Fair value accounting and pro-cyclicality : market response to changes in fair value accounting during the global financial crisis

Amsterdam Business School

MSc Accountancy & Control, variant Control & Accountancy

Faculty of Economics and Business, University of Amsterdam

Master Thesis:

Fair value accounting and pro-cyclicality:

market response to changes in fair value

accounting during the Global Financial Crisis

Final Version

Koen Smit (10662650)

20

th

of June 2015

Supervisor: Dr. Alexandros Sikalidis

Word Count: 14.731

2

Statement of Originality

This document is written by Koen Smit who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3

Content

1. Introduction ... 5

2. Literature review and hypotheses ... 8

2.1 Fair value accounting ... 8

2.2 The events preceding the Fair value accounting amendment ... 10

2.3 Fair value accounting and the reclassification amendments ... 13

2.4 Theory ... 15

2.5 Hypotheses development... 16

3. Sample and Methodology ... 20

3.1 Sample selection ... 20

3.2 Methodology ... 20

4. Empirical findings ... 23

4.1 Stock price reaction IASB fair value accounting relaxation ... 24

4.2 Stock price reaction EU carve out threat ... 24

4.3 Stock price reaction additional analysis ... 25

4.4 Trading volume reaction ... 28

5. Conclusion ... 29

6. Reference List ... 32

7. Attachments ... 35

4

Abstract

Fair value accounting allegedly exacerbated the recent financial crisis. It has been blamed for amplifying the economic cycles. This study investigates the market response across the European financial industry amid a period in which the fair value accounting legislation was relaxed. For this purpose, seven events have been identified throughout October 2008. I have documented that investors at large responded positively to fair value accounting relaxation by the IASB. Contrary, these investors’ responded negatively at the threat of a European carve out. Additionally, the abnormal stock price reactions to events that increased the probability of a fair value accounting relaxation are strongly influenced by a firms’ risk of default, as measured by the corporate credit rating. The underlying distribution of the market responses are tested with a stratified sample based on the corporate credit rating. The stratified sample showed that the higher credit rating group responded negatively throughout the entire event period, whereas the lower credit rating group responded positively. This in turn, suggests that a pro-cyclicality effect may have been incorporated in to fair value accounting, as perceived by investors at large. Moreover, a positive abnormal trading volume has been reported throughout the entire event period. This signals that the events conveyed significant information, indicating that fair value accounting is perceived as value relevant.

This study contributes to the literature regarding the standard setter’s performance, fair value accounting and the noisy information hypothesis. First, it shows that investors at large favorably perceived the decision of the IASB to amend the fair value accounting legislation in order to avoid a clash with the European Union. Second, the results indicate that investors at large seem to believe that fair value accounting was harmful, particularly to troubled banks (lower credit rating group). Finally, the results suggest that a pro-cyclicality effect is incorporated in to fair value accounting, as predicted under the noisy information hypothesis.

5

1. Introduction

The purpose of this study is to provide empirical evidence on how investors perceived the standard setters’ amendments made to the fair value accounting and impairment legislation amid the financial crisis. My motivation for this study is to evaluate whether the claims of so called fire-sales of financial instruments were perceived to be caused by fair value accounting. This in turn allows for an evaluation of the efforts of standard setters and will contribute to the fair value accounting debate in this regard.

The benefits of fair value accounting have been extensively documented over the past decades (Barth, Beaver and Landsman, 2001). It has been empirically documented that fair value accounting provides more value relevant information to the users of financial statements (Landsman, 2007). The merits of fair value accounting lay in this provision of more value relevant information. Barth, Beaver and Landsman (2001) demonstrated this alleged superior relevance by finding that fair value accounting is more closely associated with the stock price than historic cost accounting. Despite fair value accounting’s merits, it has received considerable criticism amid and following the financial crisis.

This spark of criticism has ignited a major policy debate regarding fair-value accounting. A major issue in this debate has been the concerns regarding fair value accounting in times of market turmoil. The SEC (2008, pp. 182-184) shares these concerns, they worry that fair value accounting induces a pro-cyclical downward pressure in assets prices, leading assets prices to fall strongly below to what some believe is their fundamental value (incremental value). Fundamental value is considered to be the estimated future cash flow over the assets life. In order to off-set these write-downs, financial institutions are compelled to sell their assets in illiquid markets. These fire-sales allegedly further lowered the observed market price, thereby compelling additional sales to raise capital and therefore resulting in a downward spiral.

Other critics, (e.g. Boyer, 2007) argue that fair value accounting has contributed to the financial crisis and exacerbated its severity for the financial institutions by encouraging pro-cyclical risk taking. The underlying reasoning is that fair value accounting induces a short run valuation of the firms’ assets which is more erratic than the valuation over its complete time. He argues that fair value induces a more frequent valuation of assets and liabilities for banks, which increases the volatility of income and financial results. Resulting in a pro-cyclical pattern of risk taking, this, in turn, will trigger short-term investor behavior.

6

However, the IMF (2008, p. 108), argues that forced or fire-sales would not be valid determinants of market prices, because the accounting frameworks presume that the reporting entity is a going concern that does not need to or intent to liquidate its assets. The issue at hand is that this claim by the IMF was made while the market was illiquid market and arguably in a psychological state of distrust. During the financial crisis, many banks were unwilling or incapable of lending and obtaining credit from one another. Governments and institutions all across the world took extraordinary measures to bail out banks during the financial crisis. So arguably, the going concern state was being threatened. As an example used by Magnan (2009), Lehman Brothers and AIG both appeared solvent and properly capitalized with a major chunk of their balance sheet relying on fair value accounting. Unfortunately, both firms’ estimations didn’t adequately incorporate the downside risk they were facing if events didn’t evolve according to their expectations.

In prior literature it seems there is a lack of consensus. Especially if the extraordinary circumstances of a financial crisis are incorporated into the discussion, the consensus then becomes hard to find. A fair value related topic on which prior literature seems to reach an apparent understanding, is the fact that fair value accounting didn’t cause the global financial crisis. However, whether it exacerbated the financial crisis is fiercely debated. This in turn, is illustrated by the findings of Laux and Leuz (2009). Their study aims to ‘make sense’ of the fair value accounting debate, and they find that fair value accounting is neither responsible for the financial crisis nor that it’s merely a measurement system that reports asset values without having economic effects of its own. Most of the prior literature they refer to stems from research regarding the U.S. banking industry. The European banking industry has been investigated to a lesser extent, e.g. Bischoff, Brüggemann and Daske (2010, p. 7) mentioned that the lessons learned from the global financial crisis are almost exclusively based on evidence from the U.S. environment. This lack of prior research is one of the reasons for this study, which empirically investigates the European Banking industry. Moreover, the differences in the extent of the accounting changes for U.S. GAAP and IFRS are quite large. These differences will be explained in the next section.

