and is not significant. The regression gives an adjusted R2 of approximately 60%, which is higher than the adjusted R2 of the equity method (54%).
5.3.2 Test results and findings for companies, which used the Equity method According to the equity method all gains and losses are immediately recognized outside the profit and loss. Due to this it was expected that the R2 for the companies after year-end 2012 should be higher. This would indicate a more appropriate reflection of corporate pension accounting information. Table 12 provides the results.
Table 12 – Result hypothesis 2 (equity method)
Source SS df MS Number of obs
F( 4, 45 ) Prob > F R-squared Adj R-squared Root MSE = = = = = = 50 15.61 0.0000 0.5812 0.5439 0.3668 Model 8.4025 4 2.1006 Residual 6.0558 45 0.1346 Total 14.4583 49 0.2951 MVE Coef. Std.Err. t P>|t| [95% Conf. Interval]
BVE 1.2498 0.2766 4.52 0.000 0.6927 1.8070
PENSEARN -1.8921 1.2446 -1.52 0.135 -4.3989 0.6147
NPA -1.1382 0.3474 -3.28 0.002 -1.8380 -0.4385
COREEARN -0.01425 0.0076 -1.91 0.062 -0.2981 0.0008
_cons 0.2574 0.1180 2.18 0.035 0.1962 0.4951
From the results above it can be seen that all coefficients are negative. PENSEARN and COREEARN are not significant, however NPA is now significant. Even though the sample size is rather small, the results are in line with Vuijk (2010) and Nandram (2011), whom also make use of EU listed company data. The regression gives an adjusted R2 of approximately 54%.
5.3.3 Conclusion It was expected that R2
equity method > R 2
corridor method. Starting from fiscal year 2013 companies were no longer allowed to defer their actuarial gains and losses. This was done in order to improve transparency and comparability. So, it was expected that the R2 from the equity method would be higher than the R2 from the corridor method. However, the regression results do not support the expectation. This suggests that companies using the equity method will not give a more appropriate reflection of corporate pension
accounting information than companies using the corridor method. The finding, from this perspective, implies that is was not necessary to abolish the corridor method. Since the abolishment lead to a total of 23.6 billion euros in actuarial losses.
6 Conclusion
Since January 1, 2005 all European listed companies were exposed to the adoption of IFRS. They were obliged to prepare their consolidated financial statements according IFRS. IAS19 ‘employee benefits’ was one of the most challenging areas in accounting and more specifically accounting for defined benefit plans.
Many researchers came to the conclusion that firms with DB pension schemes are often misvalued by the market. However, some even concluded that stock prices are fully efficient in reflecting corporate pension accounting information. Most of these researches were based on US company data. Therefore, I wanted to see whether equity values of firms with defined benefit plans reflect true economic value of pension assets and liabilities in the Netherlands. In order to answer this question two hypothesis were tested.
The first hypothesis tested whether the funded status of a company’s pension plan is more strongly associated with its stock price. The results for both the corridor- and equity method reveal that over- and underfunded plans are not treated equally. But there is no evidence that the funded status of a company’s pension plan, are not associated with market values.
The second hypothesis tested whether the equity method leads to more value relevant information compared to the corridor method. It was expected that the equity method provided more value relevant information. However, from the results shown we can conclude that the corridor method (R2=60%) is more value relevant than the equity method (R2=56%) by 4%. This is in line with research conducted by Vuijk, S. (2010) and Nandram, K. (2011), whom also use EU company data. Based on this the regression results do not support the expectation. This suggests that companies using the equity method will not give a more appropriate reflection of corporate pension accounting information than companies using the corridor method. The finding, from this
perspective, implies that is was not necessary to abolish the corridor method. Since the abolishment lead to a total of 23.6 billion euros in actuarial losses.
The results of both hypothesis lead to the conclusion that equity values of firms with DB plans under the corridor method and under the equity method do not reflect the true economic value of pension assets and liabilities.
The first contribution of this thesis is an extend to the existing literature on value relevance of pension accounting information as over the last decades there were a lot of
value relevance studies performed on pension accounting based on US company data which prepare their consolidated financial statements under US GAAP. The second contribution of my study is that I directly address the currently improved standard, which may lead standard setters to critically think whether this change leads to more value relevance and whether it was necessary, because the abolishment of the corridor method led to a total of 23.6 billion euros in actuarial losses. The third contribution, to my knowledge, is that this is one of the few studies to examine value-relevance of pension accounting in the Netherlands.
As follow up research I would recommend to test the credit relevance of pension accounting because a multi-user perspective can lead to different conclusions and may lead to more reliable inferences (Holthausen and Watts, 2001).
References
Barth, M., Beaver, W. and Landsman, W. (1993). ‘A Structural Analysis of Pension Disclosures under SFAS 87 and their Relation to Share Prices’. Financial
Analysts Journal, 49(1): 18-26.
Beaver, W. (2002). ‘Perspectives on Recent Capital Market Research’. Accounting
Review, 77(2): 453-474.
Carpenter, B., & Mahoney, D. (2004). ‘Toward the Goal of Teaching Excellence in Accounting: Reexamining the Role of Neutrality and Conservatism in the Development of Pension Accounting Standards’, Journal of College Teaching &
Learning, 1(9), 57-64
Christian, D. and Lüdenbach, N. (2013). IFRS essentials. Chichester: West Sussex, United Kingdom: John Wiley & Sons Ltd.
Coronado, J. and Sharpe, S. (2003) ‘Did Pension Plan Accounting Contribute to a Stock Market “Bubble”?’ Brookings Papers on Economic Activity (March) 1: 323- 371.
Coronado, J., Mitchell, O., Sharpe, S. and Nesbitt, S. (2008). ‘Footnotes aren’t enough: The Impact of Pension Accounting on Stock Values’. NBER Working
Paper Series, National Bureau of Economic Research: Cambridge.
De Jong, A., Rosellón, M. and Verwijmeren, P. (2006). ‘The economic consequences of IFRS: the impact of IAS 32 on preference shares in the Netherlands’, European Accounting Review, vol. 15, Supplement: Accounting in Europe, 167-186
Deegan, C., and J. Unerman. (2006). ‘Financial Accounting Theory’. European
edition: McGraw –Hill
Dhaliwal, D. (1986) ‘Measurement of Financial Leverage in the Presence of Unfunded Pension Obligations’ The Accounting Review, 61(4): 651-661.
Donton, A., Radhakrishnan, S. & Ronen, J. (2007) ‘Is stock price a good measure for assessing value-relevance of earnings; An empirical test’. Review of managerial
science 1 (1), 3-45
Fasshauer J., Glaum, M. & Street, D., (2008) ‘Adoption of IAS19R by Europe’s premier listed companies: Corridor approach versus full recognition: Summary of an ACCA Research monograph’. Journal of International Accounting, Auditing and
