Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Net or Not, Does it Matter?1 Xu JIANG, Mingzhu LI and Woody Y. WU This paper investigates the economic consequences of banks’ disclosure of gross derivatives position as required under Statement of Financial Accounting Standard (SFAS) No. 161, “Disclosure about derivative instruments and hedging activities”. Using a sample of U.S. banks that recognize the net value of derivatives but mandatorily disclose the gross value of derivatives in the footnotes after SFAS No. 161, we conduct the value relevance test of gross derivatives and find that it provides significant explanatory power for bank share price beyond that provided by the recognized net value of derivatives. Further, using a sample of 1,672 bankquarter observations of 74 U.S. bank-holding companies during the 2006 to 2011 period, we find evidence that the mandatory disclosure of gross derivatives required by SFAS No. 161 results in a significant decrease in information asymmetry for banks that do netting practice towards derivatives. We also find that the decrease in information asymmetry is more pronounced among U.S. banks that do netting relative to European banks which have more stringent requirements for netting. Taken together, these results suggest that the disclosure of the gross value of derivatives is value relevant and reduces information asymmetry, which are also consistent with the goal of SFAS No. 161 that aims at improving the transparency of derivative instruments. Keywords: FAS No.161, Derivative Disclosure, Value Relevance, Information Asymmetry 1. Introduction The global OTC derivatives market has grown dramatically since late 1990’s and is blamed for one of the main causes of the 2008 financial crisis (Volcker, 2011). Many investors complain that banks' exposures are opaque, making it difficult to determine exactly how safe a lender is as accounting treatment for derivatives differs sharply, with netting allowed for most derivatives in the United States but not in the Europe. 2 Leaders of the top 20 world economies (G20) have been pushing rule-makers to iron out accounting differences (G20 Declaration, November 15, 2008), 3 and in subsequent summits urged the International Accounting Standards Board (IASB) and Financial Accounting Standards Board (FASB) to complete their convergence projects by June 1 We acknowledged helpful comments from Gary Biddle, Jeff Callen, Zhaoyang Gu, Xiaohong Liu, Chul Park, Suresh Radhakrishna and workshop participants at City University of Hong Kong, Hong Kong Polytechnic University, Jinan University, Shanghai University of Finance and Economics, Southwest Jiaotong University, and the University of Hong Kong. Xu Jiang (xu.jiang@duke.edu) is at the Fuqua School of Business, Duke University. Mingzhu Li (mzli@baf.cuhk.edu.hk) and Woody Y. Wu (woody@baf.msmail.cuhk.edu.hk) are at the Chinese University of Hong Kong. 2 See section 2 for more background information. 3 Another G20 mandate is to force large users of swaps, such as banks, to process trades through clearing houses, which guarantee trades between two parties if one defaults. It has become effective from March 11,2013 in the US (“US and Europe launch derivatives reform”, March 10, 2013, Financial Times). 1 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 2011. Responding to the G20 mandate, the IASB and the FASB proposed to converge the derivative disclosure standard and eliminate the discrepancies in the offsetting treatment. on January 28, 2011 (FASB, 2011, IASB 2011). The proposals were strongly opposed by the Wall Street banks as they believe these would only ―exaggerate risks‖. Subsequently, FASB decided to be apart from IASB. This study aims to provide some evidence about this unsettled issue by examining the market consequences of gross value derivative disclosure required by Statement of Financial Accounting Standard (SFAS) No. 161, ―Disclosure about derivative instruments and hedging activities‖. Recently, Beyer (2013) theoretically documented that netted fair value accounting numbers result in a loss of information than the gross value. We provide initial empirical evidence on the incremental ability of gross value disclosure to explain stock price and bid-ask spreads. Collectively, We find that for a sample of bank-holding companies (hereafter referred to as banks) that mandatorily disclose the gross value of derivatives, the valuation coefficients on the netting adjustments towards gross derivatives are statistically significant, whereas the valuation coefficients on the net value of derivatives are not significant. Furthermore, we carry out a difference-in-difference test around the mandatory disclosure of gross derivatives by SFAS No. 161 and find that the test reveals a significant decrease in information asymmetry for banks that do netting practice towards derivatives.4 Overall, these results suggest that the disclosure of the gross value of derivatives is valuerelevant and decreases information asymmetry, which are also consistent with the objective of SFAS No. 161 to improve the disclosure transparency of derivative instruments. We first test the value relevance of gross value of derivatives. For banks that only disclose and recognize the net value of their derivatives, SFAS No. 161 forces them to disclose the gross value of derivatives in the footnotes, which is a huge amount relative to the net value. Of primary interest is whether these new disclosures are useful to investors in equity valuation relative to the recognized net value. Our value relevance test differs from prior research in two aspects. First, unlike prior studies, this paper examines the derivative assets and derivative liabilities separately rather than the aggregate value of derivatives. We partition the gross value into two parts, the net value and the netted amount of gross value. Second, while prior studies examine the fair value and notional value disclosure of derivatives, their primary focus is the derivatives used for purposes other than trading, that is, for hedging purpose. In this study, we investigate the gross value of derivatives used for both trading and hedging. Using a sample of banks that recognize the net value of derivatives and mandatorily disclose the gross value of derivatives in the footnotes after SFAS No. 161, we find evidence that the valuation coefficient on the netted amount of derivatives is statistically significant, whereas the valuation coefficient on the net value of derivatives is not significant. 4 Please note that banks that do netting practice towards derivative instrument refer to banks that disclose and recognize only the net value of derivatives prior to the implementation of SFAS No. 161. Such banks began to disclose the gross value of derivatives in the footnotes after SFAS No. 161 but still continue to recognize the net value of derivatives on their balance sheets. 2 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Furthermore, we find that the off-balance sheet derivatives for trading purpose are the main factor of the value relevance test. Our second test examines whether the disclosure of gross value of derivative instruments is informative to investors in the sense that it reduces the information asymmetry. Theoretically, more disclosure alleviates the adverse selection problem and decreases information asymmetries between informed and uninformed investors (Diamond et al. 1991). FASB also assert that disclosing fair value of derivatives on a gross basis better conveys information regarding how firms’ risks are managed to investors. Thus, we expect that the mandatory disclosure of gross amount of derivatives reduces the information asymmetry for banks that did not disclose the gross value until 2009. Skeptics of gross value disclosure, however, argue that the gross value of derivatives may mislead the investors and may have the unintended consequences of making the firm appear more risky. If their argument is valid, then it is the net value of derivatives that reflect the banks’ true underlying risk. The gross value would then at best be very noisy information and at worst further increase information asymmetry. To differentiate between those two opposing arguments, we do the difference-in-difference analysis using a sample of 1,672 bank-quarter observations of 74 U.S. banks during the 2006 to 2011 period. We use U.S. banks that self-selected to disclose and recognize the gross value of derivative instruments prior to SFAS No. 161 as a control sample. We also use a group of large European banks as another control sample. We assume that European banks are not able to do much netting because of the stringent requirement of IFRS regarding netting. We find that banks that recognize the net value of derivatives but were forced to disclose the gross value because of SFAS No. 161 have their information asymmetry among investors significantly reduced in the post-SFAS No. 161 period. These results suggest that the gross value of derivative contains valuable information and plays an informational role instead of a ―noisy‖ role in the capital market, which is consistent with the goal of SFAS No. 161 to improve the transparency of derivative instruments disclosure. Finally, we perform sensitivity test and find that the results are robust to (1) controlling for the gross notional amount of derivative assets and derivative liabilities; (2) correcting for the self- selection bias that banks choose to do netting practice. Overall, we conclude that disclosure of gross value derivatives as required by SFAS No. 161 provides useful information to investors and mitigate the information asymmetry problem. This paper makes three primary contributions to the literature. First, to the best of our knowledge, this is the first study to investigate the economic consequences of disclosure of gross value of derivatives introduced by SFAS No. 161. We add to the debate on the disclosure of netting adjustments by providing empirical evidence on the valuation and information asymmetry perspective. Our results suggest that the disclosure of fair value of derivatives in a gross basis improves the financial reporting transparency of financial institutions and helps investors to better understand how derivatives are managed and used. This evidence, however, should be interpreted with caution, as the above documented association can be driven by unobservable 3 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 characteristics of the derivatives or an imperfect control for bank characteristics, rather than by gross value disclosure to market per se. Second, this study contributes to the ongoing debate regarding the disparity in the derivative instrument presentation between current IFRS and the U.S. GAAP. The findings of this study highlight the importance of the gross value disclosure of derivatives in both useful for valuation as well as reducing information asymmetry. To some extent, the results in this study also provide implications for future accounting rules regarding derivative disclosure for regulators and standard setters. Finally, this study complements the extant value relevance literature of derivative instruments. Prior studies only focused on the derivatives for hedging purpose and were often based on the aggregate fair value of derivatives. This study focuses on the fair value of all derivatives including trading and hedging purpose and disaggregates the derivative that represents assets or liabilities position. The research design directly reflects the different properties of each component of the derivative instruments. The rest of the paper is organized as follows. Section 2 provides background information regarding accounting for the practice of netting derivatives. Section 3 reviews the literature and develops the hypothesis. Section 4 defines the variables and provides the research methodology. Section 5 describes the sample and data. Section 6 presents the empirical results. Section 7 concludes the paper. 