This study will measure the market response as a whole. Investors may perceive fair value accounting to harm banks’ financial stability. However, fair value accounting may also be perceived as merely the messenger; in the sense that investors at large believe that fair value accounting didn’t exacerbate the financial crisis. Therefore, in order to contribute to the fair value accounting debate, the following research question has been formulated:

7

Were banks’ stock prices affected by major events that (could potentially change) changed fair value accounting regulation?

If fair value accounting was perceived as a key contributor to the financial crisis, relaxation of fair value accounting legislation would have resulted in an increased stock price. Following this line of reasoning, changes that would support or solidify fair value accounting would result in a decreased stock price. Furthermore, the extent of the response illustrates whether investors perceive fair value as value relevant in times of turmoil. A strong and positive abnormal response indicates that investors agree that fair value accounting harms banks and this should be supported by an underlying assumption or belief that the incremental value of financial assets may exceed the market value.

Alternatively, if there is a negative reaction, investors perceive that the market value reflects the underlying economics correctly and that relaxing fair value accounting will assist managers in concealing true performance, which would limit their ability to assess firm value. In the end, adjustments to accounting standards and legislation should facilitate the users of financial statements to make better informed decisions. Researching how the stock prices responded, following a major event yields an interesting contribution regarding the fair value accounting debate. This study evaluates how investors perceived fair value accounting during the financial crisis and whether they perceived the measures taken by the IASB to be just.

8

2. Literature review and hypotheses

This section will explain the prior literature and background information upon which this study builds. At first, fair value accounting in times of market turmoil is discussed along with alternative perspectives that are incorporated in prior literature. Second, the institutional background and the politics involved in the fair value accounting debate will be discussed. Third, the reclassification amendments will be explained along with the differences between U.S. GAAP and IFRS. Fourth, the events leading up to the amendments of IAS 39 and IFRS7 will be discussed. Fifth, the theory will be discussed. Lastly, the hypothesis development will be disclosed.

2.1 Fair value accounting

This paragraph will discuss the various papers related to fair value accounting in times of market turmoil. Prior to the financial crisis, Walton (2004) already observed that the banks are unhappy about having to value ‘available for sale’ and ‘held for trading’ assets and liabilities at fair value and believe this will cause great fluctuations on a period-to-period basis as a reflection of short term shifts in the market. The SEC (2008, p. 182) described these fluctuations as pro-cyclicality, “the amplification of otherwise normal cyclical business fluctuations”. During the financial crisis, the SEC (2008) and IMF (2008) expressed significant concerns regarding the pro-cyclical effect fair value accounting could have on the financial system. The issue is whether these fluctuations reflect the underlying economics. Allegedly, the market prices in times of market turmoil deviate from the fundamental value. If the market value is lower than the fundamental value, selling these assets can be a suboptimal decision that harms the company and therefore the shareholders. Arguably, only firms that are pressured by liquidity and solvency requirements would be most willing to take such decisions. Regardless, the question remains, is there a pro-cyclicality effect incorporated in fair value accounting?

In order to investigate the aforementioned concerns, the SEC (2008, pp. 180-181) computed a proxy for full fair value income for a sample of commercial banks. This full fair value proxy estimates what income would have been if fair value would have also been applied to instruments that are otherwise not measured at fair value, with all unrealized gains and losses taken into income. This study found that income volatility under the full fair value proxy is more than three times higher than the volatility of then-current U.S. GAAP comprehensive income. This indicates that accounting can amplify the fluctuations of the

9

underlying economics. Magnan (2009, p. 208) shares these concerns and points out the extreme volatility during the financial crisis, with stock prices fluctuating five to ten percent from previous days, arguing that fair value is informative but not sufficient for regulators in planning and targeting their interventions. However, he also argues that fair value accounting values may not be the issue. Rather, the accompanying disclosure may be flawed.

Bowen, Khan and Rajgopal (2010) investigated whether the concerns of the SEC were shared by investors. They examined U.S. banks’ stock price response on specific event dates at which the probability increased that fair value accounting legislation would be relaxed. They found a positive relationship between fair value accounting relaxation and banks’ stock prices. Therefore, they argue that in the U.S. banking industry, investors believed that pro-cyclicality harmed banks.

Arnold (2009, pp. 805-808) argues that the financial crisis challenges the credibility of market theory: “if we acknowledge that capital markets do not price assets efficiently, what then is the economic impact of fair value?”. Once again, it all comes down to whether market prices reflect the underlying economics adequately. In this regard, there is a considerable amount of literature that argues that capital markets do price assets efficiently and they challenge the pro-cyclicality concerns made earlier.

Badertscher, Burks and Easton (2011) researched the fair value accounting losses and the sales of securities during the crisis. They used a sample consisting of 150 bank holding companies that were among the largest holders of non-Treasury available-for-sale (AFS) and held-to-maturity securities (HTM) as of June 30, 2006. They find that industry- and firm-level sales of AFS and HTM during the crisis were similar to before the crisis. Furthermore, they find no evidence that banks increasingly sold securities at losses during the crisis, providing no support for claims of “fire-sales” of assets. The findings contradict many claims made by critics of fair value accounting during market turmoil. Furthermore, Ryan (2008) argues that fair value accounting adequately reflects the economic cycles, and merely functioned as a mechanism that recalibrated as soon as the economic cycle turned. He points out that the entire economic system failed to identify the risks of the rapid-growth in subprime mortgages and the inevitable reversal of home price appreciation. The consequences of these, among other factors, can’t be eliminated by accounting. It can merely provide a common information set upon which market participants can recalibrate their decisions. Bleck and Liu’s (2007, p. 252) findings illustrate the relevance of a common information set that reflects the economic cycle. They illustrate that under historic cost accounting managers can use opaqueness in the financial market to hide project performance.

10

This in turn, can lead to inefficient continuation of projects by the shareholder. They also find that volatility overall increases under historic cost accounting and that managers can use historic cost accounting opportunistically. They furthermore support the fair value accounting method because it could function as an ‘early warning system’. Moreover, Barth, Beaver and Landsman (2001) and Landsman (2006) found that a firm’s stock price is more closely associated with the market value of the underlying asset than historic cost accounting. Therefore, they argue that fair value accounting is more value relevant than historic cost accounting.

