Influence of fair value adjustments of investment property on dividend policy for the United Kingdom and Germany

(1)

Amsterdam Business School

Influence of fair value adjustments of investment property on dividend policy for the United Kingdom and Germany.

Name: Anna-Lin Kloosterman

Student number: 10344950

Thesis supervisor: Alexandros Sikalidis

Date: 20 June 2016

Word count: 11286

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Statement of Originality

This document is written by student Anna-Lin Kloosterman, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are 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

(3)

Abstract

In this study I follow the research of Goncharov and van Triest (2011), and the research of Sikalidis and Leventis (2016), who examined if fair value adjustment of investment property and financial securities have influence on dividend policies in respectively Russia and Greece. I want to implement this study for the United Kingdom and for Germany. For this study the model of Lintner (1956) is used, who stated that if income is persistent, it could be included in distributions to shareholders, or in other words included in dividends. These fair value adjustments are unrealized income and could therefore not be included in the

distributions according to the regulations in the United Kingdom. In Germany the regulations allow the distribution of unrealized income if some requirements are met. So the first aspect I want to examine is the persistence of income, which according to the Lintner theory (1956) is a requirement for distribution. Thereafter, I want to investigate the effects of the fair value adjustments on the dividend policy, and if these distributions are also affected by the quality of a corporate governance system that a firm has and the corporate leverage ratio of a firm. I found for both countries that earnings from revaluations of investment property are persistent, and should therefore according to the theory be available for distribution to shareholders. For the United Kingdom the fair value adjustments are negatively related with future earnings, and this leads to lower dividends. This is contrary to the theory of Lintner (1956), because the persistent income components should not lead to higher dividend payouts. In Germany a positive relationship was found between fair value adjustments and future earnings and dividends. In some circumstances companies in Germany are allowed to distribute the unrealized earnings, as recommended by the Lintner theory (1956).

(4)

Table of Contents Abstract ... 3 1 Introduction ... 5 2 Theory ... 8 2.1 Investment property ... 8 2.2 Earnings persistence ... 8 3 Institutional Framework ... 9

3.1 IFRS and UK GAAP for investment property. ... 9

3.2 Dividends in the United Kingdom ... 10

3.3 German GAAP for investment property. ... 11

3.4 Dividend policies in Germany ... 11

3.5 Hypotheses ... 12

4 Research design ... 15

4.1 Sample selection ... 15

4.2 Benchmark Equation and Dividend policy ... 16

5 Results ... 19

5.1 Descriptive Statistics and correlationmatrix ... 19

5.1.1 Descriptive Statistics and correlation matrix United Kingdom ... 19

5.1.2 Descriptive Statistics and correlation matrix Germany ... 21

5.2 Earnings persistence from fair value adjustments ... 22

5.2.1 Earnings persistence from fair value adjustments United Kingdom ... 22

5.2.1 Earnings persistence from fair value adjustments Germany ... 25

5.3 Effect of fair value adjustments on dividend policy ... 27

5.3.1 Effect of fair value adjustments on dividend policy United Kingdom ... 27

5.3.2 Effect of fair value adjustments on dividend policy Germany ... 32

5.4 Comparasion between the United Kingdom and Germany ... 36

6 Conclusion and Discussion ... 37

Reference list ... 39

(5)

1 Introduction

The US Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) decided in the 1980s that fair value accounting should be the standard instead of historical cost accounting; so market-based measures should be used rather than cost-based measures (Hitz, 2007, p. 324). This means that assets and liabilities on the balance sheet are valued at fair value and changes in fair value are recognized as gains and losses in the income statement (Laux & Leuz, 2009, p. 1).

However, since 1930s there has been debate over the fair value accounting (FVA), and regulators, academics, the media and various market participants take part in it and discuss what the best valuation method should be (Sikalidis et al., 2016, p. 3).

Ball (2006) stated that fair value accounting (FVA) is difficult and very costly to implement and thereby it decreases the reliability of financial reporting. In other words, you can say that FVA obstructs decision-making (Sikalidis et al., 2016, p. 3). FVA can lead to higher net incomes, because of positive fair value adjustments, and therefore net income is potentially used to calculate excessive executive bonuses and high dividends, because of the transitory components created by FVA in net income (Sikalidis et al., 2016, p. 3). Besides the above critics, there are also proponents of FVA who argues that FVA leads to a more

transparant view of the financial statements, so the users are able to make better decisions (Barth, 2007, p. 12).

After the implementation of IFRS in January 2005, there are more companies for example in Australia and the UK that use fair value measurements (Cairns et al., 2011, p. 1). Changing from national GAAP to IFRS was part of the global trend, and was the most significant change in history concerning financial reporting (Cairns et al., 2011, p. 1). The reason for this significant change was international comparability of financial statements (Cairns et al., 2011, p. 2). Because of the increased comparability of financial statements mentioned by Cairns et al. (2011), investors get better informed and therefore can make better decisions.

Concerning these changes, Goncharov and van Triest (2013) examined the impact of positive fair value adjustments on dividend policy in a sample of Russian listed firms during the period 2003-2006. They have chosen this period because in that time there was strong economic growth, before the financial crisis came (Goncharov and van Triest, 2011, p. 52). They found that dividend payouts became lower when associated fair value adjustments were positive. So they stated that the positive fair value adjustments are associated with lower

(6)

dividend payouts, instead of higher dividend payments. Thereby, they showed that fair value accounting for financial instruments does not increase dividends during the period of

economic growth (Goncharov and van Triest, 2011, p. 52).

In this study I want to implement the research of Goncharov and van Triest (2011) in the United Kingdom and Germany; I want to investigate if fair value adjustments of investment property in the United Kingdom and Germany affects the dividend policy of a firm. This leads to the following research question:

Do fair value adjustments of investment property influence the dividend policy, in the United Kingdom and in Germany?

Furthermore, I want to compare the results of both countries. This subject is interesting because it is remarkable that the regulations changed concerning the use of fair value

accounting, with the adjustments as a consequence; but there was no change in global regulation regarding the distribution of accounting profits as dividends to shareholders. In IFRS there’s nothing specific stated about the distribution of unrealized incomes from fair value adjustments on investment properties. So the local accounting rules should have regulations regarding the distribution of the unrealized incomes.

In 1956 Lintner made a model about dividend policies; this theoretical framework sets out the link between dividends and earnings components (Goncharov and van Triest, 2011, p. 52). Lintner (1956) stated in this framework that firms strive to achieve a stable and constant dividend development in association with their earnings. For this, so-called earnings

persistence, it is necessary that firms argue why their earnings are persistent (Lintner, 1956, p. 108). If firms are able to explain why their earnings from fair value adjustments are persistent, they could distribute them to their shareholders (Lintner, 1956, p. 108). But there are some disadvantages of distributing fair value adjustments to shareholders. If a firm wants to payout income that they have not yet realized, they should use other monetary sources to payout the dividends, or it might be the case that they have to borrow money to have cash to pay out the shareholders (Enria et al, 2004, p. 8). In 1996 Baker and Powell also researched dividend payouts. They did a survey and asked managers what their view is about setting dividend payments nowadays. Their answers were in accordance with the theory of Lintner from 1956, and the respondents were highly concerned about continuity of dividends (Baker and Powell, 1999, p. 33). This means that the theory is recently confirmed and we assume that it is applicable in the period of this study.