2. Background Information and Institutional Details 2.1. Development of OTC Derivatives Market The use of derivative instruments has increased dramatically over the past two decades. According to the Bank for International Settlements, the nominal or notional amounts outstanding, defined as the gross nominal or notional value of derivatives on all types of risks concluded and not yet settled on the reporting date, have grown from $80.3 trillion in 1998 to $647.8 trillion by 2011 and peaked at $706.9 in June, 2011 (Panel A, Table 1) . Bank for International Settlements defines gross market values as the sum of the absolute values of all open contracts with either positive or negative replacement values evaluated at market prices prevailing on the reporting date. They supply information about the potential scale of market risk in derivatives transactions. Furthermore, gross market value at current market prices provides a measure of economic significance that is readily comparable across markets and products. As shown in Panel A of Table 1, the gross market value also increased substantially from $3.2 trillion in 1998 to $27.3 trillion in 2011. It reached $35.3 trillion in the second half of 2008, due to the sharp asset price movements following the bankruptcy of Lehman Brothers in September 2008. Gross market values declined quite rapidly thereafter as asset prices moved closer to their pre-crisis values, but have increased again since the first half of 2010 as markets went through another bout of turbulence in European crisis. 4 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Panel B of Table 1 indicates that most of the derivatives are foreign exchange and interest rate contracts. A total of 78% of notional amounts outstanding are from interest rate contracts and 10% are from foreign exchange contracts. Similarly, 73% of gross market value is from interest rate contracts and 9% is from foreign exchange contracts. It seems that derivatives have become a great source of revenue for banks (Aubin 2011). Mr. Volcker, appointed by President Barack Obama as the chair of the President's Economic Recovery Advisory Board on February 6, 2009 to advise the Obama Administration on economic recovery matters, argued that the vast increase in the use of derivatives, designed to mitigate risk in the system, had produced exactly the opposite effect (Volcker, 2011). Many investors complain that banks' derivative exposure is opaque, making it difficult to determine exactly how safe a lender is as accounting treatment for derivatives differs sharply, with netting allowed for most derivatives in the United States but not under IFRS. The issue has come under scrutiny by regulators worldwide since the global financial crisis (Basel Committee, 2009). Leaders of the top 20 world economies have been pushing rule-makers to iron out accounting differences (G20 Declaration, November 15, 2008). In September 2009, the G20 representatives required that global standard setters should ―make significant progress towards a single set of high quality global accounting standards.‖ The Financial Stability Board’s progress report stated: ―Moreover, continuing differences in accounting requirements of the IASB and FASB for netting/offsetting of assets and liabilities also result in significant differences in banks’ total assets, posing problems for framing an international leverage ratio .‖ (Financial Stability Board, 2009) 2.2. The Legal Issue of Netting 2.2.1 An Introductory Example Suppose that Bank GS has entered into a large number of derivative transactions with Insurer AG, and expects to receive $2.45 billion from AG in respect of in-the-money transactions and to pay $2.475 billion to AG in respect of out of-the-money transactions. AG discovers that its trader, Dick Leeson has made unauthorized trades with crippling exposures to third parties. Insolvency proceedings commence promptly against AG and it is not expected that there will be any significant distribution for AG’s creditors. GS faces at least two possibilities. The first is that the gross amounts of $2.45 billion and $2.475 billion will be netted against each other to produce a single net amount of $25 million which GS owes to AG. This would leave GS broadly in the same economic position it would have been in had AG remained solvent. The second possibility is that GS will have to pay $2.475 billion in full to AG, or to AG’s representative in insolvency, and claim as an unsecured creditor to recover $2.45 billion in AG’s insolvency. The AG’s insolvency will have converted GS’s net liability of $25 million into a gross exposure of $2.475 billion, which would only be reduced, if at all, 5 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 by whatever distribution GS ultimately receives in AG’s insolvency as an unsecured creditor for $2.45 billion. The second scenario could make GS insolvent. A domino effect might ensue, in which the insolvency of one institution (AG) might lead to insolvencies of other financial institutions such as GS. Thus the first possibility, in which GS owes a single net amount of $25 million, is good for the security of financial markets generally. Under what circumstance can GS report on its balance sheet in this way? 2.2.2 Close-out Netting The International Swaps and Derivatives Association (ISDA) is a trade organization of participants in the market for over-the-counter derivatives. Initially created in 1982, it has been credited for helping defeat US Congressional efforts to regulate derivatives in 1994 and again in 1998. In 1992, the ISDA published a standardized contract (the ISDA Master Agreement), typically used between a derivatives dealer and its counterparty when discussions begin surrounding a derivatives trade. In response to market difficulties in the late 1990s, a second edition was published in 2002. It is considered ―fundamental to, and provides a template for, the derivatives market.‖ (Ishmael 2009). Possibly the most important aspect of the ISDA Master Agreement is that it allows the parties to aggregate the amounts owed by each and replace them with a single net amount payable by one party to the other. Netting, dealt with under section 2(c) of the ISDA Master Agreement, allows the parties to net out amounts payable on the same day and in the same currency. The more important use of netting is close-out netting under Section 6(e) of the ISDA Master Agreement. Pursuant to this section, when an ISDA Master Agreement is terminated (normally following a credit event), the value of each of the Terminated Transactions is assessed and converted into the Termination Currency and any outstanding Unpaid Amounts are taken into account. 5 The Settlement Amount and Unpaid Amount are added up and a single figure in the Termination Currency is determined payable by one party or the other. In the preceding example, the successful application of close-out netting to the open transactions between GS and AG should yield a single net amount of $25 million payable by GS to AG. 5 Large users of swaps in the US, such as banks, will be required from March 11,2013 to process trades through clearing houses as the industry is forced to comply with a mandate agreed by the G20 more than three years ago. Regulators are pushing for more trades to move on to electronic trading venues and be processed through clearing houses, which guarantee trades between two parties if one defaults. (“US and Europe launch derivatives reform”, March 10, 2013, Financial Times) 6 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 2.3. Financial Reporting of Netting 2.3.1 Netting under U.S. GAAP The netting adjustments are the amount of derivatives companies has netted under legally enforceable master netting agreements and cash collateral receivable and payable with the same counterparty in accordance with ASC Subtopic 210-20, Balance Sheet – Offsetting (formerly FASB Interpretation No. 39, ―Offsetting of Amounts Related to Certain Contracts,‖).6 FIN-39 identifies four conditions that must be met for the right of setoff to exist: (1) Each of the two parties owns the other determinable amounts. (2) The reporting party has the right to set off the amount owned by itself with the amount owned by the other party. (3) The reporting party intends to set off. (4) The right of setoff is enforceable by law. Note that even though the right of ―setoff‖ is a ―legal‖ term, for accounting purpose, the right of setoff does not exist unless the reporting entity intends to set off the two amounts. Essentially, offsetting is generally an accounting option when certain conditions are met, rather than an accounting requirement. In other words, companies have the choice to offset assets against liabilities if companies meet the four criteria. However, there are two important exceptions that allow companies to offset assets against liabilities even if they don’t intend to settle on a net basis. One is for derivatives subject to a master netting agreement with the same counterparty and the other is for netting of repos and reverse repos that meet certain criteria. The motivation for companies to do a netting adjustment is to reduce credit, settlement and other counterparty risks of financial contracts by aggregating (combing) two or more obligations to achieve a reduced net obligation. Offsetting can affect the firms’ key financial ratios and improve firms’ perceived profitability and liquidity. Before FASB introduced SFAS No. 161, banks selectively disclosed their derivatives either on a net basis or a gross basis. As permitted under U.S. GAAP, banks can net derivative assets and liabilities, and the related cash collateral received and paid, when a legally enforceable master netting agreement exists between the firm and the same counterparty. Derivatives are recognized on a net-by-counterparty basis when a legal right of setoff exists under an enforceable master netting agreement (MNA). This practice is called netting and has a significant impact on the stock price and bid-ask spread, as examined in this study. For example, Bank A had an interest rate risk derivative with a $100 asset fair value and a foreign currency exchange risk derivative with a $60 liability fair value that was subject to master netting arrangements with the same counterparty Bank B. Bank A would report those derivatives a $40 net asset on the balance sheet in the event of bank B defaulting on the contract. Generally, some banks voluntarily disclose and recognize the gross value of derivative, even if the derivative instruments are subject to the MNA, e.g., Fifth Third Bank. On the other hand, some banks that disclosed and recognized only the net amount of derivative prior to 6 Arthur Levitt, the former Chairman of SEC, raised question about the FASB interpretations in his March 9, 2007 Wall Street Journal editorial-page commentary: “FAS 133, for example, deals with the accounting for derivatives. When it was first proposed, the standard was significantly simpler and easier to understand and -- we expect -- to apply. Yet, as different interests asked for exceptions to the rule, it metastasized into an 800-page treatise of rules and interpretations that continues to grow with each passing month.” 