However, even under the assumption that fair value is more applicable, investors will still discount the fair value estimates based on their own estimations of the fundamental values of assets. This is illustrated by Goh, Ng and Young (2009); they find that investors differently price mark-to-model and mark-to-market assets relative to the fair value estimates reported by the banks. They used a combination of time-series and cross-sectional analysis to measure how investors price the fair value assets reported under SFAS 157. SFAS 157, is the three-level hierarchy based on the inputs available to estimate fair value. They find that investors price mark-to-model assets less than mark-to-market assets, which indicates that fair value estimates of mark-to-model assets do not sufficiently factor in the risks of illiquidity and information risk. Moreover, they observed that pricing of mark-to-model assets declines over the course of 2008. This indicates that investors at large do not trust the banks’ fair value estimates for mark-to-model assets and that this lack of trust exacerbated during the financial crisis. For this study, based on these results, it can be argued that if fair value accounting were to be relaxed, investors at large have already discounted the fair value estimations and under the assumption that investors are rational, they will most likely maintain these estimations as soon as banks start reclassifying fair value assets to historical cost. However, the underlying reasoning of this study is that fair value accounting is harmful due to the pro-cyclicality effect incorporated in fair value accounting. Banks, especially those more threatened by liquidity and solvency requirements would therefore still benefit from a (temporary) relief in pressures from potential fair-value accounting write-downs.

2.2 The events preceding the Fair value accounting amendment

The global financial crisis worsened throughout 2008. Banks across the globe were pressured by considerable fair value write-downs. Mid 2008, European politicians started to become increasingly worried about the competitiveness of European banks. On the 8th July, the Economic and Financial Affairs Council (ECOFIN, 2008) met in Brussels, where these

11

concerns were extensively discussed by the European Finance Ministers. These concerns about competitiveness rose from the accounting differences between U.S. GAAP and IFRS. Prior to the amendments to IAS 39 and IFRS, U.S. GAAP allegedly had a considerable advantage over IFRS. The IASB’s press release on 13 October 2008 regarding the amendments made to IAS 39 and IFRS underlines these concerns. Sir David Tweedie, Chairman of the IASB said that: “by responding to the crisis, the IASB notes the concern expressed by EU leaders and finance ministers through the ECOFIN to ensure that European financial institutions are not disadvantaged vis-à-vis their international competitors in terms of accounting rules and of their interpretation” (IASB, 2008c, p. 1)

This disadvantage stems mainly from SFAS 65. SFAS 65 is for mortgage banking activities and allows held for sale assets to be reclassified as held for investment. These transfers are permitted if the entity has the ability and intent to hold the loan for the foreseeable future or until maturity. This standard offered U.S. GAAP banks the discretion needed to avoid the pressures from possible fair-value write-downs.

Bischoff, Brüggemann and Daske (2010) argue that at first, the IASB was reluctant to change the scope of fair value measurement in IAS39. However, the breakdown of Lehman Brothers entirely changed the situation. Because of this, banks were exposed to the risk of reporting substantial losses and their capital ratios would likely be harmed for the financial year 2008. Politicians became worried and started to increasingly pressure the accounting body. This created a chain of events that lead to the amendment of IAS 39.

At first, on October 3, the IASB announced its current status of its response to the financial crisis and its next steps (IASB, 2008a). The IASB explained that it recognized the need to clarify IFRS to address new market developments. Moreover, the IASB responded to the initial concerns that politicians and bank lobbyists had expressed: “The IASB is closely monitoring developments in the United States and other jurisdictions to avoid unnecessary inconsistencies in accounting treatments under IFRSs and US GAAP” (IASB, 2008a, p. 1). The IASB also declared its commitment to establish consistency of fair value measurement guidance between IFRS and US GAAP. Furthermore, they committed immediately to consider the ability to reclassify financial instruments, mainly to tackle the aforementioned inconsistencies.

On October 4, at the summit of the G8, Nicholas Sarkozy announced: “we will ensure that European financial institutions are not disadvantaged vis-à-vis their international competitors in terms of accounting rules and of their interpretation. In this regard, European financial institutions should be given the same rules to reclassify financial instruments from

12

the trading book to the banking book including those already held or issued” (European G8 members, 2008, p. 2).

On October 7, ECOFIN (2008) met in Luxembourg and agreed upon a coordinated approach to the financial crisis. The meeting aimed at finding a solution to restore the confidence in the economy and proper functioning of the financial sector. The main results of this meeting were that members agreed to raise guarantees on deposits to a minimum of 50,000 euros and that they would amend certain accounting rules applicable to banks without a delay. The council underlined the necessity of avoiding any distortions of treatment between US and European banks due to differences in accounting rules.

On October 8, EU Commissioner for Internal Market and Services C. McCreevy declared in the European Parliament that the Commission had prepared the legislation to have a carve out from IAS 39. The legal basis for this carve out stemmed from the EU IAS Regulation (2002/1606), which requires all EU listed companies to follow IFRS since 2005 and includes a requirement that IFRS endorsed by the EU must not disadvantage companies as compared to those in other markets (Andre et al., p. 14). McCreevy therefore announced that the Commission was ready to adopt its own European version of IAS 39, if the IASB didn’t alter the accounting standard on its own. This version would allow the reclassification of financial assets without introducing any disclosure requirements (Kusano and Sanado, 2013, pp. 13-15).1 With this announcement, McCreevy, and subsequently the EU, further enlarged the pressure on the IASB to take decisive measures.

On October 9, the Trustees of the IASC foundation announced their unanimous support for the approach that the IASB laid out on the 3rd of October (IASB, 2008b). As previously mentioned, the most important steps were: the immediate consideration of the ability to reclassify financial instruments and the consistency of fair value measurement guidance between the IFRS and U.S. GAAP (IASB, 2008a). The IASB announced that it would seek to eliminate the differences between IFRS and U.S. GAAP regarding the reclassification of financial instruments. The trustees, the IASB’s oversight body, hereby agreed that the IASB could suspend its due process (IASB, 2008b). This allowed the IASB to rapidly align the standards regarding the reclassification of financial instrument with the FASB. The objective was to create a level playing field for those companies applying IFRS and those using U.S. GAAP.

