The influence of fair value adjustments on dividend policy : with an interest in investment property

(1)

Deborah Ruurs Student number 5897513 deborah.ruurs@student.uva.nl

MSc Accountancy & Control, variant Accountancy & Control Amsterdam Business School

Faculty of Economics and Business, University of Amsterdam

Supervisor

dr. A. (Alexandros) Sikalidis

Final version 18th June 2014

The Influence of Fair Value Adjustments on

Dividend Policy

(2)

Abstract

The purpose of this paper is to investigate the impact of fair value adjustments on dividend policy. When fair value adjustments are transitory in nature and if managers are able to estimate their implications for future earnings, then fair value adjustments in net income are expected to have no distribution consequences. At the same time, fair value adjustments may lead to higher dividends if management incorrectly estimate their persistence. This can cause a cyclical impact, because higher dividends increase leverage and consequently risk. Corresponding with the relevant framework (Lintner, 1956), only persistent income should be included in dividend payouts. Within this thesis a sample of 485 U.K. firm-year observations in the period 2006-2012 is used together with 79 observations from the Netherlands.

In order to accomplish this purpose, this thesis contains three main chapters. The first chapter focuses on the literature review, whereby fair value is explained through some theoretical background and the corporate framework. The second chapter covers the research design, whereby the United Kingdom is the main country chosen to investigate fair value adjustments, since they require some form of modification when IFRS-profit is required. The third chapter then shows the empirical findings on fair value adjustments on dividend policy within the U.K. and compared to the Netherlands, another European country.

Following from the analysis in this thesis study, I could argue that fair value adjustments do influence dividend policy. Since empirical evidence is found for the concern that fair value revaluations, both total revaluations and revaluations on investment property, influence the distribution of income coming from unrealised income. However, the results also suggest that IFRS fair value adjustments are not persistent, meaning that they cannot be used to reliably predict future income.

(3)

Master Thesis Accountancy & Control 3

1. Introduction

Fair value accounting (FVA) is becoming increasingly important within the accounting standards. Since fair value numbers are timely and more reliable as a decision mechanism, regulators suggest that fair values lead improved financial reporting (FASB, 2000). Therefore, investors should benefit from the introduction of fair value accounting, because it grants them to make better and more informed decisions (Barth & Landsman, 1995); (Barth, 2007). But, according to Penman (2007) and Plantin et al. (2008) fair value accounting can lead to large transitory changes in net income. When important stakeholders fail to correctly estimate the information in earnings components, embodying volatile fair value adjustments in net income may introduce noise in decision making (Petroni & Wahlen, 1995); (Cornett, Rezaee, & Terhranian, 1996); (Hung & Subramanyam, 2007). Now to investigate whether fair value accounting has an effect on decision-making, the impact of mark-to-market adjustments on dividend policies is examined within this thesis.

The effect of positive fair value adjustments on dividends is estimated using the framework of Lintner (1956), which shapes the link between dividends and earnings components (Jagannathan, Stephens, & Weisbach, 2000); (Brav, Graham, Harvey, & Michaely, 2005). When following the Lintner Framework, it is stated that firms strive for a stable dividend development in relation to their earnings. When the management of the firm correctly estimates the implications of transitory fair value increments for future earnings, fair value adjustments are not relevant for distribution, and there is no relationship expected between fair value adjustments and dividends. But, if the management values the adjustments to be persistent, this then may impact dividend distributions. According to Enria et al. (2004), regulators have expressed their concerns that positive fair value adjustments will lead to increased dividend payouts and because of this to increased leverage. This may then result in firms being financially overrated and are vulnerable to economic shocks. So, if management increases dividends ensuing fair value adjustments, the assumption that fair value accounting can cause an exceptional degree of cyclicality (i.e. strengthen swings in the real economy) is supported (Plantin, Sapra, & Shin, 2008); (Laux & Leuz, 2009). Hence, in the view of regulatory considerations and the interest of the investment community, ‘whether and how dividend policy is influenced by fair value accounting is an important empirical question’ (Goncharov & van Triest, 2011).

(5)

Master Thesis Accountancy & Control 5 With this in mind, the research question of this thesis will be:

 How do fair value adjustments influence dividend policy?

To be able to answer this question, a literature part wherein fair value accounting/adjustments will be explained and an empirical research part wherein fair value adjustments will be investigated are needed. This thesis then aims to develop a better understanding of the influence and impact of fair value adjustments on dividend policy with the use of a sample of 151 U.K. listed firms during the period 2006-2012. Whereby these years represent a period of economic growth and financial crisis. Then analyzing and comparing the outcomes of U.K. firms with the outcomes of the Netherlands, another selected European country, will result in accepting or rejecting the hypotheses.

1.1 Motivation for this study

The motivation for this thesis subject comes from discussing fair value accounting within the master courses. After being introduced to the fair value accounting subject and discussing the possibilities concerning a thesis about fair value adjustments with my supervisor, I found it very interesting on how to deal with those fair value adjustments on dividend policy and so I decided to take on the subject for my thesis. In that way I could find out what fair value adjustments really are, what can and cannot be done with revaluations and how to work with and account for fair value adjustments.

Hereby I then contribute to the literature on the economic consequences of fair value accounting. It has been argued that fair value accounting exacerbates swings in the financial system by either increasing leverage during periods of growth or by inducing contamination in financial markets during economic downturns (Boyer, 2007); (Caruana & Pazarbasioglu, 2008); (Plantin, Sapra, & Shin, 2008). There is a concern that fair value accounting is cyclical and increases leverage during growth periods, however the empirical evidence on whether fair value accounting affects decision making is scarce (Beatty, 2007). So in this thesis it is shown that fair value accounting, or mark-to-market accounting, for fair value revaluations, and specifically for investment property does not increase dividends during the period of economic growth.

(6)

Master Thesis Accountancy & Control 6 1.2 Structure of the thesis

The thesis basically consist of three parts, whereby the first part contains a theoretical field, the second part provides a more practical view and the third part discusses the findings. In order to fairly answer the research question and accept or reject the hypotheses several academic articles and papers are used along with an empirical, data research. In part one the institutional framework and some theoretical background are described. Where after the hypotheses will be posed and explained with the use of earnings persistence and dividends. Then in the second part which builds on the hypothesis development, the practical research design will be illustrated by stating the research sample and explaining the benchmark equation and dividend policy. Thereafter the actual data collection will be provided and explained with the use of descriptive statistics and multivariate analysis. In the third and final part of this thesis, the hypotheses will be accepted or rejected and a conclusion will be made. But first a summary of the study will be given, because from this it will be easier for the reader to interpret the conclusions.