7 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 SFAS No. 161 mandatorily disclose the gross value after SFAS No. 161 in the footnotes, although they still recognize the net value of derivatives after SFAS No. 161. For example, the gross amount of derivatives of Bank of America is more than $1,000 billion in the past two years and only 5 percent of it is recognized on its balance sheet. One could imagine the substantial impact on the financial reports if Bank of America recognizes the gross amount of derivatives on its balance sheet. Therefore, netting practice could lead to systematic and significant underestimate of the size of derivatives positions that banks hold. More importantly, there would be a lack of understanding about the underlying type of risk being managed since the netting practice is based on the counterparty basis. As a result, the gross amount of derivative may better convey the information of how the derivative instrument used and managed. This paper focuses on the netting adjustment towards derivative assets and derivative liabilities. Generally, the fair value of derivative assets should not be offset against derivatives liabilities unless the four criteria for offsetting are met. However, when derivatives are entered into with the same counterparty under a master netting agreement, even if the reporting entity does not have the intent to settle on a net basis, the reporting entity may offset the fair value amounts recognized for derivative instruments and the fair value amounts for the right and obligation to reclaim the cash collateral, namely the payable or receivable. The master netting agreement provides one party the right to terminate the entire arrangement and demand the net settlement of all contracts, which is conditional on the event of default or bankruptcy of the other party (FIN 39, pars. 21 and 30). 2.3.2 Netting under IFRS Ever since the introduction of IFRS in Europe, the offsetting of financial assets and liabilities on the balance sheet has been a controversial issue. The ability to offset under IFRS is limited in comparison with U.S. GAAP, especially for derivatives traded with the same counterparty under an ISDA Master Netting Agreement (MNA). Historically, the Europe-based International Accounting Standards Board (IASB) has permitted significantly less balance sheet offsetting than has the FASB. Under IFRS, assets are required to be offset against liabilities when the company has the legally enforceable right to set-off and intends to settle on a net basis. Although seemingly similar to that in the U.S. GAAP, there are two major differences between the two standards regarding netting. First, offsetting is optional under U.S. GAAP while it’s mandatory under IFRS. Second, IFRS do not provide exceptions for derivatives and repos in which there is no intent to offset as does U.S. GAAP. As a result, most derivatives don’t meet the criteria to be offset under IFRS and are recognized on a gross basis on the balance sheet. The difficulty to determine exactly how safe a lender is due to different accounting treatment, with netting allowed for most derivatives in the United States but not under International Financial Reporting Standards (IFRS), is best illustrated by Dr. Josef Ackermann, Chairman of the Management Board of Deutsche Bank in his presentation, ―Financial Transparency‖, in Montreal and Toronto during February, 19-20, 2009. Appendix B shows that, Deutsche Bank had around €2.2 trillion of assets on the 8 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 balance sheet at the end of 2008 under IFRS, a quarterly increase of €140 billion. But under the US GAAP, its balance sheet would only show €1.03 trillion of assets, a quarterly decline of 22%, €290 billion, with roughly the same level of equity. 7 2.3.3 The Convergence Project ―The fact that companies can, in some instances, report IFRS balance sheet figures that are double the size of their US GAAP numbers is not acceptable in global capital markets.‖ commented Sir David Tweedie, Chairman of the IASB. Responding to the requests from the G20 and the Financial Stability Board (FSB), the IASB and the FASB proposed on January 28, 2011 new rules for when companies would be allowed to offset financial assets and financial liabilities on the balance sheet (FASB 2011, IASB 2011). The boards proposed that offsetting should apply only when the right of set-off is enforceable at all times, including in default and bankruptcy, and the ability to exercise this right is unconditional, that is, it does not depend on a future event. 8 The entities involved must intend to settle the amounts due with a single payment or simultaneously. Provided all of these requirements are met, offsetting would be required. The proposals would converge IFRS and US GAAP and eliminate several industry-specific netting practices. According to Leslie Seidman, Chairman of the FASB, ―This proposal would change US GAAP to require netting in a narrower set of circumstances, but the effect of other forms of credit mitigation would be disclosed in the footnotes.‖ The proposed standards would bring U.S. rules closer to existing international rules and dramatically gross up U.S. balance sheets. The change in standard could bloat the balance sheet figures by as much as $7 trillion (Credit Suisse 2011).The proposals would eliminate the exception under US GAAP that allows offsetting for some arrangements in which the ability to offset is conditional and there is no intention to offset or the intention is conditional (KPMG 2011). However, the top U.S. banks expressed strong opposition to this proposal as they believed that these proposed accounting standards would only mislead the users of financial reports as they tend to ―obscure or create‖ risks which do not necessarily exist (Goldman Sachs, 2010). According to them, banks have a typical practice of netting or offsetting their derivatives exposure against each other that prevents them from being exposed to the gross amounts of losses. In a letter addressed to FASB, Robert Traficanti, deputy controller at Citigroup, said ―the flawed offsetting model‖ in the proposed accounting standards would only mislead the users of financial reports as they tend to ―obscure or create‖ risks which do not necessarily exist. 7 Appendix C provides an excerpt from ISDA “Netting and Offsetting: Reporting derivatives under U.S. GAAP and under IFRS” and this table summarizes the key differences for derivatives between current IFRS and U.S. GAAP. 8 Basel Committee for Banking Supervision (article 215, 2009) explicitly stated in its consultative document that “Consistent with taking a non-risk based approach and international comparability the proposed measure of exposure does not permit netting. This applies to netting of derivatives, repo style transactions, and the netting of loans against deposits”. 9 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 FASB eventually compromised with the top banks and decided to retain its current standard for banks to continue disclosing the net position of derivatives that are subject to master netting agreements and cash collateral. This compromise keeps the disparity between FASB and IFRS on how they offset the financial assets and financial liabilities, especially derivatives, on the main body of balance sheet. Given the highly controversial nature of the requirement and the fact that FASB itself reverses its decision, it is important to provide evidence regarding the economic importance of gross amount disclosure of derivatives that can help both regulators and academic researchers to evaluate the incremental usefulness of the gross value versus the net value of derivative instruments. 2.4. SFAS No. 161 In March 2008, the FASB issued SFAS No. 161 as an amendment to SFAS No. 133, after the board received concerns that the existing legislation did not provide adequate information about the effect of derivative contracts on the financial position and performance. FASB was concerned that disclosing information on a net basis could make it difficult to analyze (1) the risks being managed with derivatives and (2) the relationship between the fair values of derivatives and the associated gains or losses reported. FASB believe that disclosing the fair value amounts on a gross basis would help users understand how and why an entity uses derivative instruments. SFAS 161 was established to improve the transparency of derivative instruments disclosure. Two features of the disclosure stipulated under SFAS 161 are especially noteworthy. First, SFAS No. 161 requires banks to disclose the fair value of derivative instruments on a gross basis, even if the derivatives are qualified for recognized net value under FIN 39. Second, banks need to disclose the gross fair value amounts for assets and liabilities separately between derivatives that are designated for hedging and those that are not. Appendix D provides an excerpt from Goldman Sachs 2010 Annual Report, Note 7 Derivatives and Hedging Activities to illustrate netting practice under SFAS 161. SFAS No. 161 (ASC 815-10-50) requires companies to disclose the gross value of derivatives in the footnote. It has no impact on companies that elect to report gross value of derivatives on the balance sheet. However, it will significantly increase the derivative assets and derivative liabilities of companies that elect to report net value of derivatives on balance sheet. Specifically, assuming no netting, the financial companies in the S&P 500 could bring up to $6.9 trillion and $6.8 trillion of off-balance sheet derivatives assets and derivative liabilities onto their balance sheet. The assets and liabilities are highly concentrated among five companies - Bank of America, Citigroup, Goldman Sachs, J.P. Morgan, and Morgan Stanley, which accounts for 97% of the total amount of off-balance sheet derivative assets and liabilities (Zion et al. 2011). Bringing the huge off-balance sheet value of derivatives onto the balance sheet would increase the reported leverage ratio while decreasing return on assets. Consequently, it may imply that these five banks have more leverage and exposure to market risks than previously thought, which in turn could affect how investors and analysts value them. That probably explains why top banks strongly oppose using the gross presentation of their balance sheets. 