1

13

On October 13, the IASB used the support from the trustees of the IASC to bypass the due process, thereby neglecting lengthy consultation procedures, which in turn allowed the IASB to finalize the amendments to IAS 39 (IASB, 2008c). After the amendment was made, David Tweedie, Chairman of the IASB, argued that the move in order to level the playing field was made because companies felt it was unfair that they were being hit by a different loss than those applying U.S. GAAP.2 He further claims that political pressures didn’t impact the IASB’s decision. He called the amendment, a classic example of why we need one set of accounting standards. Furthermore, David Tweedie (House of Commons, 2008) provided evidence that the carve out would have been disastrous. As Andre et al (2009, p. 15) explain, the Accounting Regulatory Committee (ARC) is only allowed to remove material from standards and is therefore incapable of adding material. Therefore, the removal of the paragraphs that prevented reclassification would subsequently remove all safeguards that are incorporated in those removed paragraphs. Hence, David Tweedie claimed that accounting would have been totally out of control, if the carve out wasn’t prevented. Moreover, Kusano and Sanado (2013, pp. 10-13) argue that the carve out would have seriously harmed the legitimacy of the IASB. By issuing amendments to IAS 39 the IASB avoided another EU carve out and maintained most of its legitimacy. However, it still caused concerns about the IASB’s governance because it skipped out on its regular due process (Andre et al., p. 15).

On October 15, the European Commission (2008) adopted the amendments to the accounting standards. At the summit in Brussels, Internal Market Commissioner Charlie McCreevy, commended the swift response of the Commission to the ECOFIN request from October 7. The European Commission (2008) claimed that the current financial crisis justified the use of reclassification by companies. It argued that the changes to the accounting standards were intended to mitigate the consequences of the market turmoil.

2.3 Fair value accounting and the reclassification amendments

Prior to the reclassification amendments, banks that were reporting under IFRS couldn’t avoid the reporting of fair value losses. IAS 39, paragraph 50 required that “an entity shall not reclassify a financial instrument into or out of the fair value through profit or loss category while it is held or issued”. This was contrary to SFAS 65, which offered more discretion through a reclassification option. In order to level the playing field, banks and politicians lobbied with the IASB to amend the standards. As a response, the reclassification

2

14

option was included in IAS 39 under paragraph IN8A (IASB, 2008e). It permits entities to reclassify non-derivative financial assets out of the fair value through profit or loss category in particular circumstances. The amendment also permits an entity to transfer from the available-for-sale category to the loans and receivables category if the entity has the intention and ability to hold that financial asset for the foreseeable future. This option enabled banks that applied IFRS to reclassify assets out of the fair value category. Furthermore, paragraph 50 of IAS 39 was amended and paragraphs 50B-50F and 103G are added (IASB, 2008e). These changes further facilitate the reclassification option and include safeguards to ensure that the standard is used accordingly. Moreover, it limits the discretion given to preparers of financial statements that apply IFRS. Arguably the most important paragraph, 103G, sets the effective date and the transition; it enables preparers of financial statements to retrospectively reclassify financial assets up until the 1 July, 2008. The IASB also proposed amendments to require companies applying IFRS to include additional disclosures (IASB, 2008D). The IASB has proposed these amendments at the request of users of financial statements that suggested that enhanced disclosures about fair value measurements are required (SEC, 2008 p. 193). This exposure draft proposes amendments to disclosure requirements based on a three-level hierarchy, which is similar to the one used under SFAS 157. This exposure draft was open to comments until December 15 and was eventually adopted in March 2009. The aforementioned amendments further aligned fair value accounting as adopted by the IASB and FASB.

The definitions for fair value accounting were, prior to the amendments, already very similar (Laux and Leuz, 2008, p. 827). FAS 157, defines fair value as “the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date.” Whereas, IFRS 7 defines fair value as “the amount for which an asset could be exchanged or a liability be settled, between knowledgeable, willing parties, in an arm’s length, orderly transaction”

The definitions are quite similar, however, FAS 157 includes the measurement date, which emphasizes that fair value is measured at a particular moment in time and is therefore subject to changes on a daily basis. An important aspect is the ‘orderly transaction’ which is incorporated in both definitions. Laux and Leuz (2009, p. 827) explain that in order to arrive at an orderly transaction, entities might have to make significant adjustments. Goh, Ng and Young (2009, p. 6) explain that the stated objective of fair value accounting is to increase the transparency of asset valuation, which allow investors to make better informed financial decisions. However, fair value accounting contains a fundamental assumption, which is the

15

notion of orderly transactions. Therefore, they question whether fair value accounting can still continue to provide fair value estimates that investors can rely upon in disorderly markets to price the banks’ assets. Ryan (2008, p. 1608) shares these concerns and he argues that amid the financial crisis the notion of orderly transaction became practically useless for preparers’ measurements of fair value. It is particularly difficult for these preparers to convince auditors, among others, of the reasonableness of measurements when numerous fire sales by illiquid firms in a distressed market occur at considerable lower levels.

Bischoff, Brüggemann and Daske (2010) investigated the economic consequences of the reclassification amendment of IAS 39. They point out the advantages for European Banks relatively to the U.S. Banks due to differences in the extent of the amended IAS 39 versus the content of SFAS 65 and 115. First, the reclassification option under U.S. GAAP is hardly used by U.S. Banks. The reason for this is that under IAS 39 banks can retrospectively exercise their reclassification option. Furthermore, impairment rules after (re)classification are more restrictive under U.S. GAAP. The authors used a comprehensive sample of 302 IFRS reporting banks. They found that the amendments produced short-term benefits and long-term costs. The relaxation of fair value accounting did provide short-term relief for most troubled banks. On the other hand, they also found some evidence that banks abuse the reclassification option to decrease transparency, resulting in a significant discrepancy between the ask-bid prices.

2.4 Theory

The theory used for formulating the hypotheses is the noisy market hypothesis. Bowen, Khan and Rajgopal (2010) made use of this hypothesis for their research on U.S. Banks as well. The noisy market hypothesis contrasts the efficient-market hypothesis, in that it claims that the prices of securities are not always the best estimate of the true underlying value of the firm. Boyer (2007, p. 2) neatly explains this by arguing that the proponents of fair value do assume that markets are efficient. This technically means that all relevant information is incorporated into the quoted prices. However, he argues that: “it does not imply that existing markets do provide an approximation of the fundamental value of an asset, computed from its expected return, given a long-term interest”. The theory doesn’t invalidate the efficient-market hypothesis; it merely points out that in times of market turmoil emotions might impair investor rationality. It argues that prices can be influenced by speculators, insiders and institutions that buy and sell stocks for reasons unrelated to

16

fundamental value, such as for diversification, liquidity and taxes. Moreover, in illiquid markets there may not be sufficient price observations to adequately price the fundamental value of assets. This theory therefore aligns with the circumstances during the financial crisis, namely times of market turmoil and illiquidity.