1.3 Limitations

There are some mentionable limitations coming with this thesis study.  There are no industry and year dummies included

 A rather large country, United Kingdom, is compared to a rather small one, the Netherlands

 Numbers are not always available which makes it harder to conduct a fair research  No sensitivity analysis has been conducted

These limitations provide a possibility for further research since it might be necessary to extend the research by including dummies, conducting a sensitivity analysis and using more European countries for comparisons.

(7)

2. Literature Review

The literature part of the study describes the important theoretical elements of the concept of fair value accounting/adjustments whereby there will be a nice and clear structure so that the overall story has a neat flow. Therefore a start will be made with the corporate framework concerning the United Kingdom and the Netherlands. Thereafter fair value accounting will be explained by its goal, its definition and its benefits and disadvantages. This is then followed by the description of earnings persistence and dividends and finally, the development of the hypotheses will be discussed.

2.1 Corporate Framework

2.1.1 Corporate Framework United Kingdom

Located in London City, the London Stock Exchange (LSE) is the oldest and fourth-largest stock exchange in the world. The LSE is part of a diversified international exchange Group being the London Stock Exchange Group (LSE.L). The Exchange was founded in 1081 and its current premises are situated in Paternoster Square. The LSE is the most international of all the world’s stock exchanges with a market value of £ 4,261,682 and around 2,500 companies from over 70 countries admitted to trading on its markets (LSE, 2014). Furthermore, the London Stock Exchange runs several markets for listing whereby an opportunity is given for different sized companies to list. Like for the biggest companies exists the Premium Listed Main Market, while for smaller SME’s the Stock Exchange operates the Alternative Investment Market and for international companies that fall outside the EU it operates the Depository Receipt schema as a form of listing and raising capital (WSE, 2014).

The first ‘Recommendations on Accounting Principles’ were presented in 1942 on the subjects of Tax Reserve Certificates and War Damage Contributions, and Premiums and Claims. These recommendations provided members of the Institute of Chartered Accountants in England and Wales (ICAEW) with early guidance on accounting practice. Subsequently, in 1969 the ICAEW released a ‘Statement of Intent on Accounting Standards in the 1970s’ which showed the determination of the council to advance accounting standards

(8)

Master Thesis Accountancy & Control 8 whereby the steps the institute felt that would be necessary to achieve this are pointed out. Then in 1971 the first ‘Statement of Standard Accounting Practice (SSAP) on Accounting for the results of associated companies was issued, whereby a total of 34 statements were released over time. The Council’s ‘Recommendations on Accounting Principles’ are persuasive in intent and departures from them do not necessarily require disclosure like departures from accounting standards do. During 1990, the Accounting Standards Board (ASB) adopted a number of SSAPs that had been released by the Accounting Standards Committee (ASC) so that they were brought within the legal definition of accounting standards according to the Companies Act 1985. In 2004, the government took the decision to strengthen the regulatory system in the U.K. after the major corporate collapses in the U.S. Because of this, the FRC’s role was enlarged to become the single independent regulator of the accounting and auditing profession as well as being responsible for adopting accounting standards and dealing with their enforcement (IASPLUS, 2014).

From 2005 on, listed groups in the U.K. have been required to prepare their consolidated financial statements in accordance with International Financial Reporting Standards (IFRSs). Almost all other companies and groups have a choice whereby they can choose to follow IFRSs or UK GAAP, and there is an additional option for small companies to follow the Financial Reporting Standard for Smaller Entities (FRSSE). Now for the periods beginning on or after 1 January 2015 three new Financial Reporting Standards, being FRS 100, 101 and 102, will take effect bringing with them a number of new options for all UK companies and groups. Even though these new standards are not yet mandatorily effective, they have been published and so are available for early, voluntarily adoption. The three new FRSs have been evolved by the Accounting Standards Board (ASB) to replace the old UK GAAP and to present an IFRS-based concentrated framework for certain companies (ICAEW, 2014).

2.1.2 Corporate Framework the Netherlands

Located in Amsterdam and established four centuries ago, the Amsterdam stock exchange is the oldest exchange in the world. The trading of corporate shares started in Amsterdam in 1607. At this time the shipping giant ‘Verenigde Oostindische Compagnie’ was in permanent need of funds to finance the transport of goods from the Far East. At first, shares that could

(9)

Master Thesis Accountancy & Control 9 be delivered immediately were traded, but soon futures and options were being moved as well. By the end of the 17th century, the Amsterdam stock exchange had a permanent place at the heart of the financial sector.

Euronext is the first pan-European exchange spread over Belgium, France, the Netherlands, Portugal and the U.K. It is created in 2000 and unites markets which date back to the start of the 17th century. The Euronext is the primary exchange in the Eurozone with over 1,300 issuers worth € 2,6 trillion in market capitalization. Besides it has an unmatched blue chip franchise consisting of 20+ issuers in the EURO STOXX 50 ® benchmark and a strong diverse domestic and international client base (Euronext, 2014). By offering market participants a comprehensive range of services to meet their needs, the Euronext operated regulated and transparent equity together with derivatives markets.

2.2 Theoretical Background and Hypotheses Development

Fair value accounting, also ascribed as mark-to-market accounting, has played an important role in the U.S. GAAP (generally accepted accounting principles) for more than 50 years (Ryan, 2008). In 1975, the modern standard-setting history of fair value accounting under U.S. GAAP began with guidance standards issued by the FASB (Financial Accounting Standards Board) that demands that marketable securities are recorded at the lower of cost or fair value (FASB, 1975). Then in 1991, the FASB made up the original fair value standard, when it issued guidance requiring the fair value of financial instruments to be disclosed in a company its financial statements (FASB, 1991). In 1993, the FASB expanded the fair value recognition requirements by publishing a standard that required debt and equity securities that were held for trading or held for sale to be carried at fair value in the balance sheet, and the FASB required changes in fair value to be recognized in the income statement or in a category of equity being ‘other comprehensive income’ (FASB, 1993). And in 2006, the FASB published a new standard, FAS 157 ‘Fair Value Measurements’, which foresees a single and logic definition of fair value, determines a common framework for developing fair value estimates, and requires extensive disclosures about those fair value estimates (Ryan, 2008).

(10)

2.2.1 Fair Value Accounting (FVA)

2.2.1.1 The goal and use of Fair Value Accounting (FVA)

The goal of fair value accounting is to estimate as best as possible the prices at which the positions they currently hold would change hands in transactions based on current information and conditions. In order to meet this goal, firms should fully include current information about future cash flows and current risk adjusted discount rates in their fair value measurements (Ryan, 2008). Therefore, market prices should reflect all publicly available information about future cash flows. When using unadjusted or adjusted market prices to estimate fair values, they are referred to as mark-to-market values. However, if market prices for the same or similar situations are not available, firm should estimate fair values using valuation models. When using valuation models to estimate fair values, they are referred to as mark-to-model values.