10 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 3. Literature Review and Hypothesis Development 3.1. Value Relevance Accounting information is considered to be value relevant when it is associated with market value of equity (Barth et al. 2001). If a significant association is found, then we may infer that the accounting information is relevant to investors and reliable enough to be reflected in the share price. In this section, we discuss the literature related to the value relevance of derivative instruments measures. After FASB required mandatory disclosure of fair value of financial instruments in SFAS No. 107 and No. 119, several studies examine the value relevance of those fair value disclosures. Barth et al. (1996) and Eccher et al. (1996) employ a cross sectional valuation framework and find that fair value disclosures for investments and loans provide significant explanatory power beyond historical costs in valuing firms. Nelson (1996) finds that fair value disclosure of financial assets explains differences in market-to-book ratios. Venkatachalam (1996) documents that the fair value of derivatives helps explain cross-sectional variation in bank share price. These findings indicate that investors utilize fair value disclosures when valuing firms. SFAS No. 161 introduces the mandatory disclosure of fair value of derivatives on a gross basis. For banks that are forced to disclose the gross value but only recognize the net value of derivatives, prior literature shows that the net value of derivatives has a predictable association with stock price. The incremental explanatory power of gross value of derivatives for stock price, however, has not been examined. Ahmed et al. (2006) document that the recognized fair value of derivatives are value relevant while the disclosed fair value of derivatives are not. This paper only investigates the derivatives for risk management and does not distinguish whether banks practice netting or not. We develop our first hypothesis stated in null as followings, H1: The gross value of derivatives is not value relevant to stock price. 3.2. Information asymmetry Financial reporting and disclosure has been identified by prior studies as an important mechanism through which firms communicate firm-specific information to investors with the purpose of reducing information asymmetry between management and investors (Healy et al. (1999)). These disclosures are either mandatory or voluntary. Prior studies suggest that disclosure alleviates the adverse selection problem as well as information asymmetries between management and investors as well as informed and uninformed investors (Diamond et al. 1991, Lambert et al. 2007, and Leuz et al. 2000). Prior to SFAS No. 161, accounting standard allowed firms to disclose only their net derivative positions. This treatment prevented investors from fully unraveling the gross value of those positions. After SFAS No. 161, information regarding gross positions is available to investors. Following the implications from the theoretical literature, this mandatory disclosure requirement would result in a reduction in the information asymmetry between banks and their investors. In this study we use the bid-ask spread to proxy for the extent of information asymmetry. On the other hand, some practitioners believe that the disclosure of gross amounts can mislead investors by obscuring banks’ actual riskiness (Goldman Sachs 2010). They argue that the net amount measures 11 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 firms’ leverage and other key ratios more accurately while the gross amount serves as pure noise and may result in investors making incorrect conjecture. Given the preceding argument, if the disclosure of gross derivatives does provide relevant information and does not mislead investors, information asymmetry will be reduced; otherwise, it will be increased. Thus we develop our second hypothesis stated in null, H2: The gross value of derivatives does not affect the information asymmetry. 4. Research Design 4.1. Value Relevance Test To test the value relevance of gross value, we estimate the association between share price and gross value per share using a modified Ohlson (1995) model, which has been extensively used in the literature. Barth and Clinch (2009) provide evidence that share-deflated specifications (as opposed to equity book value-deflated, returns or equity market value-deflated specifications) perform the best in reducing scale effects in the modified Ohlson (1995) model. First, we run regression only using the recognized net value of the derivative instruments to confirm the findings in the prior literature that disclosure of net value is relevant to investors. In model (1), we express stock price as a function of net value derivative assets and derivative liabilities, on-balance-sheet non-derivative assets and liabilities and net income. Model (1): where PRC is the per share price measured at the end of the quarter, FVA is the aggregate fair value assets excluding the net value of derivative assets, FVL is the aggregate fair value liabilities excluding the net value of derivative liabilities, NETBV is the net book value of non-fair value assets and non-fair value liabilities, NDERA is the net value of derivative assets and NDERL is the net value of derivative liabilities, and NI is the net income. The independent variables are scaled by the outstanding shares. Second, to test the value relevance of disclosure of gross value, we replace the main independent variables, net value of derivative assets and liabilities, with the gross value of derivative assets and liabilities. Model (2): Where GDERA is the gross value of derivative assets and GDERL is the gross value of derivative liabilities. Third, we partition the gross value of derivatives into two parts, the net value of derivatives and the netted amount of derivatives. The latter part is the difference 12 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 between the gross value and net value of derivatives, which is the off-balance sheet amount of derivative exposure because of netting treatment. Model (3): + + (3) Where ADERA is the netted amount of derivative assets and ADERL is the netted amount of derivative liabilities. Finally, we further partition the off-balance sheet derivative assets and liabilities depending on whether the derivative instruments are designated as hedging purpose or trading purpose. Model (4): + + (4) Where ATDERA is the netted amount of derivatives assets for trading purpose, ATDERL is the netted amount of derivative liabilities for trading purpose.9 We correct stand errors and related t-statistics based on two dimensional clustering (i.e., banks and quarters) following Petersen (2009). 4.2. Information Asymmetry Test We use the relative bid-ask spread (SPREAD) to measure information asymmetry between informed traders and uninformed traders. Bagehot (1971) first discussed the relation between information asymmetry and the bid-ask spread, suggesting that more information asymmetry will result in a larger bid-ask spread. Bagehot’s intuition was subsequently modeled by Copeland and Galai (1983), Kyle (1985) and Glosten and Milgrom (1985). Ball et.al. (2011) define SPREAD as the quarterly average of the difference between the ask and the bid quotes scaled by the average of the ask and the bid, expressed in percentage terms. Specifically, ∑ Where is the number of months in quarter q for bank i for which closing monthly bids ( ) and closing monthly asks ( ) are available. We use SPREAD defined by Ball et al. (2011) to proxy the information asymmetry variable in this paper. 9 Since hedging purpose derivatives accounts for only a tiny amount of total derivatives, we do not include them in our main analysis. Untabulated test shows that adding the amount of derivatives for hedging purpose does not change our results qualitatively. 13 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 We explore the impact of mandatory disclosure of the gross value of derivatives on information asymmetry using a difference- in- difference design. Specifically, we regress SPREAD on an indicator variable Netting indicating whether banks do netting practice, a dummy variable POST indicating the time period (pre-versus post SFAS No. 161), the interaction between these two indicators, and a set of control variables. This research design allows us to investigate the change in information asymmetry in the pre- versus post- SFAS No. 161 periods for banks that engage in netting relative to the change for banks that do not over the same period. Using the banks that voluntarily disclose gross value as a control group helps to isolate the effect of mandatory disclosure of gross value of derivatives by differencing out possible confounding factors that change around SFAS No. 161’s implementation. The specific regression model of information asymmetry test is as follows, + (5) The choice of control variables follows the research design of Ball et al. (2011). The characteristics of banks’ balance sheet composition are associated with information asymmetry (Morgan, 2002; Flannery et al., 2004). We control for loans and leases (LOAN) and loan loss allowance for loan and lease (LLA), both scaled by market value of equity. Following Fahlenbrach and Stulz (2011), we also control for Tier 1 capital ratio (TIREONE). Firm and market characteristics such as banks size and stock liquidity are important determinants of the bid-ask spread (Stoll, 2000). We control for size (LNMVE) and stock liquidity using turnover (TURN). We also control for daily stock return volatility (RETVOL) and the inverse of the quarterly end closing stock price (PRCINV). 5. Data and Sample Descriptive Statistics Financial statement data for U.S. bank holding companies are from the Federal Reserve’s Consolidated Financial Statements for Bank Holding Companies (FRY-9C). Data on bid-ask spread and other microstructure variables of U.S. banks are from CRSP. The financial data of European banks are from Bankscope database and other microstructure data of European banks are from Datastream. Panel A of table 2 describes the sample selection process. We identify the Top 100 U.S. banks according to the amount of total assets at the end of 2009: Q1 in FRY-9C report. Then we exclude U.S. subsidiaries of foreign banks as well as private banks, resulting in 74 unique banks. Panel B shows the sample selection process of the top 43 European banks. Panel C of table 2 presents the composition of banks that do netting practice and that do not in the pre- and post- 2009 periods. The final sample with non-missing data for all variables covers the period from 2006: Q1 to 2011: Q4 and comprises 20 U.S. banks that do netting practice before 2009 with 198 observations pre-SFAS No. 161 and 230 observations post-SFAS No. 161; 54 U.S. banks do not do netting practice until 2009 with 629 observation pre-SFAS No. 161 and 615 observation post-SFAS No. 161; and 14 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 43 European banks with 316 pre- and 404 post-SFAS 161. Appendix E shows all sample firms. Panel A of Table 3 presents the descriptive statistics for variables used to do the value relevance test for the sample of banks that did not report the gross value of derivative until 2009 in the post- SFAS No. 161 period, which are those banks that engage in netting but are forced to disclose the gross value. All variables are per share numbers. Since the gross amounts of derivatives item at FRY-9C are available from 2009:Q2, the sample here includes 189 bank-quarter observations for 20 banks from 2009:Q2 to 2011: Q4. The mean of the share price is 32.115. The average of the fair value assets (FVA) and the fair value liabilities (FVL) excluding derivatives are 146.310 and 47.921 respectively. The mean of the net book value of non-fair value assets liabilities (NETBV) is -65.244. The average net value for derivative assets (NDERA) and derivative liabilities (NDERL) are 15.901 and 12.152, respectively, while the means of the gross values of derivative assets (GDERA) and derivative liabilities are significantly larger, 183.935 and 167.045, respectively. The netted amount of derivatives are mostly concentrated in the adjustments towards derivatives for trading purpose, 167.524 for assets and 154.632 for liabilities 10 , while the adjustments towards derivatives for hedging purpose are quite small, 0.51 for assets and 0.26 for liabilities. The mean of the notional amount of derivatives is 9751.436, of which 9675.859 is the notional amount of derivatives for trading. Panel B of Table 3 presents the descriptive statistics for variables used in the information asymmetry test. The mean (median) of bid-ask spread for the U.S. banks that do netting is 0.174 (0.073), which is lower than the mean (median) for the U.S. banks not practicing netting, 0.196 (0.112), but higher than the mean (median) of 0.154 (0.091) for the European banks. Loans are more prevalent on the balance sheet of the European banks which have stringent requirements for netting. However, the U.S. banks exhibit great turnover and tier one ratio but lower stock return volatility than the European banks. Panel C of Table 3 shows the Pearson and Spearman Correlation across the variables used in the information asymmetry specification. 