2.5 Hypotheses development

In order to examine whether fair value accounting was perceived to be harmful to banks, two hypotheses have been formulated. Intuitively, if fair value accounting triggered fire-sales of financial products that resulted in a further decrease in market price, then there is a possibility that a pro-cyclicality effect is incorporated in fair value accounting. As argued by the SEC (2008), the aforementioned effect is pro-cyclical since the initial decrease in market price will pressure banks to revalue and sell their financial assets again. This will result in an even greater decrease in market price, thereby creating a downward spiral. Therefore, if fair value accounting was harmful to banks because illiquid assets were inappropriately valued (noisy information hypothesis) due to the market turmoil, then a positive stock price reaction should be observed following an announcement that increased the probability that fair value accounting rules would be relaxed. It is predicted that events that indicate an increase in the probability of fair value accounting rules relaxed by the IASB will result in a positive stock price reaction. The related events are number 1, 5, 6 and 7, as described Table 1. Therefore, and as aligned with the research question, the following hypothesis has been formulated:

Hypothesis 1: Events that indicate an increase in the probability of fair value accounting rules being relaxed by the IASB are associated with positive stock price reactions among banks across Europe.

[Insert Table 1 here]

As previously argued, relaxation of fair value accounting by the IASB should be beneficial for firms facing financial pressures. However, not all forms of fair value accounting relaxation will be beneficial for shareholders, it is important to consider the extent of relaxation when determining the predicted stock price reactions. In the event description there are two situations described, relaxation according to the IASB and the threat of a European carve out. There is an accompanying disadvantage of a European carve out that would be disastrous for investors. Andre et al. (2009, p. 15) explain that the ARC can remove

17

material from standards, but isn’t allowed to add material. Hence, removing the paragraphs that prevent reclassification would grant a great amount of discretion to the preparers of financial statements since all accompanying safeguards would evaporate. This discretion could very well be used at the expense of investors. Andre et al. (2009, p. 15) acknowledge David Tweedie’s evidence and his claim that it would have enabled a free for all in accounting for financial instruments. Therefore, a carve out would have reduced the transparency and impair the ability of investors to adequately monitor financial instruments. The events that increased the probability of a European carve out are event numbers 2-4, as described in Table 1.

Hypothesis 2: Events that indicate an increase in the probability of a European carve out from IAS 39 are associated with negative stock price reactions among banks across Europe.

Khan and Rajgopal (2010) have investigated cross-sectional reactions to factors that would arguably make firms more susceptible to fair value accounting changes, low regulatory capital, more assets recorded at fair value and poor asset liquidity. Unfortunately, their findings were rather modest. Their line of reasoning remains however solid, solvency, and liquidity, among others should have an influence on the extent to which a firm is pressured to sell financial assets below their fundamental value. This research will therefore use a comprehensive measure that arguably reflects the combined effect of the aforementioned factors; among other circumstances, that impact a firm’s risk of default. This combined effect is measured through the corporate variable credit rating, which is used as a comprehensive measure, and is defined similarly to Weber (2006, p. 246): “the rating agent’s assessment of the probability that a firm will default on interest or principle payments”. Banks that have a lower credit rating are likely to respond contrary to a relaxation of fair value accounting than banks that have a higher credit rating. Bischoff, Brüggemann and Daske’s (2010, pp. 48-49) findings already suggest that the relaxation of IAS 39 provided short-term benefits against long-term costs. They argue that long-term costs occur due to a decrease in fair value information that impacts the reporting transparency. Moreover, they suggest that the benefits were mostly for troubled banks. Following this line of reasoning, banks that have a lower credit rating are closer to violate debt covenants and/or regulatory capital requirements. This offers them room to benefit more from reclassifying their deteriorating financial assets out of the fair value category than banks with a higher credit

18

rating. They will benefit more by avoiding the pressure to sell financial assets in a fire sale. This in turn, is beneficial to investors since the market values were allegedly below the fundamental values of the financial assets. Banks that have a high credit rating will benefit less from the relaxation in fair value accounting, primarily because they aren’t as compelled to engage in fire sales. Therefore, for banks that have a higher credit rating, it is conjectured that investors may perceive the costs associated with the decreased transparency due to the relaxation of fair value accounting, larger than the benefits stemming from this relaxation.

Hypothesis 3: The stock price reaction to events that indicate an increase in the probability of fair value accounting rules being relaxed is more positive for banks that have a lower credit rating.

Continuing on Bischoff, Brüggemann and Daske’s (2010, pp. 48-49) line of reasoning, they also suggest that troubled banks (lower credit rating) will benefit more from fair value accounting relaxation due to pressures stemming from the market turmoil(noisy information hypothesis). Furthermore, the Capital Asset Pricing Model argues that firms that have a lower credit rating should be more sensitive because they have a greater risk of default. This suggests that banks with a lower credit rating (troubled banks) will respond more strongly to fair value accounting relaxations. However, it is also predicted that the market responses are contradictory, which makes this prediction somewhat arbitrary. Regardless, it is important to test whether banks that have a lower credit rating respond more strongly than those that have a higher credit rating. Since, it is plausible that the benefits for the weaker banks may outweigh the costs for the stronger banks. This would be a valid argument for the relaxation of fair value accounting legislation. Although, it remains debatable if it is politically justifiable to relief some pressures from weaker banks with a fair value accounting relaxation. Especially, since this relaxation may have negative consequences for the stronger banks. Nevertheless, if the benefits for weaker banks exceed the costs for stronger banks, the fair value accounting amendments would arguably have a stronger public support. Lastly, it would also support H1, since the alleged pro-cyclicality effect incorporated in fair value accounting would impact weaker banks more fiercely. Hence, these banks would benefit the most from a fair value accounting relaxation.

19

Hypothesis 4: The stock price reaction to events that indicate an increase in the probability of fair value accounting rules being relaxed is larger for banks that have a lower credit rating.

Furthermore, besides stock price reactions it is also expected that there is a trading volume reaction. Kim and Verrechia (1991) investigated the differences and relations between trading volume studies and price studies. Their main finding is that trading volume is proportional to both the absolute price change and a measure of differential precision across traders. Their first argument is consistent with Chen et al. (2002, p. 171), they also found that there is a positive correlation between trading volume and the absolute value of stock price change. Their causality results suggest that returns cause volume, and to a lesser extent, that volume causes returns. Therefore, it is expected that the abnormal stock price reactions are associated with abnormal trading volume reactions. The second argument of Kim and Verrechia (1991) is even more relevant, since it allows assessing investor response from a different perspective. They argue that price change reflects the average change in traders’ beliefs caused by the announcement, whereas, volume change reflects the traders’ idiosyncratic reactions. These different reactions are caused by differing levels of precision of traders’ private information. Announced information is relatively more important to traders that have less precise private information, resulting in a stronger impact on their beliefs. Hence, volume change is the sum of differences in traders’ reactions.