When using fair value accounting, firms report the fair values on their balance sheets of positions they current hold. However, when fair value accounting is fully applied firms should also report the periodic changes in the fair value of the positions they currently hold on their income statements referred to as unrealized gains and losses. Those unrealized gains and losses derive from the arrival of new information about future cash flows and changes in risk-adjusted discount rates (Scott, 2012).

There is a main issue coming with fair value accounting being whether firms can and do estimate fair values accurately and without discretion. When taking a look at identical positions that are traded in liquid markets that provide unadjusted mark-to-market values, fair value accounting is generally the most accurate and least discretionary measurement attribute possible, even though liquid market get values wrong occasionally. In using and estimating mark-to-model values, firms have choices about which valuation models they could use and about which inputs they need to use in applying the chosen models. These inputs are often derived from historical data that imperfectly predict future cash flows or correspond to risk adjusted discount rated (Ryan, 2008). Besides, the periods that firms choose to analyse the historical data to determine the inputs can have high significant effects on their mark-to-model values. This issue with fair value accounting is enlightened in two ways. First, firms are required by FAS 157 (FASB, 2006) to disclose qualitative information about how they estimate fair values as well as they need to disclose quantitative

(11)

Master Thesis Accountancy & Control 11 information about their valuation inputs, unrealized gains and losses and other changes in the fair value of their positions. Second, a high number of fair value accounting standards require fair values to be re-estimates each quarter, so that past valuations errors can and should be corrected on an on-going timely basis. After all, fair value accounting should be the best possible measurement attribute causing firms’ management to make voluntary disclosures and for making investors aware of the critical questions that they need to ask management (Ryan, 2008).

The alternative to fair value accounting is some form of amortized cost accounting. This form of accounting uses historical information about future cash flows and risk adjusted discount rated from the of positions to account for them throughout their lives on firms’ balance sheets and income statements. Other than under fair value accounting, unrealized gains and losses are ignored until they are realized through the impairment in value, disposal or passage of time. When firms choose to dispose positions, a cumulative unrealized gains and losses that have developed since the inception or prior impairment will be recorded on their income statements (Scott, 2012).

2.2.1.2 FAS 157 and definition of Fair Value Accounting (FVA)

FAS 157 (FASB, 2006) hold essentially all of the current GAAP guidance regarding on how to measure fair values. However, it does not require fair value accounting for any position since it’s guidance is only relevant when other accounting standards require or permit positions to be accounted for at fair value. Even though FAS 157 became effective for fiscal years beginning at the end of 2007, most large financial institutions adopted the standard early in the first quarter of 2007. For those institutions, the standard has been applicable during the entirety of the economic crisis and as expected they have reported that a large portion of the losses results from the economic crisis.

According to FAS 157 (FASB, 2006), fair value is defined as “the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date”. Whereby at the measurement date means that the faire value should reflect the conditions that exist at the balance sheet, and an orderly transaction is one that is unforced and unhurried. Within IAS 39 (IASB, 2011) fair

(12)

Master Thesis Accountancy & Control 12 value is defined as “the amount for which an asset could be exchanged, or a liability settled, between knowledgeable, willing parties in an arm’s length transaction”.

Fair value accounting is a financial reporting approach whereby companies are required and permitted to measure and report on an ongoing basis certain assets and liabilities at estimates of the prices they would receive if they were to sell the assets or would pay if they were to be relieved of the liabilities (Ryan, 2008). When using fair value accounting, companies report losses when the fair values of their assets decrease or liabilities increase. Those losses then reduce the company it’s reported equity and it may also reduce company its reported net income.

Even though fair values have played an important role in U.S. GAAP, accounting standards that require or permit FVA have increased considerably in number and significance in recent years. When FAS 157 ‘Fair Value Measurements’ was issued by the FASB, an important and controversial new standard came to practice which provides significantly more comprehensive guidance to assist companies in estimating fair values.

2.2.1.3 Critiques and Benefits Fair Value Accounting (FVA)

The practical applicability of this guidance standard has been tested by the extreme market conditions during the financial crisis. In response to this financial crisis, there were some parties that have criticised fair value accounting and FAS 157 concerning the following: fair values are difficult to estimate and thus not reliable, reported losses are misleading because they are temporary and will reverse as markets return to normal, and reported losses have adversely affected market prices yielding further losses and increasing the overall risk of the financial system (Ryan, 2008).

These critiques have some reliability, but they are also somewhat misplaced since it is more relevant to investigate whether fair value accounting provides more useful information to investors than alternative accounting approaches. Some elements why FVA benefits inventors are: gains and losses resulting from changes in fair value estimates indicate economic events that companies and investors may find worthy of additional disclosures, FVA limits the company its ability to manipulate their net income because gains and losses on assets and liabilities are reported in the period they occur, FVA requires or permits

(13)

Master Thesis Accountancy & Control 13 companies to report amounts that are more accurate/timely/comparable than the amounts that would be reported under existing alternative accounting approaches even during extreme market conditions, and FVA requires or permits companies to report amounts that are updated on a regular and ongoing basis (Ryan, 2008).

2.2.2 Earnings Persistence and Dividends

With his model, Lintner (1956) was the first proposing that dividend payment depends on current earnings and past dividends. He also mentions that firms smooth and adjust their dividends to achieve their target payout ratio, this is done when companies aim to pay out a certain percent of permanent or core earnings as dividends and slowly adjust their current payouts to the target ratio. This then implies that dividends, or changes in dividend, can be displayed as a function of current earnings and past dividend policy.

Fama and Babiak (1968) support the Lintner model and with their study they show that in a sample of U.S. firms dividend changes are explained pretty good by using a regression model that includes current dividends, lagged earnings and a constant term. Subsequently, DeAngelo et al. (1992) report that the decision to cut dividends is strongly influenced by net income in the years surrounding the dividend decision. They find that cuts in dividend are more likely when there are high current losses and when ongoing future earnings do not increase. This linkage between earnings and dividends is also found in more recent U.S. data and studies (Brav, Graham, Harvey, & Michaely, 2005); (Jagannathan, Stephens, & Weisbach, 2000).

Studies outside the U.S. are limited, but they do support the link between dividends and earnings. For instance Kasanen, Kinnunen & Niskanen (1996) argue that Finnish companies link their dividend payouts to their earnings in order to meet dividend thresholds. Besides, Goergen, Renneboog & Correia da Silva (2005) and Correia da Silva, Marc & Renneboog (2004) find that earnings are the key determinant of dividend changes and that because dividends are not paid out of a firm its reserves, only current earnings are linked to dividend payouts. Hereby the findings of the German study of Georgen et al.’s (2005) support those of DeAngelo et al. (1992) in the U.S. and with their Dutch sample, Renneboog & Szilagyi (2007) show a strong link between past and current earnings and dividends.

