The effect of fair value accounting on the earnings response coefficient

The effect of fair value accounting on the earnings response coefficient Author: André Kip Student number: 0516821 Date and version: Course: Supervisor: December 6, 2009 - Final draft Master thesis David Veenman Master Accountancy Faculty Economics and Business University of Amsterdam

Table of contents Abstract 2 1 Introduction 3 2 Literature review 5 2.1 Fair value accounting 5 2.1.1 Shift from historical cost accounting to fair value accounting 5 2.1.2 The use of fair value accounting 8 2.1.3 When to use fair value 9 2.1.4 Pros and cons of fair value accounting 11 2.1.5 Current problems with fair value accounting 12 2.1.6 Future of accounting standards and fair value accounting 14 2.2 Earnings response coefficient 15 2.2.1 The Earnings Response Coefficient 15 2.2.2 Prior research about ERC 17 3 The effect of fair value accounting on the earnings response coefficient 22 4 Hypotheses and contribution 23 4.1 Development of hypotheses 23 4.2 Contribution 24 5 Research design and sample selection 25 5.1 Research design 26 5.2 Sample selection 26 6 Results 30 6.1 Descriptive statistics 30 6.2 Empirical results 33 6.2.1 Overall results 33 6.2.2 Results per industry 35 7 Conclusion 38 References 40 Appendix 43 1

Abstract This research focuses on the effect of fair value accounting on the earnings response coefficient. The International Financial Reporting Standards (IFRS) have been mandatory in the European Union for all listed companies since 2005. Since then, the use of fair values is also mandatory because fair value accounting is part of these standards. Fair value accounting influences earnings because more asset value and liability value fluctuations are included into earnings and earnings should now be more relevant. More relevant numbers should result because the accounting amounts better reflect the current prices compared to the historical prices. Barth, Landsman and Lang found that the accounting amounts under the International Accounting Standards (IAS) lead to more relevant accounting amounts (2008, p. 467). However, their research focused on the accounting standards as a whole and not solely on fair value accounting. The earnings response coefficient (ERC) is the measure of the relation between earnings and stock returns. It measures the informativeness of earnings in relation to the stock return. The ERC has been used in prior research to show the effect of accounting changes or the effect of asset characteristics on how the earnings (including the changes) relate to and influence the returns. For example, accounting changes affect earnings and the ERC measures if the change is reflected in the stock return. Since fair value is claimed to be more value relevant and more informative, this research examines if the earnings response coefficient changes by the use of fair values. The findings show that fair value accounting influences the earnings response coefficient. An increase in the earnings response coefficient is noticeable across almost all industries but is especially large for the finance, insurance and real estate industry. A higher ERC will show more information to the investors and will be perceived as more persistent earnings. Additionally, the value relevance of accounting amounts under fair value accounting is higher than under historical cost accounting. This also shows that fair value accounting is more informative to investors. 2

1. Introduction Since the introduction of the new International Accounting Standards (IAS), there is more emphasis on fair value. For example, more assets are being reported at fair value and fair values are also incorporated when measuring profits. Prior literature suggests that the new IAS result in accounting numbers that are more value relevant and are of a higher quality. Higher quality is described as less earnings management, more timely recognition of losses and more value relevance of accounting amounts (Barth, Landsman and Lang, 2008, p. 467). However, the effects of the IAS differ because the IAS consist not only of fair value accounting changes. The replacement of historical cost accounting by fair value accounting has been discussed in many articles and it has led to some changes in the way of accounting. The price fluctuations of certain assets and liabilities are now included into the performance measures. Therefore, the performance measures should be more informative compared to under historical cost accounting. This research examines the effect fair value accounting has on the informativeness of earnings in relation to the share price movement. The effect of fair value accounting on the earnings response coefficient (ERC) is the focus of this research. The changes caused by fair value accounting are large. The fair value is the market price a party is willing to pay for the asset or to incur for the liability (Plantin, Sapra and Shin, 2008, p. 436). If the market price cannot be observed in the market, then the fair value can be determined using a valuation model. This way, there is always a fair value even if there is no market or an illiquid market where a price cannot be stipulated. However, fair values fluctuate over time and thus lead to different amounts on the subsequent years balance sheets. These revaluations cause the results to differ even if there has not been a sale of an asset. This revaluation goes through to equity and is included in earnings (comprehensive income). Fair value accounting was implemented because the values on the balance sheets would be more informative and problems would be avoided in the future. The use of historical cost on the balance sheet created problems during the Savings and Loan crisis. Firms with large investments in stocks didn t show their true financial status and large investments were written off later. With fair values, these problems have been avoided in the future. Other benefits of fair value accounting are that the fair values give insight into the riskiness of the business, it also leads to a higher comparability between the financial statements of companies, and problems are discovered earlier compared to historical cost accounting. On the other side, fair value accounting also has its disadvantages: the possibility of manipulation 3