(7)

In the United Kingdom, ‘The Financial Reporting Standard’ (FRS) replaced the ‘old’ UK GAAP. The standard FRS 102 is about the distribution of profit. FRS 102 requires that fair value adjustments on investment property should be recognized in the profit and loss account. Fair value adjustments are unrealized income, because it can be classified as accounting income. In the UK it is not allowed under FRS 102 to distribute unrealized net income. This means, that the income can only be distributed to shareholders if it is converted to cash on balancesheet date. This is not the case for fair value adjustments on investment property gains, because that is only accounting income; it is a gain in the financial statements

but as yet not realized.

In Germany, the German GAAP has some requirements on the distributions of profits, in other words; dividends. The financial accounting system plays an important role in

restricting dividends to shareholders (Leuz et al., 1998, p. 112). According to the German GAAP, companies in Germany face detailed dividend restrictions mandated by the commercial code and corporate law (Leuz et al., 1998, p. 112). The calculation of net earnings is organized towards restricting dividends and protecting the debtholder and can be classified as fairly ‘conservative’. Later in section 3 I will describe in more detail the

restrictions on dividend policy in Germany.

The financial accounting in the UK can be characterized as more ‘investor-oriented’ and less ‘conservative’, because dividend restrictions are still based primarily on published accounting numbers (Leuz et al., 1998, p. 112). In prior literature, is made a distinction between dividend restrictions, stated in accounting terms (Leuz et al., 1998, p. 113). Kalay (1982) distinguishes direct dividend and indirect dividend contraints. Direct dividend contraints defines an upper bound on dividends, which can be based for example on

cumulative net earnings. Indirect dividend contraints requires stakeholders to maintain certain ratios, for instance current ratio or net tangible assets to liabilities, or minimum balances, such as net worth or working capital (Leuz et al., 1998, p. 113).

This study will be structured as follows. In chapter 2 I discuss the theory that is necessary for understanding the topic, in chapter 3 the institutional frameworks are stated, such as IFRS, German GAAP and UK GAAP. In chapter 4 the research methodology and models will be described. Thereby, I will explain the selection of the sample, the proxies and the variables. The purpose of chapter 5 is to show the results of the regression analysis. The main results will be discussed and concluded in chapter 6.

(8)

2 Theory

2.1 Investment property

In IAS 40 ‘Investment property’ it is stated how to apply the accounting for property (land and/or buildings) held to earn rentals or for capital appreciation, or both (Deloitte, 2003, IASPlus). In IAS 40 it is noted that investment properties initially are measured at cost, including transaction costs, and sometimes at fair value. To measure the investment properties, a cost model is used or a fair value model, and the changes in fair value are recognized in the profit and loss account (Deloitte, 2003, IASPlus). If once the method is chosen, it must be adopted for all investment properties of the entity. Changes are only

permitted when it creates a more appropriate presentation (IAS 40). IAS 40 emphasizes that it is not likely that a switch from the fair value model to the cost model will result in a more appropriate presentation. The fair value is equal to: ‘the amount for which the property could be exchanged between knowledgeable, willing parties in an arm’s length transaction’ (IAS 40.5). The gains or losses that come from the changes in fair value of the investment property must be recorded in the profit and loss account, in the period where the changes have occured (IAS 40.35). If a firm uses the cost model, the amount of the investment property is measured at cost less accumulated depreciation and accumulated impairment losses (IAS 40).

The fair value model causes some effects in the income statement; if the investment property is measured at fair value and the adjustments are recorded in the income statement as gains or losses.

The investment property can be held by the owner or by the leasee under financial lease. In IAS 40.8 there are some examples mentioned: land held for long-term capital appreciation, land held for a currently undetermined future use, buildings leased out under an operating lease, vacant buildings held to be leased out under an operating lease, and property that is being constructed or devloped for futrue use as investment property (IAS 40.8).

2.2 Earnings persistence

As mentioned earlier, Lintner (1956) proposed the formal relation between reported earnings and dividends. Lintner (1956) showed in the model that firms payout a certain per cent of their main earnings, which includes the premise that dividends are a function of current earnings and past dividend policy. In 1968, Fama and Babiak updated this model. In the first

(9)

place, dividends are changing in response to the changes in persistent earnings (Kormendi et al., 1996 , p. 141). They stated that the changes of transitory earnings have little or no impact on dividend payouts. Kormendi et al. (1996) exposed that there are two kinds of earnings and dividends; co-integrated or not co-integrated. Co-integrated means that the permanent part of the earnings is the main factor of dividend payouts, and other factors like tax policy and transactions costs et cetera, are of secondary importance (Kormendi et al., 1996, p. 141). If they are not co-integrated, earnings and dividends are as a consequence, also influenced by other factors than only the permanent part of the earnings (Kormendi et al., 1996, p. 141). Kormendi et al. (1996) showed that in accordance with the permanent earnings dividend model the changes in dividends in response to changes in earnings are positively related to the persistence of the change in earnings. Unfortunately, the model of Lintner has not progressed since 1956 (Kormendi et al., 1996, p. 142).

3 Institutional Framework

IFRS have not given any guidance for distributing income, generated from fair value

adjustments, to shareholders. This means that local legislation regarding earnings distribution needs to be applied by companies. For example, Russian firms are required to use market-to-market accounting for financial investments and changes in the fair values need to taken in to the profit and loss-statement (Goncharov et al., 2011, p. 54). Thereby, they need to disclose the exact amount of unrealised fair value adjustments in the footnotes.

To describe one of the situations; in Greece companies are recommending by the Greek Oversight Board to distribute only the realized profits, but it is not a requirement (Sikalidis et al., 2016, p. 4). In other words, in Greece companies can choose if they only distribute the realized part or also a part of the unrealized profits.

3.1 IFRS and UK GAAP for investment property.

Under IFRS the gains and losses arise by revaluation flow through the profit and loss account, in the period when they occur (EY, 2011, p. 33). However, under UK GAAP the changes in the value of investment properties are recognised through an additional account (EY, 2011, p. 33). This account is called the investment revaluation reserve and it can be positive as well as negative. Except when the change in value arose by a permanent decrease

(10)

in value, then the difference will be taken in the income statement. In the UK a company is able to distribute only the profits that are available for that purpose (ICAEW, 2010, p. 10).

3.2 Dividends in the United Kingdom

The part of the profits that are classified as available for distribution are the realised profits that are not previously distributed or capitalised, minus the accumulated realised losses (which were so far not written off in a reduction or reorganisation of its share capital ICAEW, 2010, p. 10). In other words, according to the ICAEW (2010) the realised losses may not be offset against the unrealised profits. Furthermore, in 2014 the Financial Reporting Council of the United Kingdom issued FRS 101, which determines the clarification of

dividend policy. FRS 101 required that only the realized profits could flow through the profit and loss account (Financial Reporting Council, FRS 101, 2014, p. 31). In Section A2.13 is stated that it is allowed in the UK to hold investment property and living animals and plants at fair value (FRC, FRS 101, 2014, p. 31). If the fair values of these assets will change, the adjustments should be recognized in the profit and loss account, and not seperately in a non-distributable reserve account (FRC, 2014, FRS 101, Section A2.14, p. 31). So the restrictions that only allow the realised profits and losses in the profit and loss account are not applied in this specific situation (FRC, 2014, FRS 101, Section A2.14, p. 31). In Section A2.15 is stated that companies are allowed to transfer the change in fair value of investment properties, living animals, plants or financial instruments to a seperate non-distributable reserve, but they are not required to do so, they can also recognise them in retained earnings. If a company presents the changes in fair value that are not classified as distributable profits in a seperate reserve, it can help with the identification of profits that are available for that purpose (FRC 2014, FRS 101, A2.15, p. 32). Furthermore, a company should note whether profits are available for distribution and these distributable profits should be determined in accordance with the applicable law purpose (FRC 2014, FRS 101, A2.16, p. 32). In other words, there is a difference between accounting profit and distributable profit; where accounting profit can be calculated by: revenues minus expenses and minus distributable profit (Leuz et al., 1998, p. 116). The distributable profit is an upper bound that was stated in 1985 by the Companies Act (CA) (Leuz et al., 1998, p. 117). The purpose of restrictions of dividends is to protect shareholders and creditors, and to realize that companies were ‘prohibited from paying dividends in excess of the surplus of net assets over the amount of the paid-up capital of the