(14)

Master Thesis Accountancy & Control 14 Firms with higher and more stable earnings pay higher dividends (Lintner, 1956). He furthermore states that when firms have high and ongoing earnings, managers are more susceptible to increase dividends, because sustainable dividend payouts are then highly probable. Other studies examining Linter’s argument and model confirm his conclusions. Jagannathan, Stephens & Weisbach (2000) find that dividends are related to permanent earnings components, while transitory components are not distributed. Besides, Koremendi and Zarowin (1996) show that permanent earnings are one of the important determinants of long-term dividend policy. And Skinner (2004) describes a strong link between dividend changes and long-term sustainable earnings. Overall, prior literature expresses that there is a strong link between ongoing income and dividend payouts.

2.2.3 Hypotheses Development

Fair value reporting is expected to increase the decision relevance and transparency of accounting information, because fair values reflect present economic conditions and incorporate market expectations about future cash flows (Barth, Beaver, & Landsman, 2001). Furthermore, fair value accounting is expected to decrease information asymmetries. The increase in relevance and transparency together with the decrease in information asymmetry lead to a significant decrease in dividend payouts (Hail, Tahoun, & Wang, 2013). On the other hand, FVA introduces additional transitory components in the income statement that may increase the volatility of the aggregate income and it may reduce the ability of investors and managers to accurately evaluate the long-run performance and profitability on which to base the dividend payout (Petroni & Wahlen, 1995); (Hung & Subramanyam, 2007). If investors and managers are not able to successfully deal with earnings volatility, noise might be created around decisions taken by investors or creditors. Plus it makes the managerial assessment of corporate profitability more difficult. Consequently, the examination about if fair value adjustments are considered transitory and therefore non-distributable is of particular importance (Poon, 2004).

According to the Samuelson theorem (1965); (1973) fair values move randomly so they lack predictive ability. But this is only the case for those fair values measured on theoretical values such as financial investments, whereby financial investments are valued at

(15)

Master Thesis Accountancy & Control 15 the initial/acquisition market value and posterior differences in fair value are transferred to the income statement (Hitz, 2007).

However, Hitz (2007) also mentions that in contrast to financial investments, fair value measures of non-financial assets are supplied to consider elements such as the value in use or announcements of private information regarding future cash flows. According to Christensen and Nikolaev (2013) the valuation of investment property indeed is derived from forecasting investment property net cash flows and the resulting discount at the expected rate of return. Therefore, fair value changes are more long-lasting since they are based on contracts and general market conditions which are less volatile. But it is suggested that market values of investment properties are probably influenced by expected returns rather than future cash flows (Geltner & Mei, 1995). Thus, if returns are more volatile because these returns depend on the preferences and predictions of asset owners, fair value changes together with income might be more transitory.

So when fair value adjustments are transitory and thus have no impact on the underlying core earnings (Ohlsen, 1999), no relationship is predicted between fair value adjustments and dividends, supposing that stakeholders are able to assess the implications for fair value adjustments for future earnings. That is to say that the relationship between core earnings and dividends remains after introducing a fair value adjustment. Therefore, the first hypothesis is as follows:

 H1: Fair value adjustments have no impact on dividend policy.

This hypothesis then has two underlying assumptions being that fair value adjustments are transitory and relevant stakeholders need to correctly assess the persistence of fair value adjustments. Furthermore, it can be investigated whether the evidence in support of hypothesis 1 can be generalized in the case of downward fair value adjustments. Thereby the focus is placed upon managerial optimism and debt contracting so that the conditions under which firms might choose to deviate from regulator recommendations and distribute fair value profits can be investigated.

Existing literature suggests that dividends are not related to volatile earnings components (Lintner, 1956); (Jagannathan, Stephens, & Weisbach, 2000). To the extent that fair value adjustments are persistent, the dividend distribution should be influenced by the persistent part. If fair value adjustments are transitory and therefore have no impact on the underlying or core earnings, it is predicted that there is no relationship between both

(16)

Master Thesis Accountancy & Control 16 positive and negative fair value adjustments and dividends. Thereby assuming that stakeholders are able to estimate the implications of fair value adjustments for future earnings. This meaning the relationship between core earnings and dividends persists after introducing a positive or negative fair value adjustment. Coming from the first hypothesis and following the existing literature, the second hypothesis then is:

 H2: Both positive and negative fair value adjustments have no impact on dividend

policy.

Previous studies state that managers exhibit over-optimism and tend to overestimate implications of current earnings for future earnings (Jensen M. , 1993). Likewise, there is evidence that investors fail to correctly estimate the persistence of earnings components (Sloan, 1996). Positive revaluations may be distributed as part of bottom-line income when the persistence of fair value adjustments is overrated. This way of arguing is consistent with the concern of the European Central Bank (ECB) that is saying that during economic upturns relevant stakeholders may be overly optimistic which might result in unrealised fair value adjustments being distributed as dividends (Enria, et al., 2004).

When hypothesis 2 will be rejected in favour of a positive or negative effect of fair value adjustments on dividends, the introduction of fair value accounting may have unintended consequences and it could cause risk to increase through a procyclical impact. Fair value adjustments will mostly be positive under economic upturns, causing when distributing these adjustments leverage will increase. When looking at a banking setting, where fair value revaluations are more likely to be used than in non-financial sectors, both the ECB and the IMF (International Monetary Fund) underline the importance of aligning dividend payouts with core earnings rather than with transitory adjustments. Furthermore, the Banco de España notes that faire value accounting ‘may affect the behaviour of bank managers, encouraging undesirable behaviour in terms of appropriate risk management. Thus a revaluation of assets might lead to an increase in the dividend payout to shareholders, restricting institutions’ capacity to soften intertemporal shocks’ (Viñals, 2008).

It is less clear whether fair value adjustments mitigates agency costs when companies that employ fair value adjustments use more debt than equity (Christensen & Nikolaev, 2013). Generally, managers can take actions that shift wealth directly from debt holders to shareholders, for instance by increasing debt levels or distributing cash to shareholders using dividends (Taylor, 2013). These kind of actions reduce the probability that lenders get paid

(17)

Master Thesis Accountancy & Control 17 their full amounts, whereas firms might use the fair value adjustments from investment properties and financial investment to signal out liquidation values to debt holders, through which firms are aiming for a reduced cost of debt (Shivakumar, 2013). However, the cost of debt can also be increased, because when the same values become known to the shareholders who are expecting a minimum dividend payment based on the number presented within the income statement, this higher expectation can increase the dividend payouts to the benefit of shareholders and at the expense of debt holders.