by managers, the pro-cyclical effect, increased volatile accounting numbers and the short-term orientation. Since financial statements with fair value use market values for valuing different assets and liabilities, the amounts on the balance sheet are more relevant compared to amounts with the historical cost price (Allen and Carletti, 2008, p. 359). Historical cost prices become irrelevant over time and since some assets and liabilities fluctuate much, the current price shows the most relevant and up-to-date price. This leads to more current information for the investors about the assets and liabilities of a firm but the investors can also compare amounts on balance sheets of different companies. Additionally, investors can see the riskiness of a business relying on balance sheets under fair value (Allen and Carletti, 2008, p. 358). If amounts fluctuate a lot from year to year, then the investor can see that the company is taking a lot of risk since the value of its assets and liabilities is very dependent on market conditions. However, in some situations (for example, an illiquid market) the market price can be hard to determine and a valuation model must be used. In these situations companies are not willing to show which model and what assumptions they used. This unwillingness reduces the comparability of the financial statements between companies but the fair value can also be more easily manipulated in these situations (Ramanna and Watts, 2007, p.4). Manipulation can be used by managers to show higher or lower earnings, depending on the situation. Fair value accounting also has a pro-cyclical effect. This means that assets (liabilities) are likely to be overstated (understated) during economic growth; during a recession assets (liabilities) will be understated (overstated). This leads to earnings that are higher in economic growth and lower in a recession (Sal Oppenheim, 2008, pp.1-2). Since the revaluations do not always lead to realized gains or losses, the performance measures are artificially volatile (Allen and Carletti, 2008, p. 359). Therefore, it might be that the cash flow from operations is considered more important by investors under fair value than under historical cost accounting. Some argue that fair value accounting also leads to a short-term focus of companies (ECB, 2004, p. 8). Since the fair value changes from year to year, the result depends on how the fair value fluctuates during the year. As a consequence, the value of the balance sheets of financial institutions would be driven by short-term fluctuations of the market that do not reflect the value of the fundamentals and the value at maturity of assets and liabilities (Allen and Carletti, 2008, p. 359). The findings of this research are showing that fair value accounting affects the earnings response coefficient in an upward direction. The change is most noticeable for the industry which uses a lot of fair values, the finance, insurance and real estate industry. A 4

higher ERC means that the returns are more dependent on the earnings and that the earnings are more informative since the expected reaction differs from the actual result and the share price reacts more heavily to the earnings. Additionally, the findings also show that fair value accounting numbers are more informative in comparison to the historical cost accounting numbers. The value relevance has increased but again is most noticeable for the finance, insurance and real estate industry. The remainder of this article is organized as follows. The next section will discuss the literature about fair value accounting and the earnings response coefficient. Section 3 will link fair value accounting and the earnings response coefficient. It will show how fair values influence the ERC. In section 4 the hypothesis will be developed and the contribution of this research will be explained. Section 5 will explain the research method and the sample selection. Section 6 will discuss the findings of the research. Finally, this research will summarize its findings and will answer the question what effect fair value accounting has on the earnings response coefficient. 2. Literature review This section will describe what is already known about fair value accounting and the earnings response coefficient. Prior research had found that fair value accounting was implemented because the use of historical cost caused problems, which were discovered at a late stage and these problems would not be as big as they were if more up to date prices were used. Prior research has found that the ERC is associated with risk, earnings persistence, growth opportunities, firm size and uncertainty. 2.1 Fair value accounting 2.1.1Shift from historical cost accounting to fair value accounting The shift away from historical cost accounting towards fair value accounting has been initiated by the Savings and Loan crisis of the 1980s in the United States of America. The Savings and Loan crisis was caused by large delayed write-offs. Some assets had become worthless but were still on the balance sheet at historical cost. These assets did not get written 5

down because banks, for example, treated available securities for sale as investments. These accounting practices avoided large immediate write-offs and the securities were still at book values on the balance sheet. However, when the companies of the securities went bankrupt, the securities had to be written down and large losses were incurred. The crisis showed that the use of historical costs was inadequate because the financial status of many companies could not be properly identified. The standard setters switched their focus more on the relevance aspect of the financial statements than on the reliability and created a framework for developing accounting standards. The use of market values was proposed and had to avoid these problems in the future (Hitz, 2007, p. 329). However, the current credit crunch shows that fair value caused whole different problems. These problems will be discussed later. At the end of the 1980s, the International Accounting Standards Committee (IASC) had developed a framework to help in developing accounting standards and guide in resolving accounting issues that are not addressed in the existing accounting standards (Alfredson et al., 2007, p. 62). The current IASB also uses this framework since 2001 and the FASB and IASB have now started to create an improved common conceptual framework (FASB, 2009). The Framework for the Preparation and Presentation of Financial Statements (The Framework) of the IASB describes the basic concepts that underlie financial statements but it is not a standard itself or does define any principles for recognition or measurement of assets and liabilities (Alfredson et al., 2007, p. 62). The Framework defines the objective of the financial reporting as: The objective of general purpose financial reporting is to provide financial information about the reporting entity that is useful to present and potential equity investors, lenders and other creditors in making decisions in their capacity as capital providers. Capital providers are the primary users of financial reporting. To accomplish the objective, financial reports should communicate information about an entity s economic resources, claims on those resources, and the transactions and other events and circumstances that change them. The degree to which that financial information is useful will depend on its qualitative characteristics. (Conceptual framework exposure draft may, 2008, p. 12) It is clear from this definition that firms must take into consideration the environment they are in and that some resources have to be valued at fair value since it better reflects the entity s 6