6. Results 6.1. Value relevance Tests To test the value relevance of the gross value of derivative assets and liabilities, we estimate the association between stock price and related gross value. In Panel A of Table 4, we use four model specifications to test H1 for banks that mandatorily disclose the gross value of derivatives post-SFAS 161. Column (1) first shows the association between the net value of derivatives and stock price. The coefficient on the net value of derivatives assets (NDERA) is significant (1.292 with t-value 2.38) whereas the 10 The netted amounts for assets (ATDERA) and liabilities (ATDERL) consist of two parts. One is the legally enforceable master netting agreements which have the same netted amount for assets and liabilities. The other part is cash collateral applied which are different for netted amounts towards assets and liabilities. Thus the total netted amount ATDERA and ATDERL do not show the same descriptive statistics in Panel A of Table 3. In the FR-Y9C database, only the total netted amount is available. Please refer to appendix D as an example. 15 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 coefficient on the net value of derivative liabilities (NDERL) is insignificant. In column (2), we replace the net value of derivatives with the gross value of derivatives. The coefficients on the gross value of derivative assets (GDERA) and liabilities (GDERL) are both significant with the correct sign (0.703 with t-value 6.29 and -0.719 with t-value 6.38, respectively), suggesting that the gross value derivatives are value relevant and thus H1 is rejected. Furthermore, we partition the gross value of derivatives into the net value and the related adjustment amounts towards gross value as shown in column (3). As expected, the netted amount of derivative assets and liabilities (ADERA and ADERL) due to netting practice is value relevant (0.505 with t-value 3.55 and -0.555 with t-value -4.01, respectively) and more importantly, they subsume the statistical significance of the net value of derivative assets and liabilities. Finally, we partition the netted amount of derivatives into these related to derivatives for trading purpose and hedging purpose. In column (4), the coefficient on the netted amount of derivative for trading purpose (ATDERA and ATDERL) remains significant. Figure 1 shows the total amounts of derivatives across 20 banks that do netting at the end of each quarter from the second quarter of 2009 to the fourth quarter of 2011. Panel A of Figure 1 plots derivative assets’ gross value versus net value between the full sample and the top five banks, Bank of America, Citigroup, Goldman Sachs, J.P. Morgan, and Morgan Stanley. One feature of this figure is that the gross value of derivative assets for the top five banks approximately accounts for 80 percent of the total gross amounts for the 20 banks in the sample. Another feature is that there is a substantial difference between the net value and the gross value for derivative assets, which highlights the huge impact of netting practice on the derivative assets. Panel B of Figure 1 shows the gross value versus net value for derivative liabilities, which depicts a similar pattern. Figure 1 not only points to the substantial amount of gross value relative to net value but also shows that the netting practice is mostly concentrated in the top five banks. It is thus possible that the results in Panel A of Table 4 are driven by the top five banks, rather than by the rest 15 banks. Therefore, we exclude the top five banks to check the robustness of results in Panel A. The main results remain unchanged, as shown in Panel B of Table 4, and further strengthen the value relevance results for the mandatory disclosure of gross value derivative instruments. 6.2. Information Asymmetry Test First, panel A of figure 2 presents the change in the bid-ask spread around the introduction of SFAS 161 from the first quarter of 2006 to the fourth quarter of 2011, where the solid line denotes spreads for U.S. banks that do netting practice (netting=1) and the dotted line plots spreads for U.S. banks that do not do netting practice (netting=0). There are three noteworthy points in this Figure. First, the solid line lies above the dotted line in the pre period (and significantly so), indicating that banks use netting practice had larger spreads than banks that did not. Second, there is a downward spike in both the solid line and the dotted line in the post-SFAS 161 period. However, the downward trend is more significant for banks that do netting practice. Third, the solid and dotted lines overlap significantly in the post-SFAS 161 period, indicating that the pre period difference in spreads between two groups disappears once SFAS No. 161 is introduced. These results imply that the change in spread is due 16 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 to the change in accounting treatment rather than other confounding factors. Second, using the same U.S. banks that do netting as those of panel A, panel B of figure 2 presents the change in bid-ask spread around the introduction of SFAS 161 between U.S. banks that do netting and European banks. U.S banks (netting=1) which are represented by solid lines shows similar pattern as what it has in figure 2(a), while European banks (netting=0) have relatively smooth pattern except for a sharp downward at the third quarter of 2008 due to the financial crisis. Table 5 presents the results of pooled regression analysis of the impact of SFAS No. 161 on bid-ask spread. We compare U.S. banks that do netting versus those do not as well as U.S. banks that do netting versus European banks. The European banks are assumed to be not doing netting because of the stringent requirements of netting practice by IFRS. Panel A reports the coefficients, t-statistics and two tailed p values of the regression model for the full sample period (2006:Q1 to 2011:Q4). To examine the relation between the spread and the netting practice (banks affected by the mandatory disclosure of gross value derivatives), we first combine some of the coefficients in Panel A and test the significance of the aggregated coefficients. Panel B and Panel C presents the restricted coefficients and the significance levels in a two by two analysis for the full sample period. The columns in Panel B and Panel C partition the sample by the pre- and post-SFAS 161 and the rows partition by Netting and Non-Netting groups. Panel B of Table 5 shows that U.S. banks that do netting practice experience a significant reduction in the spread after the mandatory disclosure of gross value of derivative instruments (0.376 versus 0.498, two tailed p<0.01). For U.S. banks that voluntarily disclose the gross value of derivatives in the pre-SFAS No. 161, their spreads also change significantly after the imposition of the mandatory gross value disclosure in 2009 (0.350 versus 0.396, two tailed p<0.01). One possible interpretation is that SFAS 161 highlights the importance of gross value disclosure and attracts greater attention from investors. More importantly, the change in spreads resulting from the mandatory accounting changes is significantly stronger for the Netting group versus the Non-Netting group (-0.122 versus -0.046, two tail p<0.01). Consistent with these results, the Netting group have a significantly higher spread than Non-Netting group in the pre-SFAS 161 (0.498 versus 0.396, two tailed p<0.01). Further, these cross sectional differences between the Netting and the Non-Netting group in the post-SFAS No. 161 become insignificant (0.376 versus 0.350, two tailed p>0.1). Panel C reports similar results when comparing U.S. banks that do netting with European banks that are subject to more stringent requirement to practice netting. To sum up, the results in Table 5 suggest that the mandatory disclosure is associated with a significant reduction in the spread for both the Netting and the NonNetting group, which suggests that the gross value of derivatives is informative to investors rather than noise and thus H2 is rejected. Moreover, the mandatory disclosure requirement levels the playing field among banks that do netting versus those that do not as the bid-ask spread between the two groups becomes insignificant in the postSFAS 161 period. 17 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 6.3. Sensitivity Tests The FASB also introduced mandatory disclosure of notional amount of derivatives. The empirical evidence is mixed on the usefulness of notional amount. 