(11)

company’ (Leuz et al., 1998, p. 117). This can be defined as: ‘the dividends that could only be paid out of profits so that the shareholders’ funds remained undiminished’ (Leuz et al., 1998, p. 117). The distributable profit can be calculated by adapting the accounting profit with regard to profits that are realised, but treated as undistributable (Leuz et al., 1998, p. 117). But, it is not exactly clear which adjustments have to be made to calculate distributable profit from accounting profit. In annual reports in the United Kingdom, the distributable profit should only be disclosed if it is necessary to give a true and fair view of the company’s status (Leuz et al., 1998, p. 119).

3.3 German GAAP for investment property.

German General Accepted Accounting Principles (German GAAP) are principle-based accounting standards and the legal requirements often lack detailed descriptions for specific accounting issues (International Accounting and reporting issues, 2006, p. 32). German GAAP has no specific guidance for investment property; such property is accounted for as property, plant and equipment (PricewaterhouseCoopers, 2010, p. 33). The property is primarily measured at acquisition cost or cost of conversion and secondarily the depreciated cost model must be applied where no revaluation of the fair value is permitted

(PricewaterhouseCoopers, 2010, p. 33).

3.4 Dividend policies in Germany

The dividends that are paid to shareholders in Germany are regulated by corporate law (Leuz et al., 1998, p. 121). Leuz et al. (1998) have explained the dividend restrictions for stock corporations and limited liability companies, which are the two main legal business forms in Germany. The Stock Act, which is called Aktiengesetz, AktG in Germany, and the

Commercial Code, which is called Handelsgesetzbuch, HGB in Germany, provided together a direct dividend constraint. The upper bound on dividends is set equal to the free reserves, in other words the retained earnings, plus the current profit adjusted for any profits or losses carried forward minus any required or statutory incrementals of the reserves (AktG 150 and 233) (Leuz et al., 1998, p. 121). It is required for stock corporations to have at least 5% of their current profits in their legal reserves unless the additional paid-in capital and the legal reserve account for at least 10% of the subscribed capital (Leuz et al., 1998, p. 121). For

(12)

distributions not in the form of dividends there are more strict requirements for German stock corporations (Leuz et al., 1998, p. 121). And thereby, the payouts out of subscribed capital and additional paid-in capital are strongly banned (AktG 57 (1) and 230).

The Limited Liability Company Act (GmbHG) and the Commercial Code (HGB) have together defined direct dividend contraints as well. The contraint has quite the same structure as the constraint mentioned above. The upper bound on dividends of a limited liability company includes the reserves, the current profit adjusted for any profits and losses carried forward (Leuz et al., 1998, p. 121). The distributions that are not in the form of dividends are again strongly prohibited, as well as for stock corporations and the payouts to shareholders out of subcribed capital are banned (Leuz et al., 1998, p. 121).

These two legal dividend restrictions are formulated in accounting terms, which do not refer to group accounts, but to individual accounts (Leuz et al., 1998, p. 121). Because of this link, there is a strong accent on the establishment of distributable profit and in turn on the prudence principle in German accounting regulation. These financial accounting rules are set by the legislature and stated in the Commercial Code (Leuz et al., 1998, p. 122). These rules are supplemented by court decisions, scholars and auditors. Leuz et al. (1998) mentioned that the restrictions on dividends valuation and recognition criteria are sharply impacted by this objective (Leuz et al., 1998, p. 122). The main driver of the German accounting and dividend regulations is the goal to prevent creditors from avoidable damage.

3.5 Hypotheses

As remarked earlier, Sikalidis and Leventis (2016) examined whether fair value adjustments had an influence on the dividend policy in Greece. In this research, I want to use the same hypothesis for the United Kingdom and Germany instead of for Greece. They defined this hypothesis:

Hypothesis 1: Unrealized income from fair value adjustments of investment properties is

persistent and therefore has an impact on dividend policy

Sikalidis et al. (2016) stated that Hitz (2007) found a trend in the changes in fair values for investment properties, because the current year fair values form the basis for the next year’s expected fair value income (Sikalidis et al., 2016, p. 11). According to Hitz (2007), this is the case when the company has private information related to its prospective cash flows, which

(13)

causes that income from fair value adjustments to be seen as persistent to a certain degree (Sikalidis et al., 2016, p. 11). This is also mentioned by Goncharov and van Triest (2011), as they state that fair value reporting increases the transparancy and decision relevance of accounting, because fair value reflects present economic conditions and incorporates market expectations about future cash flows (p. 53). The disadvantage of fair value they explain is the transitory components in the income statement that are caused by fair value accounting. These transitory components may increase the volatility of aggregate income and decrease the ability of managers and investors to monitor the long-run performance (Goncharov et al., 2011, p. 53). The long-run performance forms the basis for the dividend payouts, so if they are unable to monitor the long-run performance well, problems could occur with the

determination of dividends. However, Sikalidis et al. (2016) mentioned: ‘compared with the situation for financial securities, fair value changes are more permanent because they are based on contracts and general market conditions, which are less volatile (Christensen & Nikolaev, 2013; Geltner & Mei, 1995; Owusu-Ansah & Yeoh, 2006)’. Therefore, I expect that the parts of income that are persistent have an influence on the dividend policy of companies in the United Kingdom and Germany. This expectation is in line with the theory of Lintner (1956), who stated that when income is persistent it will have influence on the dividend policy. Also Goncharov and van Triest (2011) have found a significant relationship between the persistent unrealized income and dividend policy for Russia, namely a negative

relationship. Firms that report positive fair value adjustments in their financial statements, have significantly lower increase in dividend payouts than firms that did not report fair value adjustments (Goncharov and van Triest, 2011, p. 64). So, I suggest that in the case of the United Kingdom and Germany there is also a relationship, which could be negative or positive. Not all relationships in all countries are the same, because there are differences between countries. For instance, differences in goverance structures between different countries (Gugler et al., 2003, p. 732). In continental Europe for example, a structure of concentrated ownership is the separating characteristic, and the corporate law again plays a minor role (Gugler et al., 2003, p. 732). In that kind of situation the shareholder-manager conflict is not prominent, because of the ample incentives that large shareholders have, and because they are able to control management. The shareholder-manager conflict can also be classified as the principle-agent dilemma, described by the agency theory (Eisenhardt, 1989). To describe shortly, this theory is about the shareholder (principle) who have money, and the manager (agent) who have to invest or use this money to run the business and create more money. The principle wants to know if the manager acts in the interest of the shareholders

(14)

and not only in his own interest by taking too many risks. In continental Europe this problem is according to Gugler et al. (2003) less prominent because of the ample incentives that large shareholders have. So this is also the structure Germany and the United Kingdom have. The market of Germany exists namely of large shareholders (Gugler et al., 2003, p. 734).