resources. Also, fair values are more useful to investors because the fair values would have shown the risks that were not shown under historical cost (Allen and Carletti, 2008, p. 358). The Framework also identifies the qualitative characteristics that make financial information useful and define the basic elements of financial statements and the concepts for recognizing and measuring them in financial statements (Alfredson et al., 2007, p. 62). There are four qualitative characteristics that make financial information useful according to the IASB: understandability, relevance, reliability and comparability. Information must be presented in an understandable way for users who have reasonable knowledge of business and economic activities and accounting and who are willing to study the information diligently (IAS Plus, 2009 and IASC Framework, 1989, p. 82). The relevance characteristic means that the information must meet the decisionmaking needs of the users. Relevant information influences the economic decisions of users by helping them evaluate past, present and future events or by confirming or correcting their past evaluations (IASC Framework, 1989, pp. 82-83). Two elements of relevance are materiality and timeliness. The information is material if the decision of the user is affected by the omission or misstatement of that information (Alfredson et al., 2007, p. 71 and IASC Framework, p. 83). Timeliness is also important because the information is only relevant if the information is provided to users within a time period in which the decision of the user will be affected (Alfredson et al., 2007, p. 71). Information is reliable if there are no material errors or bias and if it can be depended on by users to represent events and transactions faithfully (Alfredson et al., 2007, p. 71). The reliability characteristic also consists of multiple elements: faithful representation, substance over form, neutrality, prudence and completeness. Information must faithfully represent the transactions and other events it either purports to represent or could reasonably be expected to represent, so that the information reflects the transactions and events that have happened or are expected to happen (IASC Framework, 1989, p. 83). Substance over form means that information must reflect the economic transactions and not the legal form of the contract (IASC Framework, 1989, p. 84). The information must also be free of bias and neutral to the decision of the user because bias might influence the decisions of users (IASC Framework, 1989, p. 84). However, the preparers of the financial statements have to account for uncertainties that are related to, for example the collectivity of the receivables. Prudence refers to a degree of caution that is exercised in the exercise of judgments with reference to some uncertain areas (IASC Framework, 1989, p. 84). Finally, the information cannot be 7

reliable if it is not complete. The information must be complete but also taking into account the materiality and the cost (IASC Framework, 1989, p. 84). Comparability describes that users must be able to compare the financial statements of a company over time so that they are able to identify trends in the financial position and performance of the company. Additionally, comparability also refers to financial statements that must be comparable across different companies. It is important that companies disclose which accounting policies the company uses in the preparation of the financial statements, if there are any changes and the effects of such changes (IASC Framework, 1989, p. 85). The accounting standard setters have been focusing on these 4 characteristics for high quality accounting information. Before the Savings and Loan crisis, the accounting standards focused on the reliable information. However, during the 1980s the standard setters have shifted their focus from reliability more to relevance due to the problems of the Savings and Loan crisis. The relevance aspect became more important because it showed the true financial position a firm is in and the decision making was influenced by reporting the historical cost price. Additionally, the reliability aspect shows that the information must be free of material errors and material differences were created by the use of historical cost prices. Overall, the broader focus of the IASC let to more useful financial reports for the investors. 2.1.2 The use of fair value accounting The use of fair values instead of historical cost prices should prevent problems in the future and the use of more relevant prices would make financial reports more relevant. The FASB and the IASB developed fair value measurements for this need (Hitz, 2007, p. 329). The IASB defines fair value as: Fair value is the amount for which an asset could be exchanged, or a liability settled, between knowledgeable willing parties in an arm s length transaction (IAS 39, 2009, p. 1) The FASB defines fair value as: The exchange price is the price in an orderly transaction between market participants to sell the asset or transfer the liability in the market in which the reporting entity 8