Therefore, we expect the public announcements, as described in the events 2.2, to transmit reliable and relevant news that will triggers investors’ to reassess their financial positions. This process is expected to result in a heightened trading volume.

Hypothesis 5: Events that indicate an increase in the probability of fair value accounting rules being relaxed are associated with trading volume reactions among banks across Europe.

This study aims to build on the limited prior literature regarding standard setter performance amid the financial crisis. It does so by, to the best of my knowledge, being the first to investigate stock price reactions by distinguishing banks based on their corporate credit rating. This is particularly useful because it allows investigating whether the troubled banks specifically benefited from the fair value accounting amendments. This in turn was also part of the intent of politicians and lobbyists, aside from leveling the playing field with

20

the U.S. Furthermore, the number of events tested in this study greatly exceeds those in Bischoff, Brüggemann and Daske (2010), which is the only identified similar study in Europe. Their study emphasized on the trade-off between short-term and long-term consequences of the fair value accounting amendments and therefore they only tested two events to map the short-term consequences. Lastly, measuring the trading volumes allows for a more holistic evaluation of the short-term consequences of the fair value accounting amendments.

3. Sample and Methodology

3.1 Sample selection

The sample is selected based on BvD Bankscope and Compustat Global. At first, all banks were selected for the time period of 2008 using the BvD Bankscope database, this in turn lead to 22,867 observations. After this, only the banks that have a listing on a stock exchange were included, resulting in 2944 observations. Furthermore, only the banks that were applying IFRS were selected, resulting in 838 observations. Afterwards only the European banks were selected, resulting in 165 observations after the duplicates were removed. The sample was matched with Compustat in order to obtain daily closing stock prices, resulting in 112 observations. The 53 observations were excluded due to infrequent trading or lack of data availability, thereby avoiding the problem of thin trading.

Credit rating information was gathered through Compustat, Thomson Reuters Datastream, annual reports, press releases and announcements from S&P, Fitch and Moody’s. In order to match the credit ratings a long-term rating scale comparison from the Bank for International Settlements was used to transform ratings from Fitch and Moody’s to S&P. This credit rating transformation model is displayed in Table 2. For the majority of the sample an S&P credit rating was available, regardless 14 credit ratings had to be transformed. This resulted in a final sample of 98 firms, 14 firms were excluded due to data unavailability.

3.2 Methodology

This research uses two models for the event study methodology. Both models have been used in prior research related to this topic, by Bischoff, Brüggemann and Daske (2010) and Bowen, Khan and Rajgopal (2010), among others. The first model is the standard event study methodology, the single index model (market model). The second model is used as additional analysis to ensure robustness of results and consists of a two-factor model. In this

21

model, one additional variable is incorporated to account for changes in the short-term interest rate. These models will be explained more thoroughly after discussing the event selection, estimation period and estimation window which apply for both models.

There are 7 events identified as described in chapter 2.2, the predicted direction of the stock response (psign) is included in Table 1. The estimation period consists of 220 days before the first event (-21 until -240). According to Sorokina, Booth and Thornton (2013), the estimation period should be between 100 and 500 days. The event day is defined as (t = 0) for each public announcement. For all events the estimation window is focused on three days, day-1 the event day (0) and day+1. This estimation window is suggested by MacKinley (1997) and is applicable since it allows evaluating the informativeness of each of the stock price reactions surrounding the events. This is essential since some events are related to each other and happen subsequently.

The primary model is the market model3, which is a single factor model; the abnormal returns are calculated using the following formula:

AR

i,t

=

R

i,t -

(α

i

+

β

i

R

m,t

).

R

i,t is the raw return of firm i for day t and

R

m,t is the return on the market portfolio on day t

proxied by the S&P Europe 350 Equal Weight Index. The market model parameters

α

i

(intercept) and

β

i(slope) are calculated during the estimation period of 220 days.

The second model is a two-factor model4. This model is used as an additional analysis. It is common to model bank-specific stock returns by also controlling for interest rate changes, e.g. see Beatty et al. (1996). Hence, the abnormal returns are calculated based on the following formula:

AR

i,t

=

R

i,t -

(α

i +

β

i

R

m,t +

β

i

R

t,t

).

Ri,t is the raw return of firm i for day t, R m,t is the return on the market portfolio on day t

proxied by the S&P Europe 350 Equal Weight Index and R t,t is the daily yield on day t of a

3-month UK treasury bill (bond).

Besides the stock price reaction, there should also be a stock volume reaction. Chen et al. (2002, p. 171) found that there is a positive correlation between trading volume and the absolute value of stock price change. Their causality results suggest that returns cause volume, and to a lesser extent, that volume causes returns. Therefore, as a part of the

3

Similar to Bischoff, Brüggemann and Daske (2010)

4

22

additional analysis, the trading volume will be tested. The trading volume is expected to be largely impacted since all events arguably reveal important information, which will trigger investors to reconsider their financial positions. Moreover, according to Chen et al (2002) the anticipated stock price reaction will be accompanied with a trading volume reaction. The abnormal volume reaction will therefore be estimated as part of the additional analysis. Following Landsman et al. (2002) and Alves and Dos Antos (2008), the abnormal volume is calculated according to the following formula:

AV

i,t

=

[V

i,t

–

E

(

V

i,t

)

]

/

σ

i.

V

i,t is the ratio of shares of firm i traded on day t to the shares outstanding of firm i on day t.

The expected trading

(E

(V

i,t

))

is the firm-specific average of daily trading

(V

i,t

)

throughout

the estimation period, which consists of 220 days (-21 until -240). The abnormal trading volume

(AV

i,t

)

is then the difference between trading

(V

i,t

)

and expected trading (E(Vi,t

)),

divided by the standard deviation

(σ

i

)

of trading volumes in the estimation period. By scaling

by σi, the formula controls for changes in normal (non-announcement period) volume across

time.

Prior research, e.g. Bowen, Khan and Rajgopal (2010), used firm-specific characters such as liquidity and capitalization in order to identify troubled banks. These banks should arguably benefit more from a relaxation of fair value accounting. In order to account for these firm-specific characters an additional factor, the corporate firm credit rating, is included. This variable is used as a comprehensive measure to identify troubled banks and account for risks regarding the liquidity and capitalization, among other variables that influence the firms’ risk of default. The sample is stratified into two groups based on the credit rating. Banks that have a credit rating of A+ or higher are allocated to group 0 and banks that have a credit rating of A or lower are allocated to group 1. This stratified sample and the accompanying credit rating distribution are displayed in the attachments, Table 3.