Furthermore, in the current study the income persistence and dividend payouts will be tested in relation with the borrowing capacity of companies. Enria et al. (2004) found that borrowing capacity at a corporate level will decrease when companies distribute their unrealized income, which causes financial distress for some firms. The decreased borrowing capacity leads to a decline in corporate flexibility to undertake investment projects because of debt covenants and therefore the chance of default will increase according to Farinha (2003). Companies can opportunistically use fair value adjustments from investment properties and financial investment to show debtholders that their liquidity is good and so they hope the costs of debt should decline or avoid debt covenants violation (Shivakumar, 2013; Sikalidis et al., 2016, p. 7). But if companies do this, shareholders are expecting a minimum dividend payout because they also read the numbers in the profit and loss account (Shivakumar, 2013; Sikalidis et al., 2016, p. 7). This supposition of shareholders can lead to higher dividend payouts and this means that debtholders face higher costs of debt. According to Leuz (1998) a solution for this shareholders and debtholders issue is to restrict dividend payouts. In the United Kingdom unrealized earnings are not distributable, so only the persistent parts of the income can be distributed (UK GAAP). In other words, the dividend constraints are indirect contractly and supplement the legal direct dividend constraints, whereas in Germany direct dividend constrains are almost exclusively legal (Leuz et al., 1998, p. 127). The accounting in Germany can be classified as convervative. I expect that firms with a superior borrowing capacity are more prone to distributing persistent fair value income in comparison with firms with inferior borrowing capacity (Sikalidis et al., 2016, p. 7). Sikalidis et al. (2016)

mentioned that this would lead to companies with increasing capital restrictions and greater probability of default. So, the second hypothesis is:

Hypothesis 2: The positive relationship between dividend payouts and persistent

income from fair value adjustments is expected to be more pronounced among firms with a greater borrowing capacity.

(15)

Subsequently, the third hypothesis is about this corporate governance structure, and the research of Mitton (2004) is used to develop this hypothesis. Mitton (2004) found that firms with better corporate governance have higher dividend payouts, consistent with agency models of dividends. And firms that have a better corporate governance system have

therefore a stronger negative relationship between dividend payouts and growth opportunities (Mitton, 2004, p. 409). This is because that the amount firms paid out to their shareholders could not be used to invest in new projects, so the growth opportunities will decrease, while dividends are increasing. Besides this Mitton found that better corporate governance leads to more profitable firms and that is an explanation for the higher dividend payouts. So,

according to Mitton (2004) corporate governance and dividend payouts are related to each other, so I expect that when dividend payouts will increase and income from fair value adjustments is persistent, the firms should have a better corporate governance system. Formally stated, the second hypothesis therefore is:

Hypothesis 3: The positive relationship between dividend payouts and persistent income

from fair value adjustments is expected to be more pronounced among firms with a better corporate governance.

4 Research design

4.1 Sample selection

I examine the effect of fair value adjustments on dividend decisions for listed companies in the United Kingdom and Germany during the period of 2006-2010. This period is chosen because since 2005 listed companies are mandated to provide their financial statements in accordance with IFRS, so the data after this year is useful for this investigation. Through datastream I collected the data for United Kingdom companies that were listed on the Financial Times Stock Exchange Index, one of the most important indicators for the stockmarket in London. The data for German companies are also collected through

datastream, and I picked the companies that are listed on the German stockmarket. For the UK I started with 4456 firm years observations, and after excluding companies that did not have sufficient data, 1851 firm years observations remained. For Germany, the sample

(16)

existed of 2188 firm years observations and after excluding companies that did not have sufficient data, there remained 915 firm years observations.

4.2 Benchmark Equation and Dividend policy

To test the hypotheses that are stated above, I want to use an adaption of the Lintner (1956) model, which was used by Gugler and Yurtoglu (2003) and Correia da Silva et al. (2004). The model for the regression analysis is:

Where ∆𝐷!" is the current change in dividends (D) from the year t-1 to year t. The year is indicated with a t and the firms are indicated with an i. The variable 𝐸_!" is the distributable earnings of the firms for year t. The 𝑎_! is a constant term, and 𝜀_!" specifies the error term. Furthermore, 𝛽! and 𝛽! are model coefficients.

It is necessary to test the persistence of fair value income components, because this is a requirement for all of the three hypotheses. The research of Sloan (1996) and the study of Goncharov and van Triest (2011) form the basis to test this requirement, by the following earnings persistence regression:

In this model, 𝑅𝑂𝐴_𝐵𝐹𝑉_!,!!! is the ROA before fair value adjustments from investment for year t+1, and 𝑅𝑂𝐴_𝐵𝐹𝑉_!,! for year t, INVPR_REV is the fair value revaluation adjustment of property (Sikalidis et al., 2016, p. 9). Total assets scale all the figures, to create a situation where it is possible to compare the results. And all the variables are described in the Appendix below. If fair value adjustments predict future income, they should consist of persistent components and as a consequence they should be part of the distributable earnings (Lintner, 1956). So, the variable 𝑎_!, adjustment coefficient, should be different from zero. In case of a positive coefficient, the fair value adjustment signals good news about future profitability, and a decline in future earnings is conducted in negative fair value adjustments.

(17)

The fair value income components can be differentiated into two categories, notably postive and negative fair value adjustments, which are shown in the following model:

Model 3 examines whether an assymmetrical effect exists between positive and negative fair value adjustments on future earnings (Sikalidis et al., 2016, p. 9). In the formula,

𝐼𝑁𝑉𝑃𝑅_𝑅𝐸𝑉_!,!!_and_{𝐼𝑁𝑉𝑃𝑅_𝑅𝐸𝑉}

!,!! stands for respectively negative and positve fair value adjustments of investment property over total assets.

Furthermore, in order to test the hypotheses, I want to examine whether fair value

adjustments affect dividend payouts. 𝐷𝐷𝐼𝐹𝐹!,! stands for the difference of total divididends between year t and year t-1 and 𝐷_!,!!! is the lagged dividend over total assets.

In the above model, INVPR_REV is defined the same as the previous models. ROA_BFV and lagged ROA_BFV are proxies for current and past profitability, SIZE is the logaritm of the book value of assets and is a proxy for firm size; DEBT is the debt-to-asset ratio and is a proxy for leverage; and CASH is the cash-to-asset ratio and is a proxy for free cash flow. (Sikalidis et al., 2016, p. 9). For the GROWTH variable I use the ratio of sales to the book value of assets The control variables in this model are the following: 𝑅𝑂𝐴_𝐵𝐹𝑉!,!!!, SIZE, DEBT, CASH, and GROWTH. The error term in this formula are clustered by the firms (Sikalidis et al., 2016, p. 9).

To examine the borrowing capacity, I use median split variables to distinguish firms from each other (Sikalidis et al., 2016, p. 10). Firms with a coverage ratio of financial expenses

(18)

above the median take the value 1 and if the coverage ratio is below the median the firms take 0. The DHighCOV and DLowCOV are used to permit the possibility that the link between changes in dividends and fair value adjustments is not symmetric for high and low levels of borrowing capacity. This method was also used in 2001 by Nissim and Ziv (Sikalidis et al., 2016, p. 10). I want to operationalize borrowing capacity by applying the financial expenses on debt coverage ratio, which is operating profit over financial expenses on debt according to Whited (1992), for two reasons. Dichev and Skinner (2002) mentioned that the leverage variable is a quite noisy proxy for closeness to debt covenants (Sikalidis et al., 2016, p. 10). And thereby, previous studies conclude that debt covenants most commonly relate to coverage of financial expenses (Sikalidis et al., 2016, p. 10).