would transact for the asset or liability, that is, the principal or most advantageous market for the asset or liability (SFAS 157, 2009) So, both definitions define the fair value as the market price another party is willing to pay for the asset or the liability to settle. The fair value can be determined in three different ways. The first method of determining the fair value is determining the price in an active and liquid market (Level 1 assets). This is an easy and reliable way for multiple assets and liabilities. For example, prices of securities are usually determined this way because stock markets are liquid and active markets. However, there are no active or liquid markets for a lot of assets and liabilities. The price should then be determined by modifying market prices of comparable items or using a mark-to-market model (Level 2). Finally, if this also is not possible, the price should be determined by using a valuation model with assumptions that are based on internal estimates and calculations (Level 3) (Hitz, 2007, p. 326). By observing the way the fair value is determined, it is clear that the relevance aspect seems more important than the reliability aspect. Especially for the assets and liabilities for which there is no active or liquid market, the price is determined in a way that is hard to verify for people outside the firm. Companies are not willing to disclose which model and what assumptions they used. This unwillingness reduces the comparability of the financial statements between companies but the fair value can also be more easily manipulated in these situations (Ramanna and Watts, 2007, p.4). Manipulation can be used by managers to show higher or lower earnings, depending on the situation. 2.1.3 When to use fair value Although fair value is mandatory in some situations or an option for the valuation of some assets and liabilities, the general idea is that all assets and liabilities have to be valued at fair value and that the changes in fair value are included in the income statement (Cairns, 2006, p. 5). However, this is not the case. Fair value accounting only has to be used in some situations. This misconception is due to the press failing to understand and report the underlying concept and that the focus on some elements of the IFRS (Cairns, 2006, p. 6). Since the IFRS are mandatory in the European Union, some assets and liabilities (mostly financial assets and liabilities) have been in the spotlight and people assume that fair value accounting applies to all assets and liabilities (Cairns, 2006, p. 6). 9

Fair value accounting has to be used at initial measurement to measure transactions and non-cash transactions (Cairns, 2006, p. 11). A transaction is an agreement between two parties to exchange goods or services in exchange cash. A non-cash transaction is an agreement to exchange goods or services in exchange for other goods or services. The cost of the assets and liabilities involved in a non-cash transaction must be determined as if it has been paid in cash or cash equivalents and thus fair value is used to determine these amounts. Fair values show the amounts that would have been paid in cash or the amount that the assets and liabilities are worth. The transaction is now represented faithfully in the financial statements and old irrelevant prices are not used. Subsequent measurement of fair values for these assets or liabilities is not required (Cairns, 2006, p. 11). A list of the required use of fair value in the initial measurement of transactions can be seen in Table 1. Fair values also have to be used to allocate the cost of compound transactions (IAS 23 and IFRS 3) over the acquired assets and liabilities (Cairns, 2006, p. 12). A good example is a business combination. The party that takes over the other party must measure all the fair values of the assets and liabilities it acquires. The price that was paid for the acquisition minus the fair values is the goodwill and will be stored separately on the balance sheet. Hence, the transaction is measured as if all assets and liabilities were acquired separately and were paid with cash and this can be disclosed in the financial statements (Cairns, 2006, pp. 11-12). Subsequent measurement of assets and liabilities at fair value has to be used for derivatives, other held for trading financial assets and financial liabilities, available-for-sale financial assets and agricultural produce at the point of harvest (Cairns, 2006, p. 16). A full list of the use of fair value in subsequent measurement (option or mandatory) can be seen in Table 2. The idea behind subsequent measurement at fair value only applies to these assets and liabilities is that the amounts should be reliably determined (Cairns, 2006, p. 9) and that certain non-cash transactions are disclosed in the financial statements (Cairns, 2006, p. 11). In other cases, one of these criteria does not apply and therefore fair value is not used to determine the amount as it might lead to unreliable and irrelevant financial information. Finally, fair value accounting has to be used for impairment testing. Assets must not be carried at an amount above the recoverable amount. The recoverable amount is the amount the entity expects to recover from the sale (fair value) or the use of the asset (Cairns, 2006, p. 16). Impairment of assets applies to a lot of assets and liabilities (IAS 2, 11, 17, 36, 39, 40, 41, IFRS 4 and 5) (Cairns, 2006, pp. 16-21 and Alfredson et al., 2007, p. 551). Only the gains and losses from changes in fair value from subsequent measurement and changes in fair values caused by impairment testing have an impact on earnings. The 10