[Insert Table 2 and 3 here]

Finally, tests are run for each of the models. The first set of tests involves analyzing the stock price reaction for each event based on the average abnormal return (AAR), as displayed in the attachments, Table 1. The second set of tests analyzes the stock price reaction by combining all events and events that are related to each other. For each group of events the

23

stock price reaction is analyzed according to the cumulative average abnormal return (CAAR). The third set of tests analyzes the influence of a firms’ credit rating by comparing the AAR and CAAR based on the stratified sample. The fourth test analyzes the difference in the mean of CAAR for the entire sample period based on a stratified sample that is modified to show only positive abnormal returns. This method enables analyzing whether firms with a lower credit rating respond more strongly than firms with a high credit rating. This could yield valuable insights in whether the costs for the high credit rating firms exceed the benefits for the low credit rating firms. The final set of tests analyzes whether the trading volume was influenced by the events.

4. Empirical findings

First of all, as mentioned earlier the events are summarized in Table 1, including the predictive market reactions (psign). Each positive predictive sign indicates an increase in the probability of fair value accounting rules being relaxed by the IASB. The negative predictive signs indicate an increase in the probability of a European carve out.

The stock price reactions are reported in 4 different tables. The first two (4 and 5) tables are based on the single index model; the latter two (6 and 7) are computed using the two-factor model. The structure of the tables is similar for both models. The first table for each model (Table 4 and 6) contains the AAR based on the events and within the estimation period of 3 days. Since the events occur in a short event period (11 days), relatively short windows are used to better isolate the effects. For this reason, when estimation periods overlap until day (0), only the new event is shown in Table 4 and 6. For interpretations purposes, the full sets of results for single events (event day -1, 0 and +1) are displayed in table A. It is hereby acknowledged that there is a strong overlap in events; this in turn reduces the consistency of pinpointing the market reactions to specific events. Nevertheless, some clear patterns have emerged and events (or event days) are therefore combined in the second table for each model (Table 5 and 7) in order to make these patterns salient. These tables contain the CAAR as computed by both models.

The results for both models are reported in the description of the results. The results can be more readily interpreted by focusing on the tables from model 1, particularly because the results are quite robust for both models.

24

4.1 Stock price reaction IASB fair value accounting relaxation

The first set of tests evaluated whether (an indication of) relaxation of fair value accounting rules by the IASB (EU) resulted in positive (negative) reactions. On average banks had significant positive reactions to 3 out of the 4 events that have a positive prediction sign. These results are consistent with Bowen, Khan and Rajgopal (2010) whom only tested two events, which are similar to event 6 and 7. When the events are combined for the full period (events 1-7) the results are insignificant and the CAAR is -0.02%. This is not unexpected, the predict market response for a fair value accounting relaxation by the IASB (+) is contrary to the expected market response for the threat of a European carve out (-). However, when the events are combined excluding the negative predictive sign events, 4 out of 4 events report significant positive stock price reactions. Event 1 combined (day -1 and day 0) reports a CAAR is 2.44%, for event 5-7 combined the CAAR is even 2.89%. These results are even slightly stronger for model two (2.57% and 2.92%), which controls for changes in the interest rate. Throughout all tests the market model seems robust; the results are rather consistent with the two-factor model.

The aforementioned results provide sufficient evidence to support hypothesis 1, events that indicate an increase in probability of fair value accounting rules being relaxed by the IASB are associated with positive stock price reactions. These results illustrate the value-relevance of fair value accounting rules, as determined by the stock market participants. Assuming that the market reaction captures the underlying economics, it appears that investors at large did perceive fair value accounting to be harmful to banks.

[Insert Table 4 and 5 here]

4.2 Stock price reaction EU carve out threat

A fair value accounting relaxation of the magnitude of a European carve out is expected to be perceived contrary to a relaxation of the IASB. It is expected to be perceived negatively by investors. This is due to the inherent removal of all safeguards incorporated in the removed paragraph. Andre et al (2009, p. 15) also mention this threat, the ARC can only remove an entire paragraph; it isn’t allowed to add material. Three events have been identified that arguably increased the likelihood of a European carve out, event number 2, 3 and 4. On average banks had significant negative reactions for all three events that have a negative prediction sign. Event 2 and 3 are significant at 1% for both models on the event

25

date, as shown in Table 4 and 6. The strongest reaction is on the day of event 3, on average banks responded respectively -2.95% and -2.79%. When events 2-4 are combined, the CAAR for the three events is respectively -5.36% (Table 5) and -5.04% (Table 7) for model 2, both are significant at 1%. Overall, these results strongly support hypothesis 2. The findings indicate that investors perceived the threat of a European carve out as damaging. It seems that investors at large, feared for reduced transparency, leaving them incapable of adequately monitoring the accounting for financial instruments if a European carve out were to occur. This in turn, is consistent with claims and evidence from David Tweedie (House of Commons, 2008), Chairman of the IASB.

Even though the empirical results are strong, there is one limitation regarding the third event. On this day, ECOFIN (2008) underlined the necessity of avoiding any distortions of treatment between the U.S. and European Banks. This was a clear indication that the accounting rules and legislations were to be amended without a delay. However, it was also announced, that the members committed to raise guarantees on deposits to a minimum of 50,000 euros. Although this is an immediate response to the financial turmoil, it’s debatable whether investors perceived this news positively. Investors may have perceived it as a warning signal, indicating that the financial instability is more severe than they had previously assumed. It is unclear to which extent the results have been influenced by this announcement that is outside the scope of this study. However, the effect of this announcement is arguably somewhat dampened; some members had already raised guarantees on deposits prior to the ECOFIN meeting. Regardless, the results should be interpreted with this limitation in mind.