To test the hypothesis about the corporate governance of a firm, I use model 6, where the dummy variable DUnqual and DQual are added. If a firm received an unqualified audit opinion, the value of the dummy variable is 1 and otherwise the value is 0 by a qualified report or a firm that is not audited the year. This proxy is used in prior literature because Price et al. (2011) found that the opinion of auditors is closely related to corporate

governance activities. In accordance with the second hypothesis, I expect that firms with a better corporate governance system, thus firms that got an unqualified opinion from their auditor, will be more pronounced to include the unrealized gains from fair value adjustments of investment property in their dividend payouts. In other words, for that reason the

(19)

5 Results

In this part I want to show and discuss the results of the research. I start with the descriptive statistics of the sample. Subsequently the results of the main analysis will be discussed, the effect of fair value adjustments on the dividend policy. Each paragraph is divided into two parts; the first section is about results of firms in the United Kingdom and the second section is about the results for firms in Germany.

5.1 Descriptive Statistics and correlationmatrix

5.1.1 Descriptive Statistics and correlation matrix United Kingdom

The descriptive statistics in table 1 belong to the sample of the United Kingdom

The mean of the difference in dividends between year t-1 and t (DDIFF) is negative (-0.00065), tell us that sometimes the dividends decreased in the year relative to the previous year. However, the median (0.00021) of the difference in dividends is positive, which means that most common dividends are positive. The mean (0.73860) and the median (0.54314) of the return on assets before fair value adjustments (ROA_BFVt) are reasonably close to each

Table 1. Descriptive Statistics United Kingdom

MEAN MEDIAN St. Dev. MINIMUM MAXIMUM P25 P75

DDIFF -,00065 ,00021 ,08978 -3,92407 ,63196 ,00000 ,00325 ROA_BFVt ,73860 ,54314 ,78688 -,99902 4,85299 ,07601 1,12427 Dt ,0301 ,0188 ,11198 0,00 5,01 ,0082 ,0329 SIZE 13,49 13,22 1,731 9 20 12,22 14,43 ROA ,73942 ,54335 ,78560 -,99902 4,85299 ,06935 1,12427 CASH ,07979 ,04526 ,10283 -,00019 ,87165 ,01689 ,10131 GROWTH -,2323 ,0152 10,93451 -499,94 41,97 -,0146 ,1469 INVPR_REV -,00082 ,00000 ,03174 -,49565 ,20178 ,00000 ,00000 DEBT ,19861 ,15708 ,19425 ,00000 1,67237 ,04361 ,30285

(20)

other and both positive numbers, so there is not a broad distribution of the observations. The mean and median of debt (DEBT) and cash (CASH) are quite close to each other,

respectively: 0.19861 and 0.15708 for debt, 0.07979 and 0.02385 for cash. For growth opportunities (GROWTH) the mean is negative (-0.23230) and the median is positive (0.01520), which means that there are companies in the sample that have strong decreases in their sales relative to the previous year, so the mean will be negative. However, the most common number (the median) is a positive number. The negative mean for growth opportunities and the low median could be caused by the financial crisis that occurred in 2008, because the observations are from the period of 2006 till the end of 2010. The value of fair value adjustments on investment property are quite the same for the median and the mean, the only difference is the negative number for the mean (-0.00082) and the positive value for the median (0.00000). That the median is zero, means that in the most firm years there were no fair value adjustments, which could be caused by their valuation method (historical cost pricing) or simply that they had no adjustments in fair value for that year. Furthermore, table 2 below presents the correlations between the variables for the United Kingdom.

Table 2. Correlations United Kingdom

Variables DDIFF Dt-1 ROA SIZE CASH DEBT GROWTH ROA_BFVt INVPR_REV

DDIFF 1 -0.191** -0.002 -0.014 0.037 -0.085** 0.006 -0.002 0 Dt-1 -0.191** 1 0.02 0.445** -0.058* 0.067** 0.006 0.02 0.003 ROA -0.002 0.02 1 -0.067** 0.213** 0.042* 0.041* 0.999** 0.02 SIZE -0.014 0.445** -0.067** 1 -0.160** 0.247** 0.028 -0.067** -0.013 CASH 0.037 -0.058* 0.213** -0.160** 1 -0.121** 0.014 0.213** -0.014 DEBT -0.085** 0.067** 0.042* 0.247** -0.121** 1 0.023 0.039 -0.070** GROWTH 0.006 0.006 0.041* 0.028 0.014 0.023 1 0.041* 0.001 ROA_BFVt -0.002 0.02 0.999** -0.067** 0.213** 0.039 0.041* 1 0.060** INVPR_REV 0 0.003 0.02 -0.013 -0.014 -0.070** 0.001 0.060** 1

** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

(21)

5.1.2 Descriptive Statistics and correlation matrix Germany

The descriptive statistics in table 3 belong to the sample of Germany.

In table 3 we can find a positive mean for the difference in dividends between year t-1 and year t (DDIFF) with a value of 0.01, and the median is also positive with a value of 0.00. This means that the diffences in dividends between the firm years are very small and in general positive so the dividends are in most cases increased during the years 2006-2010. The mean as well as the median for income before fair value adjustments over assets (ROA_BFVt)are

quite high, respectively 1.11 and 0.930. The distribution of the size is not very broad, which means that the sizes of the firms in the sample are more or less close to each other. The values of the mean (0.320) and the median (0.010) of the dividends paid out (Dt) differs a bit

from each other. This means that the observations are broader distributed, which includes in this case that some firms have big dividend payouts that cause a higher mean, but that most of the firms have a lower dividend payout which causes the lower median. The broader

distribution of observations also applies to the amount cash (CASH) firms have, the mean (0.320) is slightly larger than the median (0.080). Therefore, some firms have a high amount of cash relatively to other firms which causes the mean to be slightly higher than the most common observations (the median). The mean and the median for growth opportunities (GROWTH) are the both the same, positive, amount (0.050). The median for fair value revaluations on investment property (INVPR_REV) is 0, which means that these firms have

Table 3. Descriptive Statistics Germany

MEAN MEDIAN St. Dev. MINIMUM MAXIMUM P25 P75

DDIFF 0.01 0.00 0.293 -1.00 8.00 0.00 0.00 ROA_BFVt 1.11 0.930 1.596 0.00 28.00 0,6 1.35 Dt 0.320 0.010 4.517 0.00 73.00 0.00 0.02 SIZE 13.530 13.240 2.354 3.00 21.00 11.91 14.79 ROA 1.04 0.930 0.814 0.00 12.00 0.600 1.35 CASH 0.520 0.080 13.321 0.00 451.00 0.03 0.17 GROWTH 0.050 0.050 0.489 -9.00 7.00 -0.02 0.13 INVPR_REV _0.070 ₀ _0.991 _0.00 _17.00 _0.00 _0.00 DEBT 0.470 0.170 3.914 0.00 63.00 0.04 0.32

(22)

no fair value adjustments for that year. This could be caused by the use of another valuation system like the valuation system that uses the historical cost price. The mean of the fair value revaluations (INVPR_REV) is 0.070. The distribution of observations for debt (DEBT) is quite broad, and the mean (0.470) is a bit higher than the median (0.170), so in other words there are some firms that have a high amount of debt which lead to a higher mean but the most firms have a lower amount which results in the lower median.

Furthermore, table 4 illustrates the correlationmatrix for the sample of Germany.