11 Venkatachalam (1996) finds that the notional amount of derivatives to be negatively related to the price and inferior to the explanatory power of fair values. On the contrary, Riffe (1997) reported that notional amounts of derivatives are positively related to the market value of equity. Wang et al. (2005) indicates that the notional amounts of derivatives are economically significant and provide incremental information beyond earnings and book value. Eccher et al. (1996) and Nelson (1996) control for notional principal amounts, but provide limited results pertaining to the information content of fair value disclosures for 1992-1993. Using a longer time series, Seow and Tam (2002) show that the notional amount of derivatives are not value relevant after including the other derivative disclosures for 1990-1996. In this section, we provide initial empirical evidence on the value relevance of the gross notional amounts. Using the preceding framework of the value relevance test, we add the gross notional amount of derivatives to the original regression model. Specifically, we partition the gross notional amounts of derivatives into those for derivatives of trading purpose and those for derivatives of hedging purpose. Panel A of Table 6 reports the value relevance results of gross notional amounts of derivatives. In columns (1), (2) and (3), the coefficients on the gross notional amounts of derivatives for trading and hedging purpose (NPTDER and NPHDER) are significant, indicating the value relevance of gross notional amounts of derivatives. More importantly, the significant value relevance of gross value derivatives remains the same as in prior main tests. Similarly, the Panel B of Table 6 excludes the top 5 banks and the results are also robust. Secondly, firms’ selection of netting practice is endogenous and this raises a selfselection problem. The netting practice enables banks to hide a substantial amount of the gross value of derivatives and the banks recognize a tiny portion of this gross value as net value on the balance sheet. As a result, the netting practice is adopted by banks that have a relatively substantial gross value. We correct for this self -selection problem using a two-stage approach. In the first stage, we estimate a probit model of banks choosing netting practice. In the second stage, we add the inverse Mills ratio from the first stage to the original spread model. In table 7, we only use the U.S. banks that do netting and U.S. banks that do not due to the data availability of European banks. Panel A of Table 7 reports the results of 11 According to the Bank for International Settlements, the nominal amounts outstanding provide a measure of market size and a reference from which contractual payments are determined in derivatives markets. However, such amounts are generally not those truly at risk. The amounts at risk in derivatives contracts are a function of the price level and/or volatility of the financial reference index used in the determination of contract payments, the duration and liquidity of contracts, and the creditworthiness of counterparties. They are also a function of whether an exchange of notional principal takes place between counterparties. Gross market values provide a more accurate measure of the scale of financial risk transfer taking place in derivatives markets. 18 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 the first stage. The coefficient on the notional amounts of derivatives (DERIV) is significantly positive, indicating that banks with more derivatives are more likely to use netting practice. Panel B of Table 7 presents the second stage results of changes in the bid-ask spreads around SFAS 161 with the inverse mills ratio (MILLS) from the first stage. We find that the coefficient on MILLS is significantly negative but does not subsume the statistical significance of the indicator variable Netting, which implies that the probit model may not fully capture the differences in economic characteristics between these two groups. However, the coefficient on the Netting*POST is significantly negative, which is consistent with H2 and suggests that U.S. banks that do netting experience an decrease in spread relative to banks that do not. Generally, the results are robust after controlling for the self-selection problem. 7. Conclusion Using data disclosed by the top U.S. bank holding companies we find that the mandatory disclosure of the gross value derivative is value relevant and reduces the information asymmetry as evidenced by the decrease in spread. These results indicate that the gross value of derivatives provide incremental power of value relevance tests and are informative to investors. As the first empirical study to compare the gross versus the net value of derivatives, this paper not only contributes to the recent debate over the gross versus net value of derivative disclosure and but also have significant policy implications for regulators and standard setters. References Aboody, D. 1996. Recognition versus Disclosure in the Oil and Gas Industry. Journal of Accounting Research 34:21-32. Ackermann, J. 2009. Financial Transparency. Presentation at Montreal and Toronto: 19-20. Ahmed, A. S., E. Kilic, and G. J. Lobo. 2006. Does Recognition versus Disclosure Matter? Evidence from Value‐Relevance of Banks' Recognized and Disclosed Derivative Financial Instruments. The Accounting Review 81 (3):567-588. Aubin, D. May 6, 2011. U.S. banks oppose derivatives accounting plan. Reuter. Ball, R., S. Jayaraman, and L. Shivakumar. 2012. Mark-to-Market Accounting and Information asymmetry in banks. Working paper, University of Chicago, Booth School of business. Basel Committee for Banking Supervision. 2009. Strengthening the resilience of the banking sector (http://www.bis.org/publ/bcbs164.pdf). Bank of International Settlements. 2010. Foreign exchange and derivatives market activity in 2010. Barth, M. E., W. H. Beaver, and W. R. Landsman. 1996. Value-Relevance of Banks' Fair Value Disclosures under SFAS No. 107. The Accounting Review 71 (4):513-537. ———. 2001. The relevance of the value relevance literature for financial accounting standard setting: another view. Journal of Accounting and Economics 31 (1–3):77-104. Barth, M. E., and G. Clinch. 2009. Scale Effects in Capital Markets-Based Accounting Research. Journal of Business Finance & Accounting 36 (3-4):253-288. Beatty, A. L., and B. Bettinghaus. 1997. Interest Rate Risk Management of Bank Holding Companies: An Examination of Trade-offs in the Use of Investment Securities and Interest Rate Swaps. Working paper, The Pennsylvania State University. Beyer, A. 2013. Conservatism and Aggregation: The Effect on Cost of Equity Capital and the 19 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Efficiency of Debt Contracts. Working paper, Stanford University. Copeland, T. E., and D. A. N. Galai. 1983. Information Effects on the Bid-Ask Spread. The Journal of Finance 38 (5):1457-1469. Diamond, D. W., and R. E. Verrecchia. 1991. Disclosure, Liquidity, and the Cost of Capital. The Journal of Finance 46 (4):1325-1359. Eccher, E. A., K. Ramesh, and S. R. Thiagarajan. 1996. Fair value disclosures by bank holding companies. Journal of Accounting and Economics 22 (1–3):79-117. Fahlenbrach, R., and R. M. Stulz. 2011. Bank CEO incentives and the credit crisis. Journal of Financial Economics 99 (1):11-26. FASB. 1991. Statement of Financial Accounting Standards No. 107 Disclosure about Fair Value of Financial Instruments. Stamford, CT: FASB. ———. 1994. Statement of Financial Accounting Standards No. 119 Disclosure about Derivative Financial Instruments and Fair Value of Financial Instruments. Stamford, CT: FASB. ———. 1998. Statement of Financial Accounting Standards No. 133 Accounting for Derivative Instruments and Hedging Activities. Stamford, CT: FASB. ———. 2008. Statement of Financial Accounting Standards No. 161 Disclosure about derivative instruments and Hedging activities an amendment of FASB No. 133. Stamford, CT: FASB. ———. 2011. Exposure Drafts on Offsetting Financial Assets and Financial Liabilities. Stamford, CT: FASB. FASB Interpretation No. 39, ―Offsetting of Amounts Related to Certain Contracts‖. Financial Stability Board. September 2009. Improving Financial Regulation (http://www.financialstabilityboard.org/publications/r_090925b.pdf). Flannery, M. J., S. H. Kwan, and M. Nimalendran. 2004. Market evidence on the opaqueness of banking firms’ assets. Journal of Financial Economics 71 (3):419-460. Glosten, L. R., and P. R. Milgrom. 1985. Bid, ask and transaction prices in a specialist market with heterogeneously informed traders. Journal of Financial Economics 14 (1):71-100. Goldman Sachs. Nov. 11, 2010. Comment letter to FASB, Re: Offsetting of Derivative Assets and Liabilities. Gros, D. 2010. Too interconnected to fail = too big to fail: What's in a leverage ratio? Working paper, Center for European Policy Studies. G20. Nov. 15, 2008. Declaration of the Summit on Financial Markets and the World Economy. Healy, P. M., A. P. Hutton, and K. G. Palepu. 1999. Stock Performance and Intermediation Changes Surrounding Sustained Increases in Disclosure. Contemporary Accounting Research 16 (3):485-520. Heckman, J. J. 1979. Sample Selection Bias as a Specification Error. Econometrica 47 (1):153-161. International Accounting Standard Board. 2011. Exposure Drafts on Offsetting Financial Assets and Financial Liabilities. London: IASB. International Swaps and Derivatives Association. 2012. Netting and Offsetting: Reporting derivatives under U.S. GAAP and under IFRS. Ishmael, S.-M. September 30, 2009. Lehman, Metavante and the ISDA Master agreement. Financial Times Alphaville. Knight, E., April 22, 2010. Reputational Risk May Turn Out the Largest of All, Financial Times. KPMG. February 2011. New on the Horizon: Offsetting financial assets and financial liabilities. Kyle, A. S. 1985. Continuous Auctions and Insider Trading. Econometrica 53 (6):1315-1335. Lambert, R., C. Leuz, and R. E. Verrecchia. 2007. Accounting Information, Disclosure, and the Cost of Capital. Journal of Accounting Research 45 (2):385-420. 20 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Leuz, C., and R. E. Verrecchia. 2000. The Economic Consequences of Increased Disclosure. Journal of Accounting Research 38:91-124. Levitt, A. March 9, 2007. Standards Deviation, Wall Street Journal. Mackenzie M., P. Stafford, and N. Munshi. March 10, 2013. US and Europe launch derivatives reform, Financial Times. Nelson, K. K. 1996. Fair Value Accounting for Commercial Banks: An Empirical Analysis of SFAS No. 107. The Accounting Review 71 (2):161-182. Ohlson, J. A. 1995. Earnings, Book Values, and Dividends in Equity Valuation. Contemporary Accounting Research 11 (2):661-687. Petersen, M. A. 2009. Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches. The Review of Financial Studies 22 (1):435-480. Riffe, S. 1996. The valuation of off-balance-sheet financial instrument disclosures in the banking industry. In Derivatives, Regulation and Banking: Advances in Finance, Investment and Banking Series Amsterdam, The Netherlands: North-Holland. Ryan, S. G. 2007. Financial Instruments and Institutions: Accounting and Disclosure Rules New York, NY: John Wiley & Sons, Inc. Seow, G. S., and T. Kinsun. 2002. The Usefulness of Derivative-related Accounting Disclosures. Review of Quantitative Finance & Accounting 18 (3):273. Venkatachalam, M. 1996. Value-relevance of banks' derivatives disclosures. Journal of Accounting and Economics 22 (1–3):327-355. Zion, D., Varshney A., and Burnap N. 2011. Grossing up the balance sheet. Credit Suisse. 