Restrictive debt covenants as dividend restrictions have been outlined as effectively mitigating debt holder – shareholder conflicts (Jensen & Meckling, 1976). Prior literature suggests that at a corporate level leverage might increase by the distribution of non-realised income, which potentially brings some firms into financial distress (Enria, et al., 2004). Specifically, Farinha (2003) states that increased leverage decreases corporate flexibility in undertaking investment projects due to debt covenants and so it increases the probability of default. However, at a macro-level, increased dividend payouts might contribute to the cyclicality of the financial system due to positive fair value adjustments (Laux & Leuz, 2009). Based on the above discussion and the question whether to distribute non-realised income or not, it is investigated whether firms’ borrowing capacity affects their decisions regarding the distribution of fair value income. Therefore the third hypothesis is as follows:

 H3: The distribution of income coming from fair value adjustments is not influenced

by corporate leverage.

The hypotheses are being tested for total fair value adjustments whereby on some points a specific interest within fair value adjustments on investment property will be shown. As for the persistence it will be examined whether fair value adjustments are persistent and whether investment property revaluations are persistent. The same will be done for the effect of dividend policy and fair value adjustments.

(18)

3. Research Design

Within this part of the study the important elements of the research design are described whereby there will be a nice and clear structure so that the overall design has a neat flow. Therefore a start will be made with the research sample concerning the time period and brief explanation of the observations. Thereafter the benchmark equations and dividend policy will be provided. This is then followed by the description selection issues paragraph in order to describe the selection issues.

3.1 Research Sample

The sample in this thesis covers firms listed on the LSE and AEX over the period 2006 – 2012. This timeframe is selected because since 2005 it is required to produce the financial statements according to IFRSs. Furthermore, this timeframe contains the economic crisis period, from this it can then be seen whether the crisis and fair value adjustments might influence each other. Fair value adjustment data on totals and on investment property from the annual reports of 151 companies listed on the LSE for the above period are collected. Whereby the final sample contains 485 firm observations and includes all the sectors of the LSE. The accounting data for each firm were taken from the Datastream database whereby they refer to single entity accounts. And for the Netherlands fair value adjustment data on totals and on investment property from annual reports of 25 companies are collected. Whereby the final sample contains 79 firm observations that also includes all the industry sectors. The included sectors for both the U.K. and the Netherlands are described within the appendix.

A comparison will be made between the United Kingdom fair value adjustments and another European country following the same category that they have a mandatory or possible profit distribution based on IFRS financial statements and whereby some form of modification is required if IFRS profit. This other European country then is the Netherlands. When making this comparison it can become clear how fair value adjustments influence dividend policy within different countries within the European Union that have the same requirements in following IFRSs.

(19)

Master Thesis Accountancy & Control 19 The initial sample contains 41,592 observations over the period 2005-2012. Here from, the observations over 2005 are initially excluded since they are only necessary to make comparisons and useable with lagged variables. Then also one observation is excluded that is a 2013 observation, this was included within the data sample because of the fiscal year end. Furthermore, all the N/A accounts, being the not available or not applicable ones, are excluded since only available data can be used in order to run regressions and obtain an outcome. Besides, 154 more N/A accounts are excluded since some lagged calculations could not have been made because of missing numbers. Therefore, the final sample contains 485 observations. This is done in the same way for the Netherlands, where the final sample contains 79 observations. Table 1 then summarizes the sample selection criteria.

Table 1: Sample Selection Criteria

Panel A (United Kingdom)

Description Deletion Number of observations

Initial sample 41,592

N/A data over Year 23,370 18,222

Data over 2005 2,568 15,654

Data over 2013 1 15,653

FVA data FVA total

FVA Investment Property

10,669 4,261

4,984 723

N/A data over EBIT 17 706

N/A data over Paid Dividend 1 705

N/A data over Cash & Cash eq. 65 640

N/A data over Short term Debt 1 639

(20)

Master Thesis Accountancy & Control 20 Panel B (the Netherlands)

Initial sample 3,536

N/A data over Year 1,947 1,589

Data over 2005 226 1,363

FVA data FVA total

FVA Investment Property

740 497

623 126

N/A data over Paid Dividend 2 124

N/A data over Cash & Cash eq. 20 104

N/A data over Calculations 25 79

Final Sample U.K. The Netherlands

2006 - 2012 485 79

There are observations from companies within the sample that only account for a couple of years within the time span of 2006 – 2012. It is assumed that this might be due to accounting changes and the publicly available data, therefore those observations are included within the sample since it is expected that it will contribute to a representative outcome.

3.2 Benchmark Equation and Dividend Policy

In testing the hypotheses, an extension of the Lintner (1956) framework is used, according to which firms adjust dividend payments considering their defined payout ratio. Correia da Silva et al. (2004) argue that the change in dividends based on the Lintner model can be created for regression purposes as follows:

(1)

Within this equation, for any year t and firm i, ∆Di,t is the actual change in dividends (D) from year t-1 to year t. E stands for the net distributable earnings, αi is a constant term which captures the drift in dividends through time, β1 and β2 are the model coefficients, and εi,t is the error term.

Because the hypotheses stated within this thesis are of conditional nature on the persistence of fair value adjustments, Sloan (1996) and Goncharov & van Triest (2011) are

(21)

Master Thesis Accountancy & Control 21 followed, since they test this condition by employing the following earnings persistence regressions:

(2)

Within this equation ROA_BFVi,t is the return or EBIT over total assets before total fair value adjustments and ROA_BFVi,t-1 is the lagged ROA before fair value adjustments. Furthermore, FairValue_REVi,t-1 is the total fair value revaluation adjustment.

(3) Within this equation ROA_BFINVi,t is the return or EBIT over total assets before fair value adjustments on investment property and ROA_BFINVi,t-1 is the lagged ROA before investment property adjustments. Furthermore, InvestmentProperty_REV is the fair value revaluation adjustment of investment property. All the numbers are scaled by average total assets. If fair value adjustments predict future income, they establish earnings components that are persistent and therefore should be part of the distributable earnings (Lintner, 1956). Coming from this, the adjustment coefficient, α2, should be different from zero. When the coefficients are positive, fair value adjustments bring good news about future profitability, and when the coefficients are negative, fair value adjustments signal a decrease in future earnings.