changes from subsequent measurement affect earnings because the changes go thought profit or loss or changes affect earnings when the impairment test is done or the asset is derecognized (Alfredson et al., 2007, pp. 220-226). The impairment losses are immediately recognized in profit or loss if an asset is measured using the cost model. However, if an asset is measured using the revaluation model, then the impairment loss will go to the revaluation reserve and does not affect earnings (Alfredson et al., 2007, p. 559). The changes caused by the initial measurement of fair value and the allocation of cost of compound transactions do not affect earnings since the assets and liabilities are recognized at fair value when they enter the financial statements. The fair values will be stated on the balance sheet and will be depreciated or amortized. 2.1.4 Pros and cons of fair value accounting The use of fair value accounting has its advantages and disadvantages compared to historical cost accounting. With the use of market values for valuing different assets and liabilities, the amounts on the balance sheet are more relevant compared to amounts with the historical cost price (Allen and Carletti, 2008, p. 359). Historical cost prices become irrelevant over time and since some assets and liabilities fluctuate much, the current price shows the most relevant and up-to-date price. This leads to more current information for the investors about the assets and liabilities of a firm but the investors can also compare amounts on balance sheets of different companies. Additionally, investors can see the riskiness of a business relying on balance sheets under fair value (Allen and Carletti, 2008, p. 358). If amounts fluctuate a lot from year to year, then the investor can see that the company is taking a lot of risk since the value of its assets and liabilities is very dependent on market conditions. An example of a firm where you can see the riskiness is an investment firm. The carrying amounts on the balance sheet will depend for the most part on market conditions and therefore could fluctuate much from year to year. Thus, the value of the firm depends on the market conditions. The changes will go through profit or loss and this will lead to volatile earnings. A production company will not be so risky because the value mostly depends on the value of the assets and these will not be very influenced by market conditions. In some situations (for example, an illiquid market) the market price can be hard to determine and a valuation model must be used. The unwillingness of companies to show which model and what assumptions they used reduces the comparability of the financial 11

statements between companies. Because the model and the assumptions are also not known, the manager can manipulate the amounts more easily (Ramanna and Watts, 2007, p.4). Additionally, less developed nations also have less liquid and active markets and for those countries the use of fair values might not bring more relevance into the financial statements (Ball, 2006, p. 24). The increased relevance of using fair values is limited by the political and economic environment and in an illiquid market it reduces the reliability of the amounts on the financial statements. Another disadvantage of fair value accounting is that it has a pro-cyclical effect. This means that assets (liabilities) are likely to be overstated (understated) during economic growth and during a recession assets (liabilities) will be understated (overstated). This leads to earnings that are higher in economic growth and lower in a recession (Sal Oppenheim, 2008, pp.1-2). Since the revaluations do not always lead to realized gains or losses, the performance measures are artificially volatile (Allen and Carletti, 2008, p. 359). Therefore, it might be that the cash flow from operations is considered more important by investors under fair value than under historical cost accounting. An example is volatile earnings from an investment firm. The volatility is dependent on the market values of the investments of the firm and therefore the earnings can fluctuate much from year to year. Some opponents also argue that fair value accounting leads to a short-term focus of companies. The European Central Bank has acknowledged this shortcoming. Banks are influenced the most because fair value accounting influences the amount a bank is allowed to loan. Loans are given to customers and information asymmetries are overcome, but banks are more willing to focus on short-term loans because of changing interest rates and other conditions change the fair value and thus influence earnings. Short-term loans will not change much and so the fair values and earnings will not be affected much (ECB, 2004, p. 8). Since the fair value changes from year to year, the result depends on how the fair value fluctuates during the year. As a consequence, the value of the balance sheets of financial institutions would be driven by short-term fluctuations of the market that do not reflect the value of the fundamentals and the value at maturity of assets and liabilities (Allen and Carletti, 2008, p. 359). 2.1.5 Current problems with fair value accounting The financial crisis of 2008 showed new problems with fair value accounting which had not been taken into account when fair value accounting was implemented. Critics say that the fair 12

value and the immediate recognition of losses when the market takes an excess downturn leads to an incorrect working pricing mechanism. Widened losses and a bigger capital gap are the consequence and especially financial institutions have to sell assets to meet capital requirements. This immediate reaction to the excess write-down leads to large losses and even lower prices (China Daily, 2008). Liquidity problems also led to lending at a higher interest rate. Additionally, asset-backed securities went from Level 1 to Level 3 assets because they went from liquid to illiquid and these securities had to be measured against discounted prices in the market (China Daily 2, 2008). For example, the subprime positions of banks went from Level 1 to Level 3 assets because of the increasing illiquid market of subprime positions and the high default rates of mortgages. The large write-downs also led to more uncertainty under investors. The uncertainty focused especially on the Level 3 financial assets and the pricing methods that are used by companies. A survey shows that only 3 percent of the respondents think that the valuations of banks are within 3 percent of the accurate value (Web CPA, 2009). The assumptions and models that are used are not valid because there are large write-downs. Investors want to know more about these models and assumptions on which the valuations are based but these are not required to be disclosed under fair value accounting. Since fair value accounting is pro-cyclical, it amplifies the crisis in an economic downturn and growth during an economic growth. During the growth period, the assets of banks are continuously increasing in value and this leads to additional credit provision. The banks are also looking for new investing projects and this leads to booming markets (like the house market in the USA). When the economy slows down, these markets collapse since the prices in the markets were based on excessive and overenthusiastic credit provision of banks (VKW Metena, 2008). Under historical cost accounting, these problems would not keep piling up but as history has shown there could have been other problems. The large write-downs and decreased asset prices also affect earnings. The unrealized gains and losses are included in other comprehensive income and do not show the relevant income number. Some of the losses should have been recognized earlier because the prices of the assets were not as high as estimated with the models. The income numbers during economic growth were also not reflecting the economic truth because the increases of the assets should not have been included. 13