21 Proceedings of 8th Annual London Business Research Conference Appendix A. Key Variable Definitions Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 FVA Aggregate fair value assets excluding net value of derivative assets, scaled by total shares outstanding FVL Aggregate fair value liabilities excluding net value of derivative liabilities, scaled by total shares outstanding NETBV Net book value of non-fair value assets and non-fair value liabilities, scaled by total shares outstanding NDERA Net value of derivative assets, scaled by total shares outstanding NDERL Net value of derivative liabilities, scaled by total shares outstanding GDERA Gross value of derivative assets, scaled by total shares outstanding GDERL Gross value of derivative liabilities, scaled by total shares outstanding ADERA Netting adjustments towards derivative assets, scaled by total shares outstanding ADERL Netting adjustments towards derivative liabilities, scaled by total shares outstanding ATDERA Netting adjustments towards derivative assets for trading purpose, scaled by total shares outstanding ATDERL Netting adjustments towards derivative liabilities for trading purpose, scaled by total shares outstanding AHDERA Netting adjustments towards derivative assets for other than trading purpose, scaled by total shares outstanding AHDERL Netting adjustments towards derivative liabilities for other than trading purpose, scaled by total shares outstanding NPDER Gross notional amount of all derivatives, scaled by total shares outstanding NPTDER Gross notional amount of derivatives for trading purpose, scaled by total shares outstanding NPHDER Gross notional amount of derivatives for other than trading purpose, scaled by total shares outstanding SPREAD The quarterly average of the monthly difference between the closing ask and the closing bid quotes, scaled by the average of the ask and the bid NI The quarterly income before extraordinary ,scaled by total shares outstanding NETTING An indicator variable equal to 1 if a bank does not report gross value of derivative until 2009, and 0 otherwise NON An indicator variable equal to 1 if a bank report gross value of derivative before NETTING 2009 POST An indicator variable equal to 1 if a bank-quarter observation falls in or after 2009, and 0 otherwise LOAN Total loans and leases (2122) scaled by the market value of equity LLA Loan loss allowance (3123) scaled by the market value of equity TIREONE The ratio of Tier 1 capital (8274) to total assets (2170) LNMVE The log of market value of equity at the end of the quarter TURN The log of the total number of shares traded (VOL) during the quarter divided by total shares outstanding (SHROUT) at the end of the quarter RETVOL The standard deviation of daily returns over the quarter PRCINV The inverse of the average stock price during the quarter DERIV The notional amount of derivatives during the quarter scaled by total assets B/M Book value of equity divided by market value of equity over the quarter SIZE The log of the total assets over the quarter LEV The ratio of total liabilities to total assets over the quarter 22 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Appendix B A slide from the presentation by Dr. Josef Ackermann, Chairman of the Management Board of Deutsche Bank, ―Financial Transparency‖, in Montreal and Toronto during February, 19-20, 2009. 23 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Appendix C: An excerpt from ISDA ―Netting and Offsetting: Reporting derivatives under U.S. GAAP and under IFRS‖. This table summarizes the key differences for derivatives between current IFRS and U.S. GAAP. 24 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Appendix D: An Excerpt from Goldman Sachs 2010 Annual Report, Note 7 Derivatives and Hedging Activities 25 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Appendix E: Sample firms U.S. banks that do netting practice 1. AMERICAN EXPRESS CO 2. REGIONS FINANCIAL CORP NEW 3. 4. 5. U S BANCORP DEL HUNTINGTON BANCSHARES INC KEYCORP NEW 6. BANK OF AMERICA CORP 7. 8. NORTHERN TRUST CORP P N C FINANCIAL SERVICES GRP INC 9. STATE STREET CORP 10. W HOLDING CO INC 11. T C F FINANCIAL CORP 12. NEW YORK COMMUNITY BANCORP INC 13. BANK NEW YORK INC 14. JPMORGAN CHASE & CO 15. CITIGROUP INC 16. MORGAN STANLEY DEAN WITTER & CO 17. WELLS FARGO & CO NEW 18. SUNTRUST BANKS INC 19. GOLDMAN SACHS GROUP INC 20. NEWALLIANCE BANCSHARES INC U.S. banks that do not take netting practice 1. ASSOCIATED BANC CORP 28. BANCORPSOUTH INC 2. 3. POPULAR INC BANK OF HAWAII CORP 4. 5. COMMERCE BANCSHARES INC SYNOVUS FINANCIAL CORP 6. 7. 8. CULLEN FROST BANKERS INC CITY NATIONAL CORP COMERICA INC 29. F N B CORP PA 30. FIRST CITIZENS BANCSHARES INC NC 31. SOUTH FINL GROUP INC 32. WEBSTER FINL CORP WATERBURY CONN 33. FIRST BANCORP P R 34. STERLING FINANCIAL CORP WASH 35. UNITED BANKSHARES INC 9. 10. 11. 12. 13. 14. TRUSTMARK CORP M & T BANK CORP FIFTH THIRD BANCORP FIRST HORIZON NATIONAL CORP CORUS BANKSHARES INC MARSHALL & ILSLEY CORP 36. 37. 38. 39. 40. 41. 15. B B & T CORP 16. U M B FINANCIAL CORP S V B FINANCIAL GROUP CATHAY GENERAL BANCORP B O K FINANCIAL CORP M B FINANCIAL INC NEW PACIFIC CAPITAL BANCORP NEW WINTRUST FINANCIAL CORPORATION 42. UMPQUA HOLDINGS CORP 43. U C B H HOLDINGS INC 17. VALLEY NATIONAL BANCORP 18. WILMINGTON TRUST CORP 44. PROSPERITY BANCSHARES INC 45. EAST WEST BANCORP INC 26 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 19. 20. 21. 22. 23. 24. 25. 26. 27. ZIONS BANCORP FIRSTMERIT CORP FULTON FINANCIAL CORP PA FIRST MIDWEST BANCORP DE 46. 47. 48. 49. PRIVATEBANCORP INC WHITNEY HOLDING CORP CAPITAL ONE FINANCIAL CORP INTERNATIONAL BANCSHARES CORP FRANKLIN RESOURCES INC 50. SANTANDER BANCORP NATIONAL PENN BANCSHARES INC 51. METLIFE INC SUSQUEHANNA BANCSHARES INC 52. UNITED COMMUNITY BANKS INC GA PA OLD NATIONAL BANCORP 53. C I T GROUP INC NEW CITIZENS BANKING CORP MI 54. INVESTORS BANCORP INC European Banks 1. Allied Irish Banks PLC 2. BNP Paribas 3. Banca Monte DEI Paschi 4. Banca Popolare DI Milano 23. 24. 25. 26. Dexia Erste Group Bank AG HSBC Holdings PLC ING Groep NV 5. 6. 7. 8. 9. 10. 11. 12. 13. Banca popolare Emilia Romagna Banco Bilbao Vizcaya Argentaria SA Banco Comercial Portugues Banco Espanol de Credito SA Banco Espirito Santo SA Banco Popolare Banco Popular Espanol SA Banco Santander SA Banco de Sabadell SA 27. 28. 29. 30. 31. 32. 33. 34. 35. Intesa Sanpaolo KBC Groep NV Lloyds Banking Group PLC Natixis Nordea Bank AB Raiffeisen Bank International AG Royal Bank Of Scotland Group PLC SEB 'A' SA SNS Reaal NV 14. 15. 16. 17. 18. 19. 20. 21. Bankinter SA Barclays PLC Commerzbank AG Credit Industriel Et Commerical Credit Agricole SA DNB ASA Danske Bank A/S Deutsche Bank AG 36. 37. 38. 39. 40. 41. 42. 43. Societe Generale Standard Charterted PLC Svenska Handelsbanken AB Swedbank AB UBS AG Unicredit Spa Unione DI Banche Italian Yapi Ve Kredi Bankasi AS 22. Deutsche Postbank AG 27 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Figure 1: Plot of Derivative Instrument Gross value versus Net value across full sample versus Top 5 banks This figure depicts the total amounts of derivatives across 20 banks that do netting at the end of each quarter from the second quarter of 2009 to the fourth quarter of 2011. Panel A plots derivative assets’ gross value versus net value between the full sample and the top five banks. Panel B shows the gross value versus net value for derivative liabilities, which depicts a similar pattern. Panel A: Derivative Assets (in billions) Derivative Assets : Gross vs. Net 9000 8000 7000 6000 5000 4000 3000 2000 1000 0 GDERA_Total GDERA_Top 5 NDERA_Total NDERA_Top 5 09Q2 5787 4644 480 369 09Q3 5742 4707 459 349 09Q4 5666 4785 391 296 10Q1 5495 4637 370 280 10Q2 6639 5659 413 308 10Q3 7349 6282 435 324 10Q4 5606 4745 370 276 11Q1 4943 4161 345 255 11Q2 5215 4401 345 251 11Q3 7793 6639 464 338 11Q4 7085 6024 401 295 11Q3 7537 6515 464 338 11Q4 6866 5921 401 295 Panel B: Derivative Liabilities (in billions) Derivative Liabilities : Gross vs. Net 8000 7000 6000 5000 4000 3000 2000 1000 0 GDERL_Total GDERL_Top 5 NDERL_Total NDERL_Top 5 09Q2 5485 4488 480 369 09Q3 5450 4559 459 349 09Q4 5411 4665 391 296 10Q1 5275 4544 370 280 10Q2 6399 5547 413 308 28 10Q3 7107 6169 435 324 10Q4 5405 4662 370 276 11Q1 4762 4093 345 255 11Q2 5028 4332 345 251 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Figure 2: Spread changes around the introduction of SFAS No. 161 This figure depicts the spread changes around the introduction of SFAS No. 161. The xaxis plots quarters from 2006: Q1 to 2011: Q4 around the introduction of SFAS 161. The y-axis plots average residual spreads, orthogonalized w.r.t controls and year effects. The solid (dotted) line denotes banks that do (do not) take netting practice. Panel A presents the change in the bid-ask spread around the introduction of SFAS 161 from the first quarter of 2006 to the fourth quarter of 2011, Panel B presents the change in bid-ask spread around the introduction of SFAS 161 between U.S. banks that do netting and European banks. Panel A: for U.S. banks that do netting and U.S. banks that do not 0.4 Netting 0.3 No Netting 0.2 0.1 -0.1 06Q1 06Q2 06Q3 06Q4 07Q1 07Q2 07Q3 07Q4 08Q1 08Q2 08Q3 08Q4 09Q1 09Q2 09Q3 09Q4 10Q1 10Q2 10Q3 10Q4 11Q1 11Q2 11Q3 11Q4 0.0 -0.2 -0.3 Panel B: for U.S. banks that do netting and European banks that do not 0.4 Netting 0.3 No Netting 0.2 0.1 -0.1 06Q1 06Q2 06Q3 06Q4 07Q1 07Q2 07Q3 07Q4 08Q1 08Q2 08Q3 08Q4 09Q1 09Q2 09Q3 09Q4 10Q1 10Q2 10Q3 10Q4 11Q1 11Q2 11Q3 11Q4 0 -0.2 -0.3 29 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Panel A. Size in trillions of US dollars 1998/12 1999/06 1999/12 2000/06 2000/12 2001/06 2001/12 2002/06 2002/12 2003/06 2003/12 2004/06 2004/12 2005/06 2005/12 2006/06 2006/12 2007/06 2007/12 2008/06 2008/12 2009/06 2009/12 2010/06 2010/12 2011/06 2011/12 Notional Amounts Outstanding 80317 81458 88201 94037 95199 99755 111115 127564 141665 169658 197167 220058 257894 281493 297670 369906 414845 516407 595341 638725 598147 594495 603900 582655 601046 706884 647762 Gross Market Value 3231 2628 2813 2581 3183 3045 3788 4450 6360 7896 6987 6394 9377 10605 9749 10074 9691 11140 15813 20353 35281 25314 21542 24673 21296 19518 27285 Panel B. Types of derivatives, in trillions of US dollars, at the end of 2011 Foreign exchange contracts Interest rate contracts Equity-linked contracts Commodity contracts Credit default swaps Unallocated Grand Total Notional Amounts 63349 504098 5982 3091 28633 42609 647762 % of Grand Total 10% 78% 1% 0% 4% 7% 100% 30 Gross Market Value 2555 20001 679 487 1586 1977 27285 % of Grand Total 9% 73% 2% 2% 6% 7% 100% Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Table 2: Description of sample selection This table reports the sample selection for our sample of 74 U.S. banks and 43 European banks. # of firms Panel A: U.S. bank 1. Top 100 bank holding company sorted by total assets as of 2009Q1 in FRY-9C 2. Foreign bank subsidiaries in U.S. and private banks Total 100 (26) 74 # of firms Panel B: European bank 1. Top 100 European banks under IFRS sorted by total assets as of 2009Q1 in Bankscope Database 100 2. Private banks and banks with missing data (57) Total 43 Panel C: Composition of sample 3. U.S. Banks that not report gross value of derivative until 2009 (Netting = 1) 4. U.S. Banks that report gross value of derivative before 2009 (Netting = 0) 5. European Banks that report gross value of derivative before 2009 (Netting=0) Total 31 # of firms # of Obs. Pre2009 # of Obs. Post2009 20 198 230 54 629 615 43 316 404 117 1,143 1,249 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Table 3: Sample Descriptive Statistics and Correlation Matrix Panel A: Descriptive Statistics for the sample of 20 U.S. Netting Banks from 2009:Q2 to 2011:Q4 25th 50th Variables(Thousand) N Mean Std. Dev. Percentile Percentile Price/Share 189 32.115 32.886 11.500 25.620 75th Percentile 41.640 FVA/Share 189 146.310 260.261 23.350 64.121 145.373 FVL/Share 189 47.921 142.710 0.226 2.336 16.147 NETBV/Share 189 -65.244 112.555 -60.185 -24.100 -9.343 NDERA/Share 189 15.901 36.651 0.880 4.655 9.232 NDERL/Share 189 12.152 26.903 0.556 2.958 8.531 GDERA/Share 189 183.935 439.497 1.645 13.540 97.835 GDERL/Share 189 167.045 385.357 1.426 13.046 61.024 ADERA/Share 189 168.034 403.849 0.621 7.346 57.166 ADERL/Share 189 154.892 359.342 0.789 7.374 31.937 ATDERA/Share 189 167.524 402.989 0.130 5.921 57.166 ATDERL/Share 189 154.632 359.403 0.639 6.773 31.938 AHDERA/Share 189 0.510 3.174 0.000 0.000 0.028 AHDERL/Share 189 0.260 0.797 0.000 0.000 0.097 NPDER/Share 189 9751.436 23503.050 76.497 701.936 6507.194 NPTDER/Share 189 9675.859 