Equation 1 together with the model of Correia da Silva et al. (2004) forms a basis for the model for the multivariate analysis. Specifically, in equation 4 DDIFF is used, which is the difference of total dividends between year t and t-1 over total assets and Di,t-1 is the lagged dividend over total assets. The ROA_BFV variable is included to substitute the E variable for scaling purposes and to constitute the profitability measure for historical income. Besides, a fixed set of regressors is included in order to control for firm characteristics and the earnings components that are derived from fair value revaluations of aggregate earnings are separated. The model corresponding with equation 4 is as follows:

(4)

Within this equation FairValue_REV is the total value of non-realised revaluations over total assets, ROA_BFV and lagged ROA_BFV are proxies for current and past profitability and are defined as earnings before interest and taxes excluding total fair value revaluations over total assets. Furthermore SIZE is the logarithm of the book value of assets

(22)

Master Thesis Accountancy & Control 22 and is a proxy for firm size, DEBT is the debt-to-asset ratio and is a proxy for leverage, CASH is the cash-to-asset ratio and is a proxy for free cash flow, and GROWTH is the percentage of sales increase and is a proxy for growth opportunities. Within this thesis industry and year dummies are excluded from the original model. A consequence of this, as already mentioned before in the limitations part, is that this might cause the results to be somewhat different then the outcomes of similar studies.

Coming from hypothesis 1 it would be expected that α1 in equation 4 is zero, and then fair value adjustments would not affect dividend distributions. On the other hand, a significant coefficient would suggest that fair value adjustments affect dividend payouts. In this thesis a positive association between dividends and profitability measures and a negative relationship for lagged dividend payments is predicted, and there is no particular prediction formed about the coefficient of SIZE due to the ambiguity of prior results. DeAngelo et al. argue that larger firms should pay higher dividends (2004), whereas Smith & Watts argue for a negative association (1992). Further, DEBT is expected to have a negative coefficient because the flexibility of managers’ ability to use corporate resources decreases because of high debt. A positive coefficient for CASH is expected, because cash rich companies could pay dividends more easily to decrease free cash flows and potential agency costs. And a negative coefficient for GROWTH is expected, because companies with high growth opportunities are more likely to decrease dividend payments so that they can internally finance their expansion.

However, within this thesis there is only an interest in fair value adjustments concerning investment property instead of fair value revaluations as a whole (including investment property and financial securities). So coming from the fourth model to the fifth model FairValue_REV, ROA_BFVi,t and ROA_BFVi,t-1 are excluded and InvestmentProperty_REV, ROA_BINVi,t and ROA_BINVi,t-1 are included in order to estimate the relative effect of the investment property type of fair value adjustments on dividend policy. Then the fifth equation is as follows:

(5)

Within this equation InvestmentProperty_REV is the value of non-realised investment property revaluations over total assets, ROA_BINV and lagged ROA_BINV are proxies for

(23)

Master Thesis Accountancy & Control 23 current and past profitability and are defined as earnings before interest and taxes excluding total investment property fair value revaluations over total assets.

In order to investigate hypothesis 2, a distinction between positive and negative adjustments needs to be made in order to examine whether FVA affect dividend policy symmetrically. Whereby negative non-realised adjustments might not have negative effects as strong as the positive effects from positive fair value adjustments. This irregularity could be a result of the difference in reaction of investors and managers regarding expected returns, since these are based on positive or negative news. Coming from this, the sixth equation is as follows:

(6)

Within this equation FairValue_REV+ stands for positive total fair value adjustments over total assets and FairValue_REV- stands for negative total fair value adjustments over total assets. Whereby α1 and α2 are expected to be statistically non-significant if fair value adjustments are transitory. However, if fair value adjustments are persistent, the effect of positive and negative adjustments should be the same and similar to the effect of the current profitability measure. For example, a distinction within FairValue_REV can be made for InvestmentProperty_REV since it can be divided in InvestmentProperty_REV+ and InvestmentProperty_REV-.

To be able to investigate hypothesis 3 and see if corporate leverage influences the distribution of income coming from fair value adjustments, different versions of equation 4 and 5 are used. In particular for hypothesis 3 the following equation is used:

(7)

Within this equation the variable Short_Lev stands for the short term debt over total assets, and Long_Lev stands for the long term debt over total assets. The independent variables hereby are profitability, lagged profitability, lagged dividend, size, debt, cash and cash equivalents, and growth.

(24)

Master Thesis Accountancy & Control 24 3.3 Description Selection Issues

Coming from previous research, the accounting choice to apply fair value adjustments is voluntarily taken by companies and therefore it can be regarded as systematic (Goncharov & van Triest, 2011). Within the models used in this study selection biases are tested by employing the propensity score matching method (PSM). Prior studies (Goncharov & van Triest, 2011); (Hung & Subramanyam, 2007); (Leuz, 2003) are followed which suggest that financial performance, size, leverage, growth and cash levels are of influence on the accounting choices of fair value adjustments. A description is used to analyse the sample firms’ decisions to revalue total fair value adjustments and investment property adjustments alone.

For the United Kingdom a sample of 15.584 observations is used to take a close look at the use of fair value accounting on whether FVA is used or not. These consist of 4,375 firm-year observations where a fair value revaluation has taken place and 11,209 observations where no revaluation has taken place. Of these 4,375 observations, there are 720 firm-year observations in which revaluation of investment property have taken place and there are 3,655 firm-year observations in which revaluation other than of investment property have taken place. For the Netherlands a sample of 1,363 observations is used to take a close look at the use of fair value accounting on whether FVA is used or not. These consist of 552 firm-year observations where a fair value revaluation has taken place and 811 where no revaluation has taken place. Of these 552 observations, there are 125 firm-year observations in which revaluation of investment property have taken place and there are 427 firm-year observations in which revaluation other than of investment property have taken place. This is shown in table 2.

A selection issue arises when taking into account the possibility that the amount of investment property revaluation might just be a part of the overall fair value adjustments. This might be the case for firms that choose to implement fair value accounting for different types of revaluations and not just for investment property. A numerical example is given: When firm A has a fair value revaluation of € 10,000 and accounts € 7,500 of this to revaluation on investment property, the € 2,500 that is left will be accounted to another type of revaluation for instance on financial securities. Accounting for fair value adjustments in this way might cause issues in selecting a useable sample.

(25)

Table 2: Description Selection Issues

Description Number of observations

Initial sample Deleting N/A data

41,592 15,584

Yes No

Fair Value Adjustments

Fair Value Adjustments on Investment Property Fair Value Adjustments other than IP

4,375 720 3,655* 11,209 14,864 **

Panel B (the Netherlands)

Initial sample Deleting N/A data

3,536 1,363

Yes No

Fair Value Adjustments

Fair Value Adjustments on Investment Property Fair Value Adjustments other than IP

552 125 427* 811 1,238 **

* Difference between Fair Value Adjustments and Fair Value Adjustments on Investment Property, so those are revaluations but not on Investment Property.