2.1.6 Future of accounting standards and fair value accounting The International Accounting Standards Board (IASB) and the Financial Accounting Standards Board (FASB, US accounting standard setter) have started a convergence project in October 2002 (Ball, p. 11). The goal is to make the International Financial Reporting Standards (IFRS) and the US GAAP (US reporting standards) the same as far as possible across jurisdictions and also trying to improve the overall quality of the standards (Schipper, 2005, p. 102). Since the IFRS are based on principles and the US GAAP on rules, there are a lot of differences. The IFRS are more general and leave some interpretation to the user but the US GAAP are very strict and leave no standard open to interpretation. By converging these different standards, the financial statements across the globe should be more comparable and transparent (Ball, 2006, pp. 7-9). However, Ball (2006) notes that there are some difficulties in converging the two standards and that the use of IFRS all over the world might not improve the accounting quality (Ball, 2006, p. 23). The worldwide use of IFRS might bring investors more comparable financial statements and more accurate, detailed and timely financial statement information only relative to national standards (Ball, 2006, p. 11). The national standard might be poor in reflecting certain transactions, like financial instruments, and then IFRS increases the quality of the financial statement information. Nevertheless, the enforcement of IFRS is done by the national institutions and not by the IASB or some international institution. These national institutions might not enforce the use of IFRS in the correct manner and might punish the users in different ways. The adoption of IFRS will only then be perceived as a signal of quality but if the national institutions do not enforce the use of IFRS in the correct manner the higher quality of the financial statements will only be perceived on paper (Ball, 2006, p. 23). The near future will not bring any changes in the use of fair value accounting. The use of fair value will not be extended to full fair value accounting. Cairns (2006, p. 21) and Benston et al. (2006, p. 263) know of no project that will extend the use of fair value accounting. However, with the current credit crunch the IASB and the FASB will give additional guidance with regard to use of fair value accounting. Additionally, the existing standards will be adjusted so that IFRS and US GAAP will measure multiple assets and liabilities in the same way. Finally, the use of fair value will be adjusted so that the use of fair value is the same across different standards (IASB, 2009). This will make the standards easier 14

to use because the standards are based on the same principle and do not require different ways of determining fair value. 2.2 Earnings Response Coefficient 2.2.1 The Earnings Response Coefficient The earnings response coefficient (ERC) is a measure of the relation between stock returns and earnings around the time the earnings are announced. The ERC measures how much new information the earnings contain and this new information is measured by examining the influence on the stock return around the announcement. However, some researchers examine the relationship between the earnings and returns over one year. Additionally, some research also uses the ERC as a measure of the relation between the unexpected earnings and the unexpected returns. The ERC is not of particular interest for the shareholders but it is used by researchers to examine and show the effect between the earnings, which are measured in different ways, and the return. The relationship between returns and earnings is based on the idea that the stock price is a function of future expected dividends (Easton and Zmijewski, 1989, p. 118 and Collins and Kothari, 1989, p.146). The price of security i at time t may be depicted as: where P it = k=1 Ε t D it + k k τ=1 1 1 + Ε(R it + τ) Ε t D it + k = the expected dividends at time t to be received at the end of period t + k Ε R it + τ = the expected rate of return on security i from the end of t + τ 1 to the end of t + τ (Collins and Kothari, 1989, p. 146) The price of a security is a function of the expected dividends at time t and the expected rate of return at time t. The price of a security changes if investors revise their expectations of future dividends given the expected rate of return (Collins and Kothari, 1989, p. 146). To relate earnings and returns it is necessary to substitute the price formula into 15

returns. The price can be substituted in returns by dividing the formula by the beginning stock price: where P it = share price at time t RET it = P it + DIV i P it 1 ) P it 1 DIV i = received dividend per share during the year P it 1 = share price at the beginning of the year The return is a function of the price change during the year and the dividend per share divided by the price at the beginning of the year. The unexpected return, due to a revision of future earnings, can be calculated by deducting the expected return from the realized return. The expected return is calculated with the CAPM model and where the expected return is a function of the risk a company takes and the expected market rate of return and risk-free rate of return. Current earnings (X) are related to future dividends because investors revise their expectations of future earnings and their expectations of future dividends when the current earnings are announced. Easton and Zmijewski support that current earnings and expected future earnings are related. The information in the earnings announcement results in a revision of the expected earnings if the current earnings are different from what is expected (Easton and Zmijewski, 1989, p. 137). Current earnings will also affect future earnings because some earnings are permanent and some are transitory. The permanent earnings affect future earnings but the transitory earnings are only temporary and only affect this year s earnings (Freeman and Tse, 1992, p. 186). Current earnings are likely to differ from what investors expect. This unexpected part of earnings leads to a revision of future dividends and thus leads to a stock price change. The unexpected earnings can be calculated by deducting the expected earnings from the realized earnings. The expected earnings are calculated with the expected rate of return the company times the total assets of the company. Prior research sometimes did not calculate the expected earnings but took expected earnings from analysts forecasts (Imhoff Jr. and Lobo, 1992, p.430). Since investors are likely to revise their expectations of future dividends after the earnings announcement is made because the earnings differ from what is expected and the stock price reacts to this revision, it is interesting to relate the earnings to the returns. This can be estimated by using a linear model which relates the returns with the earnings. The ERC 16