5.2 Earnings persistence from fair value adjustments

5.2.1 Earnings persistence from fair value adjustments United Kingdom

In this subsection I want to test the earnings persistence of fair value adjustments of investment property in the United Kingdom. If the fair value adjustments of investment properties are persistent, they should be, according to the above theory, included in the dividends that are paid by a company. For the first hypothesis I want to test, earnings persistence is a condition. For the regression, I regress future net income for the next year before fair value adjusments on net current income before fair value adjustments and current fair value adjustments for investment properties (formula 2). Table 5 below shows the results for the United Kingdom.

Table 4. Correlations Germany

Variables DDIFF Dt-1 ROA SIZE CASH DEBT GROWTH ROA_BFVt INVPR_REV

DDIFF 1 0.608*** 0.359*** -0.053 -0.04 0.628*** -0.111*** 0.595** 0.583*** Dt-1 0.608*** 1 0.553*** -0.096*** -0.063* 0.998*** 0.145*** 0.888*** 0.997*** ROA 0.359*** 0.553*** 1 -0.231*** -0.039 0.548*** 0.297*** 0.766*** 0.560*** SIZE -0.053 -0.096*** -0.231*** 1 -0.135** -0.078*** -0.057* -0.171*** -0.092*** CASH -0.04 -0,063* -0.039 -0.135*** 1 -0.004 0.024 -0.023 -0.003 DEBT 0.628*** 0.998*** 0.548*** -0.078*** -0.004 1 0.151*** 0.888*** 0.997*** GROWTH -0.111*** 0.145*** 0.297*** -0.057* 0.024 0.151*** 1 0.305*** 0.178*** ROA_BFVt 0.595*** 0.888*** 0.766*** -0.171*** -0.023 0.888*** 0.305*** 1 0.886*** INVPR_REV 0.583*** 0.997*** 0.560*** -0.092*** -0.003 0.997*** 0.178*** 0.886*** 1

*** Correlation is significant at the 0.01 level (2-tailed). ** Correlation is significant at the 0.05 level (2-tailed). * Correlation is significant at the 0.10 level (2-tailed).

(23)

Table 5 shows that for the whole sample and for the sample with only the firms with fair value adjustments on investment property the future income and the dividends (0.820 and -0.698) are negatively related. My expectation was that the coefficient for the revaluations (INVPR_REV) should be different from zero, which is the case. And when the value of the coefficient is positive it will give a good signal about future profitability, which is not the case according the results because there is a negative relationship between next year profits (ROA_BFVt+1) and revulations (INVPR_REV). Furthermore, from these results we can state that the IFRS adoption in 2005 has a certain influence on the persistence of net income for the years 2006-2010. However, this influence is negative, so the income will decrease in general by fair value adjustments. In the United Kingdom, the authorities decided to not include the fair value adjustments, or in other words the unrealized net income. This concurs with the theory of Lintner (1956), who stated that when earnings are persistent they could be include in the dividends that are paid. Therefore the first hypothesis; ‘Unrealized income from fair value adjustments of investment properties is persistent and therefore has an impact on dividend policy’ is supported for the United Kingdom by the results of the regression. There is a negative relationship according to these results, which is parallel to the results of Goncharov and van Triest (2011), because they also found a negative relationship for Russia (p. 58). Sikalidis and Leventis (2016) found for Greece a positive relationship, which is not the same as the results of this study. However, for Greece the fair value adjustments are also persistent for investment property, unless the Greek companies are not allowed to distribute unrealized profits (Sikalidis, et al., 2016, p. 18).

Table 5. Persistence of fair value adjustments of investment property (United Kingdom)

Dependent variable ROA_BFVt+1

Sample All firms Only firms with FVA

Explanatory variables Cf. Cf. ROA_BFVt .963*** .785*** INVPR_REVt -.820*** -.698*** Intercept .015** -.008 R2 .945 .511 N 1851 131

*** indicate significance at the 1% level ** indicate significance at the 5% level

(24)

Furthermore, in formula 3 the revaluations on investment property (INVPR_REVt) are split in negative and positive revaluations (𝐼𝑁𝑉𝑃𝑅_𝑅𝐸𝑉!_{and 𝐼𝑁𝑉𝑃𝑅_𝑅𝐸𝑉}!_{). Table 6 shows the} results of the regression with the splitt revaluations on investment property.

These results are the assessment of whether the effect on future earnings is asymmetrical between positive and negative fair value adjustments. The results for the whole sample are significant, which means that both positive and negative fair value adjustments on investment property are persistent. For the sample with only firms that report fair value adjustments the positive fair value adjustments are not persistent, and the negative fair value adjustments are again significant. Goncharov and van Triest (2011) as well as Sikalidis and Leventis (2016) found both a significant result for positive and negative fair value adjustments. In other words they also found that the unrealized income for positive and negative fair value adjustments is persistent. This is also in accordance with the theory of Lintner (1956) who stated that persistent income from fair value adjustments would have impact on the dividends that are paid out. From the results of the whole sample we can conclude that hypothesis 1 also is supported when the revaluations are splitt in positive and negative adjustments. In the United Kingdom it is not allowed to distribute this income to shareholders, unless it is recommended by the research of Lintner (1956).

Table 6. Persistence of fair value adjustments of investment property, splitt +/- INVPR_REV (United Kingdom)

Explanatory variables Cf. Cf. ROA_BFVt .962*** .820*** INVPR_REVt + -.975*** -.255 INVPR_REVt - -.781*** -.889*** Intercept .015** -.035* R2 .945 .527 N 1851 131

*** indicate significance at the 1% level ** indicate significance at the 5% level * indicate significance at the 10% level

(25)

5.2.1 Earnings persistence from fair value adjustments Germany

In table 7 the results of the regression that tests the earnings persistence are presented for the sample with firms from Germany.

Following my expectations, the coeffients should be different from zero, which is the case in this situation. The coefficients for the whole sample are significant and positive, which suggests that the revaluations of investment property (INVPR_REVt) certainly have an effect on the next year income before fair value adjustments (ROA_BFVt+1), as well as the current years income before fair value adjustments (ROA_BFVt). In other words, fair value

adjustments of investment property are a persistent component of the income of a company. The positive value of the coefficient of fair value adjustments (0.243) leads to an increase in future income, and this can be taken as a good signal about future profitability. According to the Lintner (1956) model, the persistent part of net income should be available for

distribution to shareholders. In Germany this is possible, providing that the requirements for distributing unrealized gains are met. The results of the regression with the sample that only consists of firms that have reported revaluations, the coefficients are not significant. So we could not conclude for this sample that the earnings from fair value adjusments are persistent, and so they could not included in the dividend payouts. The first hypothesis is supported by the whole sample, so we can state that unrealized income from fair value adjustments of investment properties are persistent and therefore have an impact on the dividend policy for firms in Germany. This result is comparable with the result of Sikalidis and Leventis (2016),

Table 7. Persistence of fair value adjustments of investment property (Germany)

Explanatory variables Cf. Cf. ROA_BFVt 0.807*** -0.493 INVPR_REVt 0.243*** 2.164 Intercept 0.178*** 0.466*** R2 0.904 0.937 N 915 38

(26)

who also found earnings persistence of the revaluation gains on investment property in Greece.

Furthermore, in model 3 the fair value adjustments on investment property are splitt up in two components; the positive revaluations and the negative revaluations. The results of this

regression for Germany are shown in table 8.