** Amount cannot be stated, only research on Investment Property not on other types.

Final Sample U.K. The Netherlands

FVA

FVA on Investment Property

4,375 720

552 125

As can be measured only 28.1% of the companies in the United Kingdom implemented fair value accounting and uses fair value adjustments. Whereby the percentage of companies that uses fair value adjustments regarding investment property is 4.6%. These are rather low numbers. Whereas concerning the Netherlands 40.5% of the companies implemented fair value accounting and uses fair value adjustments whereby the percentage of companies that uses investment property fair value adjustments is 9.17%. The fact that the percentages of the Netherlands are higher than those of the United Kingdom is rather remarkable since the Netherlands is a much smaller country than the United Kingdom

(26)

Master Thesis Accountancy & Control 26 is. The percentages of investment property revaluations when fair value adjustments are made are 16.5% and 22.6% for the United Kingdom respectively the Netherlands. So following these percentages it can be said that fair value adjustments and fair value adjustments on investment property are more applied in the Netherlands.

(27)

4. Empirical Findings

Within this part of the study the findings of the research are described whereby there will be a nice and clear structure so that the overall analysis has a neat flow. Therefore a start will be made with the descriptive statistics, which will be explained with the use of several tables. Thereafter the multivariate analysis will be given. This including the persistence of income from fair value adjustments and the effect of fair value adjustments on dividend policy.

4.1 Descriptive Statistics

Table 3 provides an overview of the financial characteristics as the mean, median and standard deviation over the variables of the sample firms, whereby for the United Kingdom these statistics are given in panel A and for the Netherlands they are given in panel B. The median of DDIFF for the U.K. is 0, suggesting that with most firms there is no difference in dividend between year t and t-1. The mean {0.026} and the median {0.033} return of assets before any fair value adjustments {ROA_BFV} are somewhat close to similar, showing that the distribution of observations is close to normal. This also applies to the Netherlands where the mean is 0.043 and the median 0.050. For the U.K. an average firm would have a mean DEBT of 0.401 that is close to its median of 0.386, and for the Netherlands these numbers are even closer as they are 0.426 and 0.418 respectively. Then regarding GROWTH, there is a rather big difference between the mean {0.156} and the median {0.033}, showing that there are firms with high growth opportunities which affect the mean upwards. The same applies to the Netherlands where the mean is 0.152 and the median is 0.038.

Table 3: Descriptive Statistics

Variables Mean Median Std. Dev. Min. Max.

DDIFF ROAi,t Dii,t SIZEi,t 0.0002948 -0.0200824 0.0131032 12.67276 0.0000 0.0208755 0.0086407 12.77797 0.0216136 0.2075012 0.0328252 1.710604 -0.02583144 -3.064536 0.00000 8.194506 0.3154048 0.4460063 0.5935829 16.75694

(28)

Master Thesis Accountancy & Control 28 DEBTi,t CASHi,t GROWTHi,t ROA_BFVi,t FairValue_REVi,t InvestmentProperty_REVi,t 0.4013331 0.0704989 0.155665 0.0259549 -0.0460373 -0.0410512 0.3859509 0.0375054 0.0326984 0.0329283 -0.0054805 -0.0023069 0.2459265 0.1003969 2.208413 0.0652706 0.2034282 0.1999994 0.00000 0.00000 -40.16667 -0.5209447 -3.464385 -3.464385 2.04635 0.9804347 12.44985 0.4220275 0.1860787 0.1860772

Variables Mean Median Std. Dev. Min. Max.

DDIFF ROAi,t Dii,t SIZEi,t DEBTi,t CASHi,t GROWTHi,t ROA_BFVi,t FairValue_REVi,t InvestmentProperty_REVi,t 0.0021797 0.0351431 0.0189548 13.72024 0.4263814 0.0451836 0.1516871 0.0425509 -0.0074077 -0.0059992 0.0007349 0.0371077 0.0217149 13.87342 0.4147735 0.0133004 0.0377285 0.0501028 -0.0001909 -0.0000426 0.0099196 0.0658174 0.0141376 1.378275 0.186882 0.0641197 0.695783 0.0450679 0.0530949 0.0495491 -0.0345892 -0.1899473 0.00000 10.93982 0.1299589 0.00000 -0.9016112 -0.1158509 -0.2492825 -0.2480826 0.0345203 0.1445223 0.0533621 16.35127 1.127605 0.333217 4.236984 0.1223697 0.1075468 0.1029621 Table 4 provides the correlations between the variables. Off the 55 correlations, there are 16 correlations that have differences in positive and negative value between the United Kingdom and the Netherlands. However, the differences in correlations can be explained by size and amounts. So to further investigate the relations and work towards more definite conclusions, a multivariate analysis will be made whereby the correlated factors will be held constant.

(29)

Table 4: Correlations

Variables ROA Di SIZE DEBT CASH GROWTH ROA_BFV FV_REV* IP_REV* DDIFF

ROA 1.000 Di -0.3152 1.0000 SIZE 0.1466 -0.0754 1.0000 DEBT -0.3246 -0.1619 0.1933 1.0000 CASH 0.0142 0.1508 -0.2028 -0.2595 1.0000 GROWTH 0.1930 -0.6986 0.0949 0.1304 -0.0769 1.0000 ROA_BFV 0.2191 -0.2451 0.1253 0.1034 0.0386 0.3841 1.0000 FV_REV* 0.9497 -0.2428 0.1094 -0.3643 0.0021 0.0736 -0.0974 1.0000 IP_REV* 0.9450 -0.2467 0.1160 -0.3537 -0.0040 0.0806 -0.0939 0.9941 1.0000 DDIFF -0.0551 0.4314 -0.0207 -0.0649 0.0961 -0.1940 -0.0227 -0.0489 -0.0532 1.0000

Panel B (The Netherlands)

Variables ROA Di SIZE DEBT CASH GROWTH ROA_BFV FV_REV* IP_REV* DDIFF

ROA 1.000 Di 0.2909 1.0000 SIZE 0.2370 0.3492 1.0000 DEBT -0.5115 -0.3109 -0.2110 1.0000 CASH 0.0403 -0.1137 -0.0425 0.1255 1.0000 GROWTH 0.0506 -0.1816 -0.0710 -0.1802 0.0062 1.0000 ROA_BFV 0.5974 0.4906 0.2817 -0.5408 -0.1302 0.0662 1.0000 FV_REV* 0.7325 -0.0592 0.0547 -0.1750 0.1606 0.0065 -0.1083 1.0000 IP_REV* 0.7223 -0.0487 0.0683 -0.1789 -0.1725 0.0001 -0.1004 0.9806 1.0000 DDIFF 0.3034 0.4446 -0.0547 -0.1352 0.3183 -0.0225 0.1564 0.2433 0.2535 1.0000

* FV_REV abbreviation for FairValue_REV and IP_REV abbreviation for InvestmentProperty_REV

4.2 Multivariate Analysis

Coming from the previous paragraph, this paragraph will provide multivariate analyses in order to give a more accurate and precise view of the relations with fair value adjustments.