(β) is the important variable in this estimation because it shows the amount of new information that the earnings contain measured by the reaction in stock return. The model will be: RET = α + β 1 EARN + β 2 ΔEARN + ε This model relates the earnings and the changes in earnings with the returns. Easton and Harris (1991) have done this and found that the earnings and the change in earnings complement each other in explaining the returns (Easton and Harris, 1991, p. 31). The ERC is depicted in this formula by β 1. 2.2.2 Prior research on the ERC So far, there has been a lot of research on the earnings response coefficient. Most of the studies examine if the ERC is associated with other variables and explain their idea about why there should be an association. Collins and Kothari (1989) examine the cross-sectional and the temporal determinants of the ERCs. They find that the temporal variation of the ERC is negatively related to the risk-free interest rate (Collins and Kothari, 1989, p.165). The riskfree interest rate influences the stock price because the expected return depends on the amount of risk the company takes and the risk-free interest rate. If the risk-free interest rate increases, then more earnings will be expected (Collins and Kothari, 1989, p. 150). However, the ERC will be negatively related to the risk-free interest rate because an increase in the risk-free interest rate, will lead to a lower ERC. Collins and Kothari also find that the ERC is negatively related to the firm s systematic risk if the β is constant through time. This is obvious because the present value of future expected dividends will be smaller if the systematic risk is smaller because the expected return is lower (Collins and Kothari, 1989, p.147). Easton and Zmijewski also show that the ERC and risk are related. They show that the ERC and the systematic risk are negatively correlated (Easton and Zmijewski, 1989, p. 133). So firms with a higher risk are having a smaller stock price reaction to new information in the earnings announcement. Finally, Collins and Kothari find that the earnings persistence and growth opportunities are positively related to the ERC. Collins and Kothari believe that the variables do not individually affect the ERC because the proxies for growth and persistence could reflect the effect of both variables (Collins and Kothari, 1989, p. 178). The earnings persistence affects the ERC because current earnings shocks that are persistent also affect 17

future dividend expectations. More persistent earnings lead to more dividends in the future and thus a revision of the earnings expectations (Collins and Kothari, 1989, p.147). The growth opportunities are related to the ERC because they represent investment opportunities that are expected to yield an above average return. Therefore, the future dividends will be higher if the investment is successful and this is supported by findings of Easton (1985) (Collins and Kothari, 1989, p. 149). Easton found a negative relation between current dividends and future dividends (Easton, 1985, pp. 74-75). If earnings are invested now, then the dividends will be lower but once the investments are completed, the operations are expected to grow and future dividends will be higher. The firm size is also found to be influencing the ERC. This has multiple reasons. Usually large firms are less risky and therefore their expected rate of return is lower and large firms also do not react heavily on new information. A revision of future earnings then will not lead to a large stock price change. Findings of Schwert support this, he finds that stock returns are negatively correlated with firm size (Schwert, 1983, p. 10). So, it is likely that the ERC is smaller for larger firms since the stock returns will not react much to the earnings announcement. Additionally, large firms release more information between earnings announcements dates and analysts forecast dates and therefore the unexpected earnings measures for large firms have more error. Hence, the measurement error bias in the ERCs is negatively correlated with the firm size (Easton and Zmijewski, 1989, pp. 133-135). Additional research does not show much evidence for a relationship between the ERC and firm size. Easton and Zmijewski find a positive relation between the ERC and firm size but the results are not significant. When they control for firm size, the prior results were not affected (Easton and Zmijewski, 1989, pp. 135-137). Collins and Kothari also examine the effect of firm size. They divide their sample into three groups by size and see if the results differ (Collins and Kothari, 1989, p. 159). Their findings suggest that the differences are due to the different information environments and that it can be controlled for by changing the holding period from 12 to 15 months for large and medium firms. Without controlling for firm size, the earnings/return association will be understated, especially for large firms (Collins and Kothari, 1989, p. 164 and 178). Easton and Zmijewski also examine the extent to which the new information in an earnings announcement leads to an expected earnings revision (revision parameter) (Easton and Zmijewski, 1989, p. 118). The revision coefficient relating the current earnings to the future earnings is positively associated with the ERC (Easton and Zmijewski, 1989, p. 137). Other researchers refer to this as the persistence of earnings. 18