The revaluations are splitt up to check whether the effect on future earnings is asymmetrical between negative and positive fair value adjustments. For Germany, the results for the whole sample are for revaluations (INVPR_REVt+) are significant, and positive. So the positive fair value adjustsments are persistent and in accordance with Lintner’s theory (1956) and thus the revaluations could therefore be included in income that are available for distribution. The coefficient of the negative fair value adjustments (INVPR_REVt-) is not significant for the whole sample. For the sample that only consists of firms that reported fair value adjustments, the coefficient of the negative revaluations is not significant either. However, the coefficient of the positive fair value adjustments is again significant, so this part is also persistent and is available to distribute according to Lintner (1956).

Table 8. Persistence of fair value adjustments of investment property,

splitt +/- INVPR_REV (Germany)

Explanatory variables Cf. Cf. ROA_BFVt 0.811*** -0.457 INVPR_REVt + 0.237*** 2.119*** INVPR_REVt - 0.00009 0.00008 Intercept 0.175*** 0.322 R2 0.904 0.936 N 915 38

(27)

5.3 Effect of fair value adjustments on dividend policy

5.3.1 Effect of fair value adjustments on dividend policy United Kingdom

I initially assess whether the fair value adjustments of investment property influences the dividends that are paid out. In Table 5 below are the results of the regression of model 3 presented. The regression is run for the whole sample and once for firms that reported fair value adjustments. The numbers that are on the same row of the variables are the coefficients and the numbers in the parenthesis are the standard deviations.

The coefficient for fair value adjustments (INVPR_REV) for all firms is -0.028 and this result is not significant. For the sample that consists only of firms with revaluations the coefficient has a value of 0.02, which is significant. The relation is positive, which means that if a firm reports larger fair value adjustments, the difference in dividends will also be larger.

Table 9. Dividend policy and fair value adjustments (United Kingdom) DDIFF

Explanatory variables Cf. Cf. INVPR_REV -0.028 0.02* (0.072) (0.012) ROA_BFVt 0,001 -0.023*** (0.012) (0.009) ROA_BFVt-1 0 0.011* (0.012) (0.006) Dt-1 0 0** (0) (0) SIZE 0.007*** 0.002* (0.002) (0.001) DEBT -0.055*** -0.003 (0.012) (0.006) CASH 0.035 0.010 (0.024) 0.016 GROWTH 0.00004 0 (0) 0 Intercept -0.084*** -0.020 (0.021) 0.013 R2 0.055 0.104 N 1851 131

(28)

This is not in accordance with the results of Goncharov and van Triest (2011) for Russia, because they found a negative relationship in their study between fair value adjustments and dividend payouts.

The impact on firms with revaluation is significant but very small, and for the whole sample the results are not significant for revaluations. This means that the first hypothesis is supported for firms with revaluations of investment property. The revaluations have a certain effect on the dividend policy, because there is a significant positive relationship between fair value adjustments and the dividend policy. The positive relationship means the bigger the unrealized income from investment property, the bigger the dividends will be. The significant coefficient of the small sample group is small (0.02), so the dividends would not increase very much but still a little bit when unrealized income will become larger.

Table 10 illustrates the results of the fifth model of this study, where the revaluations are splitt up in interaction terms with high or low coverage ratio. The first column is for the whole sample and the second column is for the firms with only fair value adjustments.

(29)

When I run the regression for the whole sample, only Dt-1, SIZE, DEBT and the intercept are significant, and the coefficients of the interactions of the fair value adjustments and the high and low coverage ratios of financial expenses prove not significant. When the regression is runned for only the firms with fair value adjustments, the coefficients of the interactions between revaluations and high coverage ratio of financial expenses (DHighCOV*INVPR) are significant, and have a value of 0.021. The revaluations multiplied by low coverage ratio firms (DLowCOV*INVPR) are again not significant, and proved insignificant for the whole sample. I expected to find a positive relationship between dividend payouts and persistent income from fair value adjustments, which I expected to be more pronounced among firms with a greater borrowing capacity. That is to say, firms with a superior borrowing capacity are more likely to distributing persistent fair value income in comparison for firms with inferior borrowing capacity. For the whole sample there is indeed a positive relationship, but

Table 10. Dividend policy and fair value adjustments, dummy COV (United Kingdom) DDIFF

Explanatory variables Cf. Cf. DHighCOV*INVPR_REV -0.029 0.021* (0.070) (0.012) DLowCOV*INVPR_REV 2.061 0.527 (5.474) (0.630) ROA_BFVt 0.001 -0.024*** (0.012) (0.009) ROA_BFVt-1 0.000 0.011* (0.012) (0.006) Dt-1 0.000*** 0.000** (0.000) (0.000) SIZE 0.007*** 0.002** (0.002) (0.001) DEBT -0.056*** -0.004 (0.012) (0.006) CASH 0.035 0.010 (0.024) (0.016) GROWTH 0.00004 0.00007 (0.000) (0.000) Intercept -0.084*** -0.021 (0.021) (0.013) R2 0.055 0.108 N 1851 131

(30)

it isn’t significant for the firms with the low coverage ratio of financial expenses while it is significant for firms with a high coverage ratio. To conclude, there is not enough evidence to find support for the second hypothesis, therefore we cannot say that firms with a superior borrowing capacity are more prone to distributing persistent fair value income in comparison for firms with inferior borrowing capacity.

In order to test hypothesis three, the regression with model 6 is run twice; once with the whole sample and once only with firms that have reported revaluations. Table 7 shows the results of the regression with the dummy variables about the corporate governance.

For the sample with all firms, there are no significant values found for the interaction term consisting of the dummy variable unqualified (qualified) opinion and investment property revaluations. For the sample that only consists of firms with revaluations, there is a

significant result found for the interaction term ‘dummy unqualified opinion and revaluation’. The relationship found is a positive one, which means that if an auditor gives an unqualified

Table 11. Dividend policy and fair value adjustments, dummy Corporate Governance (United Kingdom)

DDIFF

Explanatory variables Cf. Cf. DUnqual*INVPR_REV -0.028 0.019* (0.07) (0.012) DQual*INVPR_REV 1.363 -0.885 8.715 (1.087) ROA_BFVt 0.001 -0.023*** (0.012) (0.009) ROA_BFVt-1 0.000078 0.010* (0.012) (0.006) Dt-1 -0.00000005*** -0.0000000319** (0.000) (0.000) SIZE 0.007*** 0.002** (0.002) (0.001) DEBT -0.055*** -0.003 (0.012) (0.006) CASH 0.035 0.01 (0.024) (0.016) GROWTH 0.0000447 0.00008913 (0.000) (0.000) Intercept -0.084*** -0.02 (0.021) (0.013) R2 0.055 0.109 N 1851 131

(31)

opinion, it is likely that the dividends will increase with a small amount (0.019). The dummy variable unqualified opinion is used as a proxy for better corporate governance. My

prediction was that firms with a better corporate governance system should have higher dividend payouts, which is the case for the sample of firms with revaluations. And I expected a negative relationship between dividend payouts and growth opportunities, which is not the case for the sample with firms with revaluations, however this coefficient is not significant. For the qualified opinion there’s a negative relationship found, which is also intuitive because it is not a good sign when an auditor gives a qualified opinion so it should be likely that dividends will decrease. However the interaction term for the dummy qualified opinion is not significant. To conclude, for the whole sample and the sample of firms with revaluations, hypothesis 3 could not be supported. Not all of the coeffients are significant and thereby, the expectation of a negative growth opportunity for firms with higher dividends proves unfound. This negative relationship was expected because if firms pay out higher dividends, it means there is less cash to invest in new projects, which could lead to a growing company.