4.2.1 Persistence of Income from Fair Value Adjustments

Within this section the persistence of fair value adjustments will be determined. Whereby the persistence of historical income is considered. Table 5 shows that there is no strong evidence that fair value adjustments would be persistent, since the coefficient of fair value

(30)

Master Thesis Accountancy & Control 30 revaluations is -0.0971 for the United Kingdom and -0.3277 for the Netherlands versus 0.0864 for the U.K. and 0.3937 for the Netherlands for the accounting income before fair value adjustments. The coefficient of fair value revaluations on investment property is -0.0782 for the United Kingdom and -0.2462 for the Netherlands versus 0.0523 for the U.K. and 0.3570 for the Netherlands for the accounting income before fair value adjustments. These outcomes however could suggest that total fair value revaluations and fair value revaluations on investment property cause a negative relationship.

Table 5: Persistence of Fair Value Adjustments

Dependent Variable ROA_BFVi,t ROA_BINVi,t

Explanatory Variables Cf. ROA_BFVi,t-1 ROA_BINVi,t-1 FairValue_REVi,t-1 InvestmentProperty_REVi,t-1 0.0864334 -0.0970743* 0.052195** -0.078179***

Industry and Year dummies No No

R2 Number of Observations Prob>F 0.0344 485 0.0002 0.0144 485 0.0305 * P > |t| 0.001, ** P > |t| 0.028, *** P > |t| 0.014

Dependent Variable ROA_BFVi,t ROA_BINVi,t

Explanatory Variables Cf. ROA_BFVi,t-1 ROA_BINVi,t-1 FairValue_REVi,t-1 InvestmentProperty_REVi,t-1 0.3936713 -0.3276982 0.3569385 -0.2461554*

R2 Number of Observations Prob>F 0.3595 79 0.0000 0.2884 79 0.0000 * P > |t| 0.005

(31)

Master Thesis Accountancy & Control 31 Therefore, the IFRS adoption of fair value adjustments does not necessarily leads to the inclusion of new income components that do not change the persistence of income. The results propose that fair value adjustments do not reliably predict future income. This is contrary to prior literature that states that fair value adjustments do reliably predict future income (Goncharov & van Triest, 2011). But the results do support prior literature and studies on the relevance of fair value accounting. According to Lintner’s (Lintner, 1956) model, where it is stated that fair value adjustments are persistent, those adjustments should be part of the distributable earnings. These results are in line with the recommendations about unrealised profits, since income should not be distributed when this income comes from unrealised profits. However, there is another side to these results, if fair value adjustments, both total adjustments and adjustments on investment property, appear to be and remain negatively related, the results then propose that fair value adjustments do reliably predict future income although this then is in a negative way.

4.2.2 The Effect of Fair Value Adjustments on Dividend Policy

In first instance, it is investigated whether the revaluation component of net income impacts dividend payout. In order to accomplish this, the models of equation 4 and 5 are used. The results are provided within table 6 where DDIFFi,t is the dependent variable and stands for the difference in dividend between year t and t-1 over total assets.

Table 6: Dividend Policy and Fair Value Adjustments

Dependent Variable DDIFFi,t

Total Fair Value

DDIFFi,t Fair Value IP Explanatory Variables Cf. Cf. FairValue_REVi,t InvestmentProperty_REVi,t ROA_BFVi,t ROA_BFVi,t-1 ROA_BINVi,t 0.0130085* 0.0185671 0.0019381 0.0127274** 0.0213609

(32)

Master Thesis Accountancy & Control 32 ROA_BINVi,t-1 Dii,t SIZEi,t DEBTi,t CASHi,t GROWTHi,t 0.4097468 -0.0002489 0.0050279 0.0059114 0.0020299 0.0022566 0.4093526 -0.0002521 0.0049922 0.0056789 0.0020028

R2 Number of Observations Prob>F 0.2204 485 0.000 0.2208 485 0.0000 * P > |t| 0.027, ** P > |t| 0.031

Dependent Variable DDIFFi,t

Total Fair Value

DDIFFi,t Fair Value IP Explanatory Variables Cf. Cf. FairValue_REVi,t InvestmentProperty_REVi,t ROA_BFVi,t ROA_BFVi,t-1 ROA_BINVi,t ROA_BINVi,t-1 Dii,t SIZEi,t DEBTi,t CASHi,t GROWTHi,t 0.0480029* -0.0007451 0.0043016 0.4164537 -0.0018657 0.0010701 0.0510746 0.001004 0.052083** -0.0002974 0.0042 0.4152537 -0.0018867 0.0011792 0.0505404 0.0010183

R2 Number of Observations Prob>F 0.4447 79 0.0000 0.4465 79 0.0000 * P > |t| 0.025, ** P > |t| 0.019

This table indicates that total fair value adjustments do affect dividend policy for the United Kingdom {coefficient of 0.0130}, and for the Netherlands {coefficient of 0.0480}. When looking at the fair value adjustments on investment property, table 6 indicates that

(33)

Master Thesis Accountancy & Control 33 these revaluations affect dividend policy for the United Kingdom {coefficient of 0.0127}, and they affect dividend policy for the Netherlands {coefficient of 0.0521}. Therefore, hypothesis 1 ‘fair value adjustments have no impact on dividend policy’ will be rejected for both the United Kingdom and the Netherlands since fair value adjustments do have an impact on dividend policy. Thereby, the coefficients of the relation between dividend policy and fair value adjustments are on both total fair value revaluations and investment property revaluations alone higher within the Netherlands than within the United Kingdom.

Concerning the difference in positive and negative fair value adjustments, there is no direct evidence that positive or negative total fair value adjustments and investment property adjustments affect dividend changes differently. Within the United Kingdom there are 213 positive fair value revaluation observations and 272 negative observations. Additionally there are 222 positive investment property fair value revaluation observations and 263 negative observations. Coming from this, the difference between positive and negative fair value adjustments, related to the sample size, is not that big to have a significant effect. For the Netherlands, 33 positive fair value revaluations observations are found and 46 negative observations. Additionally there are 37 positive investment property fair value revaluation observations and 42 negative observations. From these results the same conclusion can be drawn as with the U.K., however the difference between positive and negative fair value adjustments is smaller than with the United Kingdom, but this could be the result of the sample size used. From this hypothesis 2 ‘both positive and negative fair

value adjustments have no impact on dividend policy’ is in some way accepted and in some

way rejected. Since there is no difference found in positive and negative fair value adjustments, but fair value adjustments do have an impact on dividend policy according to the results shown in table 6. What is striking is that there are more negative than positive fair value adjustments and within the negative observations, there are more negative fair value adjustments than negative investment property fair value adjustments. So there are more positive investment property fair value adjustments than positive fair value adjustments, meaning that overall investment property revaluations are positively made and fair value revaluations including revaluations on other types than investment property are negatively made.

In order to react upon the third hypothesis ‘the distribution of income coming from