23459.870 53.362 588.562 6277.938 NPHDER/Share 189 75.578 105.766 15.495 28.429 91.598 NI/Share 189 -0.279 11.133 0.118 0.489 0.846 32 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Table 3 (Continued) Panel B: Descriptive Statistics of main variables in bid-ask spread test among U.S. banks that do netting practice, U.S. banks that don’t do netting, and European banks. Netting=1 U.S. banks Stat. SPREAD LOAN LLA TIREONE LNMVE TURN RETVOL PRCINV Obs. Netting=0 U.S. banks Std. Dev. Mean Median 0.460 27.551 0.881 0.018 1.704 0.718 0.026 0.133 428 0.174 9.289 0.254 0.076 9.749 4.160 0.031 0.072 0.073 4.290 0.054 0.074 9.950 4.147 0.023 0.037 Std. Dev. 0.258 25.188 1.330 0.063 1.159 0.867 0.025 0.289 1244 Netting=0 European banks Mean Median Std. Dev. Mean Median 0.196 10.006 0.304 0.094 7.547 3.837 0.032 0.107 0.112 5.206 0.074 0.084 7.402 3.919 0.024 0.043 0.293 20.785 16.730 0.016 1.100 1.352 0.166 0.425 720 0.154 16.970 2.560 0.043 9.744 -1.729 0.182 0.177 0.091 12.132 0.252 0.042 9.697 -1.431 0.129 0.079 Panel C: Simple correlation of key variables in bid ask spread test, with Parson correlation below the diagonal and Spearman correlation above the diagonal. (1) Between U.S. banks that do netting and U.S. banks that don’t do netting. SPREAD Netting LOAN LLA TIREONE LNMVE SPREAD -0.179 0.474 0.378 0.068 -0.475 Netting -0.030 -0.132 -0.103 -0.250 0.554 LOAN 0.468 -0.012 0.909 0.203 -0.468 LLA 0.386 -0.018 0.945 0.231 -0.356 TIREONE -0.058 -0.141 -0.103 -0.100 -0.202 LNMVE -0.335 0.589 -0.403 -0.348 0.074 TURN 0.102 0.167 0.240 0.209 -0.014 0.009 RETVOL 0.456 -0.014 0.487 0.453 -0.044 -0.307 33 TURN 0.204 0.156 0.460 0.469 0.109 0.006 0.562 RETVOL PRCINV 0.522 0.463 -0.043 -0.063 0.594 0.690 0.557 0.639 0.109 0.211 -0.329 -0.541 0.679 0.353 0.505 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 PRCINV 0.398 -0.060 0.700 0.778 -0.090 -0.377 0.194 (2) Between U.S. banks that do netting and European banks that don’t do netting. SPREAD Netting LOAN LLA TIREONE LNMVE TURN -0.063 0.412 0.314 0.023 -0.506 -0.058 SPREAD 0.027 -0.542 -0.413 0.711 0.041 0.838 Netting 0.452 -0.156 0.803 -0.411 -0.413 -0.396 LOAN -0.008 -0.084 0.022 -0.157 -0.279 -0.306 LLA 0.033 0.691 -0.202 -0.038 -0.213 0.631 TIREONE -0.402 0.002 -0.485 0.139 -0.220 0.074 LNMVE -0.002 0.927 -0.131 -0.079 0.686 0.018 TURN 0.094 -0.483 0.355 0.000 -0.359 -0.135 -0.428 RETVOL 0.299 -0.146 0.210 0.009 0.041 -0.216 -0.139 PRCINV 34 0.457 RETVOL 0.254 -0.759 0.708 0.583 -0.532 -0.182 -0.541 0.284 PRCINV 0.322 -0.284 0.522 0.527 0.015 -0.410 -0.066 0.404 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Table 4: Value Relevance Test of U.S. Netting Banks This table reports results of value relevance tests of U.S. banks that do netting practice (Netting=1) in the post-SFAS No. 161 period (from 2009:Q2 to 2011:Q4). Panel A includes 189 bank-quarter observations of 20 banks. Panel B excludes the Top 5 banks, which includes 134 bank-quarter observations of 15 banks. Column (1) presents the OLS regression results of share price on net value of derivative assets and liabilities. Column (2) replaces the net value of derivatives with the gross value of derivatives. Column (3) partitions the gross value of derivatives into the net value and the netted amounts. Column (4) partitions the netted amounts of derivatives by derivatives for trading and hedging purpose. See the Appendix A for variable definitions. T-statistics are presented in parentheses. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively, based on two-sided p-values. Panel A: Value relevance tests of 20 U.S. Netting Banks (Dependent Variable = Price) Independent (1) (2) (3) (4) Variables FVA 0.163* 0.156** 0.203*** 0.197*** (1.95) (2.03) (2.64) (2.60) FVL -0.175 -0.215** -0.311*** -0.295*** (-1.63) (-2.21) (-3.09) (-2.99) NETBV 0.193* 0.175* 0.251** 0.243** (1.68) (1.67) (2.39) (2.33) NDERA 1.292** 0.708 0.720 (2.38) (1.35) (1.41) NDERL -0.702 0.389 0.332 (-1.15) (0.66) (0.57) GDERA 0.703*** (6.29) GDERL -0.719*** (-6.38) ADERA 0.505*** (3.55) ADERL -0.555*** (-4.01) ATDERA 0.496*** (4.02) ATDERL -0.544*** (-4.55) NI 0.155 0.175 0.128 0.128 (1.24) (1.50) (1.12) (1.14) CONSTANT 17.198*** 21.793*** 18.874*** 18.943*** (8.21) (11.94) (9.32) (9.53) Observations 189 189 189 189 Adj. R-squared 0.71 0.75 0.76 0.76 35 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Table 4 (Continued) Panel B: Value relevance tests of 15 U.S. Netting Banks (Dependent Variable = Price) Independent (1) (2) (3) (4) Variables FVA FVL NETBV NDERA NDERL 0.254*** (2.69) -0.276** (-2.32) 0.337** (2.60) -0.087 (-0.11) 1.365 (1.51) GDERA 0.109 (1.25) -0.136 (-1.26) 0.126 (1.05) 0.239*** (3.10) -0.328*** (-3.36) 0.339*** (3.18) 2.045*** (2.96) 0.775 (1.03) 1.163*** (5.91) -1.264*** (-5.81) GDERL ADERA 0.761*** (4.21) -1.020*** (-5.34) ADERL ATDERA ATDERA NI CONSTANT Observations Adj. R-squared 0.239*** (3.07) -0.296*** (-3.01) 0.339*** (3.15) 1.879*** (2.69) 0.969 (1.28) 0.076 (0.58) 18.985*** (7.84) 134 0.76 0.135 (1.09) 23.956*** (11.17) 134 0.78 36 0.011 (0.10) 15.738*** (7.18) 134 0.84 0.525*** (3.72) -0.759*** (-5.19) 0.013 (0.12) 15.491*** (6.99) 134 0.84 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 TABLE 5 Change in bid-ask spread around the implementation of SFAS No. 161 Panel A reports the pooled regression results of the impact of SFAS No.161 on bidask spread from 2006:Q1 to 2011:Q4. Column (1) includes 1,672 bank-quarter observations of 20 U.S. Netting banks (Netting=1) and 54 U.S. Non-Netting banks (Netting=0). Column (2) includes 1,148 bank-quarter observations for 20 U.S. Netting banks (Netting=1) and 43 European Non-Netting banks (Netting=0). Panel B presents two-by-two way analysis of U.S. Netting banks versus U.S. Non-Netting banks by aggregating the coefficients in Column (1) of Panel A. Panel C uses the coefficients in Column (2) of Panel A. See the Appendix A for variable definitions. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively, based on two-sided p-values. Panel A : Pooled regression (Dependent Variable = Spread) (1) (2) U.S. Netting banks U.S. Netting banks vs. vs. European Non-Netting banks U.S. Non-Netting banks Independent Variables tptpCoeff. stat value Coeff. stat value Constant 0.396 7.38 0.000 0.520 4.70 0.000 Netting(1) 0.102 4.24 0.000 0.218 4.08 0.000 POST(2) -0.046 -3.03 0.002 -0.015 -0.63 0.531 Netting*POST(3) -0.076 -2.65 0.008 -0.129 -3.39 0.001 LOAN 0.011 12.94 0.000 0.006 11.10 0.000 LLA -0.194 -10.45 0.000 0.000 0.62 0.534 TIREONE 0.105 0.91 0.365 0.076 0.12 0.908 LNMVE -0.021 -3.66 0.000 -0.050 -5.73 0.000 TURN -0.064 -6.67 0.000 -0.016 -1.96 0.051 RETVOL 4.478 2.93 0.000 -0.195 -2.68 0.007 PRCINV 0.292 7.10 0.000 0.227 8.14 0.000 Test: (2) + (3)=0 0.000 0.000 Test: (1) + (3)=0 0.281 0.126 Observations 1,672 1,148 2 Adj. R 0.38 0.31 37 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 TABLE 5(Continued) Panel B: Two-by-Two Analysis of U.S. Netting banks vs. U.S. Non-Netting banks, by Period, Using the coefficients in Column (1) of Panel A. Netting No Netting Diff. Pre-SFAS No. 161 (2006-2008) Post-SFAS No. 161 (2009-2011) 0.498 N=198 0.396 N=629 0.102*** 0.376 N=230 0.350 N=615 0.026 Diff. -0.122*** -0.046*** -0.076*** Panel C: Two-by-Two Analysis of U.S. Netting banks vs. European Non-Netting banks, by Period, Using the coefficients in Column (2) of Panel A. Netting No Netting Diff. Pre-SFAS No. 161 (2007-2008) Post-SFAS No. 161 (2009-2011) 0.738 N=198 0.520 N=316 0.218*** 0.594 N=230 0.505 N=404 0.089 38 Diff. -0.144*** -0.015 -0.129*** Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 TABLE 6 Sensitivity Test: value relevance This table reports sensitivity test results for main findings of value relevance tests in table 4 by controlling for the gross notional amount of derivative, other things being equal. See the Appendix A for variable definitions. T-statistics are presented in parentheses. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively, based on twosided p-values. Panel A: Value relevance tests of 20 U.S. Netting Banks (Dependent Variable = Price) Independent (1) (2) (3) (4) Variables FVA -0.072 -0.031 -0.040 -0.046 (-0.75) (-0.37) (-0.47) (-0.53) FVL 0.164 -0.067 -0.052 -0.005 (1.36) (-0.61) (-0.46) (-0.04) NETBV -0.122 -0.073 -0.078 -0.085 (-0.93) (-0.63) (-0.66) (-0.73) NDERA 0.026 -0.704 -0.532 (0.05) (-1.28) (-0.97) NDERL 0.856 1.554** 1.407** (1.28) (2.54) (2.29) GDERA 0.555*** (5.11) GDERL -0.608*** (-5.52) ADERA 0.551*** (4.05) ADERL -0.621*** (-4.62) ATDERA 0.482*** (3.37) ATDERA -0.544*** (-3.82) AHDERA -0.322 (-0.66) AHDERL 2.906 (1.16) NPTDER -0.0006* 0.0010*** 0.0008** 0.0005 (-1.92) (2.71) (2.01) (1.22) NPHDER 0.0758*** 0.0679*** 0.0782*** 0.0608*** (4.72) (4.85) (5.42) (2.81) NI 0.126 0.135 0.091 0.096 (1.05) (1.22) (0.84) (0.90) Constant 17.050*** 19.941*** 18.463*** 18.679*** 39 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 Observations Adj. R-squared (8.45) 189 0.83 (11.19) 189 0.84 (9.66) 189 0.86 (9.69) 189 0.86 TABLE 6 (Continued) Panel B: Value relevance tests of 15 U.S. Netting Banks (Dependent Variable = Price) Independent (1) (2) (3) (4) Variables FVA -0.024 -0.161 -0.069 -0.073 (-0.23) (-1.48) (-0.72) (-0.75) FVL -0.192 0.037 0.109 0.153 (-1.30) (0.19) (0.63) (0.87) NETBV -0.043 -0.233 -0.079 -0.081 (-0.30) (-1.58) (-0.61) (-0.62) NDERA -3.190*** 1.654 2.697** (-3.99) (1.49) (2.09) NDERL 3.636*** 1.013 0.259 (4.14) (1.01) (0.23) GDERA 0.708* (1.77) GDERL -0.826** (-1.98) ADERA 1.030*** (2.70) ADERL -1.301*** (-3.19) ATDERA 1.440*** (3.08) ATDERA -1.767*** (-3.47) AHDERA 2.166** (2.49) AHDERL -0.327 (-0.12) NPTDER 0.0032*** 0.0017 -0.0011 -0.0022 (3.96) (1.16) (-0.72) (-1.28) NPHDER 0.1001*** 0.0894*** 0.0761*** 0.0554** (5.73) (5.45) (4.94) (2.08) NI 0.014 0.103 0.017 0.027 (0.12) (0.89) (0.17) (0.27) Constant 20.471*** 23.081*** 17.090*** 17.167*** (9.46) (10.91) (8.40) (8.44) Observations Adj. R-squared 134 0.82 134 0.82 40 134 0.87 134 0.87 Proceedings of 8th Annual London Business Research Conference Imperial College, London, UK, 8 - 9 July, 2013, ISBN: 978-1-922069-28-3 TABLE 7 Sensitivity test: self-selection problem This table presents the sensitivity test results for main findings of information asymmetry tests in table 5 by correcting the self-selection problem. Due to data availability of European banks, here we only test the U.S. netting banks versus U.S. Non-netting banks. Using Heckman (1979) procedure, in the first stage, we examine the determinant of netting practice, as reported in Panel A. In the second stage, we include the Inverse Mills Ratio generated from the first stage in the regression model of table 5, as reported in Panel B. See the Appendix A for variable definitions. *, **, and *** denote significance at the 10%, 5%, and 1% levels, respectively, based on twosided p-values. Panel A: First Stage - probit model (Dependent Variable = Netting) Coeff. B/M 0.017* SIZE 0.408*** ROA -0.863 LEV -0.670 DERIV 1.039*** Constant -4.882*** Observations 1,672 2 Pseudo R 0.47 z-stat 1.84 9.61 -0.22 -0.88 8.61 -6.44 Panel B: Second stage – change in bid-ask spread around SFAS No.161 (Dependent Variable = Spread) Coeff. t-stat Constant 0.355*** 6.26 Netting 0.389*** 2.89 POST -0.051*** -3.36 Netting*POST -0.077*** -2.70 LOAN 0.010*** 12.20 LLA -0.188*** -10.01 TIREONE 0.199 1.61 LNMVE -0.038*** -3.95 TURN -0.064*** -6.62 RETVOL 4.266*** 11.87 PRCINV 0.278*** 6.68 MILLS -0.188** -2.16 Observations 1,672 Adj. R-squared 0.39 41