Kormendi and Lipe (1987) examine if the magnitude of the effect of unexpected earnings on stock returns is positively correlated with the present value of the revisions in expected future earnings derived from a univariate time-series model (Kormendi and Lipe, 1987, p. 324). Current earnings possess information about current and future cash flows. Kormendi and Lipe think the measures are positively correlated because earnings will persist into the future and will lead to revisions in expected future earnings (Kormendi and Lipe, 1987, p. 330). Their findings suggest there is a positive correlation between the magnitude of the relation between unexpected earnings and stock returns and the revisions in expected future earnings. The derived effect of new information on stock returns equals 1 plus the present value of revisions in expected future earnings. Additionally, they find no evidence that stock prices react excessively sensitive on new information (Kormendi and Lipe, 1987, p.343). So the persistence in earnings affects the relationship between earnings and returns. Billings (1999) examines if the relation between the ERC and default risk is due to an association between bond ratings and expected earnings growth rather than an association between an element of equity risk that is not captured by the equity beta and bond ratings (Billings, 1999, p.510). Prior theory found a positive association between the ERC and expected earnings growth and a negative association with equity risk (Billings, 1999, p. 509). Billings uses two different default risk measures that are associated with expected earnings: bond ratings and debt-to-market equity ratios. Bond ratings are a risk measure because bond agencies take the expected earnings growth into consideration when the bond ratings are assigned. The expected earnings growth is important because a firm should grow in the future and debt agencies see it as a weakness if there is no or negative earnings growth (Billings, 1999, p. 513). Debt-to-equity ratios and expected earnings growth are also related. Firms with high debt leverage are not expected to have high earnings growth since the costs of debt are large and earnings are not expected to grow (Billings, 1999, p. 515). Billings finds that bond ratings do not have incremental information compared to expected earnings growth. The debtto-market equity ratios reflect both expected earnings growth and default risk in an ERC context (Billings, 1999, p. 518). Billings concludes that the ERC is mostly determined by the expected earnings growth and that the debt-to-market equity ratios or bond ratings do not show an additional element of risk that influences the ERC (Billings, 1999, p. 520). Prior research has also found that uncertainty affects firm value. Greater uncertainty about a firm s future cash flows will cause larger stock price fluctuations to value relevant information (Imhoff Jr. and Lobo, 1992, p. 428). The uncertainty about future cash flows affects the stock price because future cash flows influence the amount of dividends that will 19

be paid in the future to the equity holders. Alternatively, earnings uncertainty could be due to noise in the process generating the earnings signal. The noise is due to the implementation of IFRS because it gives managers more discretion and by using the accruals and fair values they can influence the earnings. If earnings uncertainty is due to noise, then greater uncertainty in earnings should results in smaller stock price responses to earnings news since it is manipulated (Imhoff Jr. and Lobo, 1992, pp. 428-429). However, since it is hard to examine and control if the uncertainty is due to noise or is fundamental uncertainty, Imhoff Jr. and Lobo examine how uncertainty affects the ERC. They measure the earnings uncertainty by the dispersion of analysts forecasts and show results for firms with low, middle and high uncertainty (Imhoff Jr. and Lobo, 1992, p. 429 and 432). They conclude that firms with lower uncertainty appear to have a larger reaction in stock price to the earnings announcement. So the ERC is higher for firms with lower earnings uncertainty and if the uncertainty is mostly due to noise, the ERC will be relatively small (Imhoff Jr. and Lobo, 1992, p. 437). Defond and Park (2001) examine if the effects of abnormal accruals are anticipated by the earnings response coefficient. Since the market anticipates accruals to be mean reversing, this effect should be seen in the price of the security and thus in the ERC. Abnormal working capital accruals have no or a very little effect on the lifetime earnings of a company since they are mean reversing. If they have no effect on earnings they should also not have an effect on the stock price. Defond and Park expect that the reported magnitude of earnings surprises that contain abnormal accruals to differ from the underlying magnitude that the market prices. They expect that a firm reporting good news and with income-increasing abnormal working capital accruals will have a lower ERC than if the firm reports good news with incomedecreasing abnormal working capital accruals. A firm reporting bad news with incomeincreasing abnormal working capital accruals will have a higher ERC than if the firm reported bad news with income-decreasing abnormal working capital accruals (Defond and Park, 2001, pp. 378-379). The market anticipates the mean-reversing nature of the accruals and incorporates it into the price. The abnormal working capital accruals are measured by the actual working capital accruals minus the expected working capital accruals. The expected working capital accruals are calculated by a percentage of sales in the same quarter of last year adjusted to this year s sales (Defond and Park, 2001, p. 380-381). Defond and Park find evidence that supports their expectations but conclude that the market does not fully adjusts for the pricing implications (Defond and Park, 2001, pp. 400-401). Prior research about the ERC also shows that the ERC is much higher when losses are not included into the sample. The informativeness of losses with respect to future cash flows 20