4.3 Stock price reaction additional analysis

The prior sections have illustrated that the conclusions from Bischoff, Brüggemann and Daske (2010) are somewhat narrow. Based on their empirical evidence they concluded that fair value accounting relaxation produced short-term benefits and long-term costs. However, it turns out that investors responded differently according to the extent of the relaxation of fair value accounting. These contradictory reactions are probably one of the reasons why the CAAR turned out to be insignificant (-0.02% and + 0.04%) throughout the entire event period, the 2nd of October until the 16th of October, based on CAAR for the full (non-stratified) sample. Moreover, the market reactions are also expected to be different depending on the bank’s financial stability, capitalization, liquidity, and risk of default,

26

among others, as measured by the credit rating. In order to test this hypothesis, the sample was stratified according to Table 3. The results are shown in Table 4 to 7 at panel B. Mean 0 is the sample group with a high credit rating and Mean 1 with a low credit rating. When individually evaluated, (Table 4 and 6) the results are modest. The results are significant for 4 out of 11 events for model 1 and 3 out of 11 according to model 2. However, there is a clear pattern which is more appropriately illustrated when the events are combined, as in Table 5 and 7. For the first combined event, event 1 (event day -1 and 0) there is a significant positive difference. The CAAR of the high credit rating group is 2.43% higher than the low credit rating group. These results suggest that initially investors perceived the proposed actions of the IASB to be more beneficial for the high credit rating group. However, the sign of the market reactions for the stratified sample already took another turn at the threat of an EU carve out. This is illustrated by the market reaction concerning events 2 to 4; the combined difference as measured by CAAR is respectively -2.75% and -1.94%, albeit insignificant. The response is also similar for the latest three events. These events are arguably the most important, primarily because event 6 is the actual adoption of the reclassification amendments to IAS39 and IFRS 7. These amendments were officially endorsed by the EU commission on event 7. Throughout this period the market reaction signaled the importance of a banks’ financial strength, as measured by credit rating and as perceived by investors at large. The results are significant at 1% based on a two-tailed test for both models; the difference in CAAR between group 0 (higher credit rating) and 1 (lower credit rating) is respectively -8.40% (Table 5) and -8.36% (Table 7). This difference is mostly caused by a strong positive reaction for groups with a weak credit rating (+7.35% for both models), whereas the high credit rating group responded respectively -1.04% and -1.01% throughout the same period. Lastly, when all events are combined, the difference in market reaction as measured in CAAR between group 0 and 1 is respectively -8.71% and -7.59%.

The aforementioned empirical evidence demonstrates that banks with a lower credit rating significantly outperformed banks with a high credit rating throughout the event period. These results support hypothesis 3 and are somewhat consistent with Bowen, Khan Rajgopal (2010). However, they used individual variables such as liquidity and capitalization, rather than a comprehensive measure. Moreover, their results were more modest, achieving a maximum significance level of 90%. The results can arguably be compared to Bischoff, Brüggemann and Daske (2010) as well. They concluded that the fair value accounting amendments resulted in short-term benefits versus long-term costs. They argued that the fair value amendments resulted in a reduced transparency, which had long-term negative

27

consequences. The empirical findings in this study suggest that there are only benefits for banks with a lower credit rating. The trade-off between the costs of reduced transparency and the benefits from increased discretion (as allowed for by the fair value amendments) seems to only favor the lower credit rating group. This suggests that the results for event 5-7 yield additional support for hypothesis 1 as well. If fair value was perceived as harmful for banks due to the alleged pro-cyclicality effect (SEC, 2008), then banks that have a weaker credit rating should benefit most from a fair value accounting relaxation. These banks will benefit more from reclassifying assets out of the fair value category, primarily because they will be less pressured to sell financial assets below the fundamental value. Whereas, for the higher credit rating banks that are not pressured to engage in to fire sales, there will be no ‘benefit’ from reclassifying assets out of the fair value category. Investors are unlikely to benefit since these financially stronger banks were less compelled to sell assets below their fundamental value. This in turn, is clearly visible in the results, which illustrate that weaker banks outperformed the stronger banks throughout the event period. Hence, investors perceived fair value accounting to harm weaker banks; this suggests that there is a pro-cyclicality effect incorporated in to fair value accounting, which is consistent with the findings from Bowen, Khan and Rajgopal (2010).

[Insert Table 6 and 7 here]

Lastly, tests are performed to measure whether the weaker banks responded more strongly to the fair value accounting amendments throughout the entire period. This should be interpreted in two ways. The first one is that individual banks with a low credit rating responded more strongly as in terms of sensitivity (e.g. all abnormal returns are modified to be positive, which sample group responds more strongly on average). The second one is more specified, in order to evaluate the trade-off between benefits for the weaker credit group versus costs for the stronger credit rating group. Only the results (the mean computed from the CAAR for all firms, throughout the entire event period) are modified to both have a positive signal. For both instances, the response is measured throughout the entire period, because this incorporates the entire market response regarding the fair value accounting amendments. The first method, showed a total CAAR (based on a modified positive sign) of 46% for group 0 and 42.3% for group 1, the difference was 3.7% and insignificant (t = 0.6260, not tabulated). Apparently, when evaluated on a daily basis, there is no support that the lower credit rating group responds more strongly than the higher credit rating group.

28

The second method used the two-sample t test with manual data entry. The combined results for the entire period of the stratified sample are used as input (Table 5 panel B, mean group 0 and 1). Table 5 already indicated that the benefits for the weaker credit rating group (+4.60%) seem to exceed the costs for group the higher credit rating group (-4.11%). The difference between the modified CAAR (+4.6% and +4.11%) is 0.49% and is not significant (t = 0.1584, not tabulated). There is therefore no evidence that supports hypothesis 4. It seems that, on average, the benefits for the lower credit rating group didn’t exceed the costs for the higher credit rating group, or the other way around. Regardless, based on these findings there are still arguments to support the decision to amend the fair value accounting legislation. Primarily because the relief in down-side risk for the lower credit rating group is arguably more valuable than the harm done to the higher credit rating group. Imagine a scenario in which a troubled bank gets more breathing room and uses this flexibility to guarantee its coexistence. The fair value accounting amendments may have saved some banks from tumbling over, or at least be in a less immediate state of turmoil. If pro-cyclicality was an issue, as the prior results have suggested, then the fair value accounting amendments seem to have been an appropriate answer. Despite this, it is acknowledged that more stable banks (higher credit rating group) may have been punished by the fair value accounting amendments. While they may have been better managed, or had a more conservative approach that allowed them to earn a higher credit rating.

4.4 Trading volume reaction

The prior sections provided evidence regarding significant stock price reactions surrounding the fair value relaxation announcements. Kim and Verrechia (1991) argue that these abnormal stock price reactions should be accompanied by abnormal trading volume reactions. Chen et al (2002, p. 171) provided further evidence that is consistent with this notion. The results in this study are therefore in line with the prior literature, as shown in table 8, which reports the average abnormal trading volume (AAV) surrounding the fair value relaxation announcements. Significant reactions were found for each event, taking in to account the estimation window (-1, 0 and +1). The strongest reaction was found the day before the IASB decided to adopt the reclassification amendments to IAS 39 and IFRS 7. The AAV on event 6 (day -1) is positive (1.59) and statistically significant at the 1% level (t = 3.5986). This is a very strong response, comparatively; Landsman and Maydew (2002) reported an AAV of 0.95 on day 0. When interpreting these results, it is important to bear in mind that the formula scales the results by the standard deviation (σi), contrary to the AAR