(32)

5.3.2 Effect of fair value adjustments on dividend policy Germany

The initial test of this study is the assessment whether fair value adjustments of investment property have an impact on the dividend policy. In table 13 below, the results for the sample of Germany are illustrated. First, the regression is runned for the whole sample, and secondly only for firms that have reported revaluations in their financial statements.

The coefficient of revaluations of investment property (INVPR_REV) for the whole sample is significant. There is a negative relationship, the coefficient is -0.758, which means that fair value adjustments will lead to lower dividend payouts. This corresponds with the result of Goncharov and van Triest (2011), because they also found a negative relationship for firms in Russia. The income before fair value adjustments over assets for the current year

(ROA_BFVt) also have a significant negative influence on the difference in dividends

Table 13. Dividend policy and fair value adjustments (Germany) DDIFF

Explanatory variables Cf. Cf. INVPR_REV -0.758*** -0.208 (0.102) (0.012) ROA_BFVt -0.232*** -0.550*** (0.017) (0.043) ROA_BFVt-1 0.287*** 0.605*** (0.015) (0.036) Dt-1 -0.182*** -0.428*** (0.020) (0.047) SIZE -0.001 0.058*** (0.002) (0.011) DEBT 0.423*** 0.504*** (0.020) (0.070) CASH 0.182*** 1.606*** (0.038) (0.542) GROWTH 0.071*** 0.307*** (0.012) (0.056) Intercept -0.165*** -1.194*** (0.035) (0.196) R2 0.795 0.994 N 915 38

(33)

(DDIFF). Furthermore, the return on assets before fair value adjustments for the previous year (ROA_BFVt-1) have also a significant effect, but this relationship is positive. So, if the return on assets of previous year will increase, the difference in dividends (DDIFF) between the years will also increase. The control variables DEBT, CASH, SIZE and GROWTH are significant as well. In order to answer hypothesis 1 for Germany, we can say that there are distributional consequences and therefore hypothesis 1 is supported. Thus the predictions for Germany are fulfilled, because for the whole sample unrealized income from fair value adjustments have an impact on the dividend policy.

The coefficient of revaluations (INVPR_REVt) of the sample that only consist of firms that have fair value adjustments is not significant. The return on assets for the current year (ROA_BFVt) and for the previous year (ROA_BFVt-1) have significant influence on the difference in dividends (DDIFF).

Furthermore, model 5 concerning whether the borrowing capacity of a firm have influence on the unrealized income that is distributed as dividends. The results for Germany are shown below in the table 14.

(34)

For the borrowing capacity of firms a dummy variable is used, which is in this regression is an interaction term with the revalutations of investment property. The coefficient of the interaction term for firms with a high coverage ratio (DHighCOV*INVPR_REV) is significant, and negatively related with difference in dividends (DDIFF), for the whole sample. The coefficient of interaction term between low coverage ratio and revaluations (DLowCOV*INVPR_REV) is for the whole sample and the sample with only firms with revaluations is not significant, as is the coefficient of the interaction term between high coverage ratio firms with fair value adjustments (DHighCOV*INVPR_REV) for the second sample. We can conclude for Germany that there’s no asymmetrical relationship between fair value adjustments and differences in dividends between firms with a high or low coverage ratio. Therefore, there is no support for the second hypothesis; I did not find sufficient evidence for the fact that firms with a greater borrowing capacity are more likely to have a

Table 14. Dividend policy and fair value adjustments, dummy COV (Germany) DDIFF

Explanatory variables Cf. Cf. DHighCOV*INVPR_REV -0.761*** -0.441 (0.102) (0.010) DLowCOV*INVPR_REV -1.528 -1.423 (1.294) (1.157) ROA_BFVt -0.232*** -0.551*** (0.017) (0.042) ROA_BFVt-1 0.287*** 0.600*** (0.015) (0.035) Dt-1 -0.182*** -0.435*** (0.020) 0.047 SIZE -0.001 0.057*** (0.002) (0.011) DEBT 0.424*** 0.515*** (0.020) (0.070) CASH 0.183*** 1.67*** (0.038) (0.540) GROWTH 0.071*** 0.302*** (0.012) (0.056) Intercept -0.164 -1.188*** (0.035) 0.195 R2 0.796 0.993 N 915 38

(35)

positive relationship between dividend payouts and persistent income from fair value adjustments.

Subsequently, model 6 is about testing whether the positive relationship between dividend payouts and persistent income from fair value adjustments is expected to be more pronounced among firms with a better corporate governance system. The results of model 5 for Germany are illustrated by table 15.

For the whole sample, the interaction term of revaluations with the dummy variable of an unqualified auditor’s opinion (DUnqual*INVPR_REV) is significant, and negative. The negative relationship means that when a company receives an unqualified auditor’s opinion the dividend payouts will decrease. This is not in line with the prospections, because I expected that when a firm received an unqualified auditor’s opinion, the difference in

dividends (DDIFF) would increase. The coefficient of the interaction term with the qualified

Table 15. Dividend policy and fair value adjustments, dummy Corporate Governance (Germany) DDIFF

Explanatory variables Cf. Cf. DUnqual*INVPR_REV -1.25*** -2.090 (0.109) (0.042) DQual*INVPR_REV 8.434 1.095 (5.858) (5.842) ROA_BFVt -0.180*** -0.548*** (0.017) (0.045) ROA_BFVt-1 0.237*** 0.606*** (0.016) (0.037) Dt-1 -0.086*** -0.425*** (0.023) (0.049) SIZE -0.001 0.057*** (0.002) (0.012) DEBT 0.447*** 0.501*** (0.019) (0.074) CASH 0.198*** 1.575*** (0.036) (0.574) GROWTH 0.063*** 0.307*** (0.012) (0.057) Intercept -0.176*** -1.18*** (0.034) (0.208) R2 0.801 0.993 N 915 38

(36)

opinion (DQual*INVPR_REV) is not significant. Alongside the whole sample, once again I ran a regression with only firms with revaluations. For the sample that consists of firms with revaluations; the interaction term of fair value adjustments with the dummy variable of unqualified auditor’s opinion (DUnqual*INVPR_REV) and qualified auditor’s opinion (DQual*INVPR_REV) are not significant. These results leads to rejection of the third hypothesis, because there is not enough evidence to state that the positive relationship between dividend payouts and persistent income from fair value adjustments is expected to be more pronounced among firms with a better corporate governance system.

5.4 Comparasion between the United Kingdom and Germany

The results show earnings persistence for both countries, so that should mean distribution consequences according to the theory. In the United Kingdom is it not allowed to distribute the unrealized income, and in Germany it is allowed in some circumstances. For the United Kingdom the coefficient of the fair value adjustements was not significant, and the coefficient for Germany was significant. So in Germany the revaluations of investment property have a certain influence on the difference in dividends but the relationship is negative. Thus when a firm reports fair value adjustment the dividends will not increase in the same way but it decrease.

For the United Kingdom as well as for Germany there’s not enough evidence to conclude that there exists a positive relationship between dividend payouts and persistent income from fair value adjustments for firms with a greater borrowing capacity. Only the coefficients for high coverage firms were significant, and that is insufficient evidence to conclude that the positive relationship is more pronounced for firms with a greater borrowing capacity.

The third hypothesis is again rejected for both countries, because I do not consider the results to be sufficient to prove the positive relationship between dividend payouts and

persistent income from fair value adjustments is expected to be more likely among firms with better corporate governance.