When LIBOR becomes LIEBOR: Reputational

When LIBOR becomes LIEBOR: Reputational penalties and bank contagion Michele Fabrizia , Xing Huanb , Antonio Parbonettic a

Department of Economics & Management, University of Padova, Via del Santo 33, Padova 35123, Italy. E-mail: [email protected] b Warwick Business School, University of Warwick, Coventry, CV4 7AL, UK. Corresponding author E-mail:[email protected] c Department of Economics & Management, University of Padova, Via del Santo 33, Padova 35123, Italy. E-mail: [email protected]

Abstract This paper examines whether commonality of incentives and opportunity to commit fraud triggers reputational contagion from culpable firms to nonculpable firm. Relying on a sample of thirty banks involved in fixing the LIBOR and a control sample of thirty banks from eleven countries, we find that banks’ reputations suffered substantial damage upon the announcement of their involvement in the scandal. We also document reputational contagion damage from banks that manipulated LIBOR to banks that shared the same incentives and opportunity to commit the fraud. Additional analyses show that the reputational contagion is more pronounced for large derivatives dealers, who have had the strongest incentive to commit the fraud. Keywords: reputational penalties, bank contagion, fraud triangle JEL Classification: G14, M41.

We are grateful to Anna Alexander, Abhinav Anand, Thomas Conlon, Elisabetta Ipino, Martin Jacob, Michel Magnan, Niall McGeever, William Megginson, Andrea Menini, Giovanna Michelon, Patricia O’Brien, Conall O’Sullivan, Stephen H. Penman, Gary J. Previts, and Giulia Redigolo for their helpful comments. This paper has also benefited from comments made by participants at seminars in Ca’ Foscari University of Venice, European School of Management and Technology, University College Dublin, University of Padova, and University of Zurich.

1. Introduction Reputation plays an important role in modern corporate life as a qualityassuring device (Klein & Leffler, 1981), and the growing integration of the international financial market has created multiple channels for the transmission of shocks. This paper addresses the reputational penalties that banks received for involvement in the London Interbank Offered Rate (LIBOR) scandal and the contagion effect on other banks in the LIBOR and connected panels (“LIBOR banks”1 hereafter). Extant literature on the spillover effect of reputational penalties has proposed three main mechanisms through which contagion might occur: through economic or exchange ties between firms (e.g., Beatty et al., 1998; Chaney & Philipich, 2002; Jensen, 2006); through inter-organizational networks, particularly director interlocks, where the directors of firms sit on each other’s boards (Kang, 2008); and through the mechanism of generalization by which reputational penalties spill over to other firms (Jonsson et al., 2009). This paper examines the mechanisms through which contagion can be channeled and reputational penalties can spill over from culpable firms to non-culpable firms. Specifically, we argue that reputational penalties spread from the firm accused of misconduct to other firms that share with it the same incentive and opportunity to commit the fraud. Rationalization, the incentive to commit a fraud and the opportunity to do so are the three elements of the so-called “fraud triangle” developed by Cressey (1973) and used by CPAs who seek to understand and manage fraud risk. Commonality of incentives and the opportunity to commit a fraudulent act represent a channel through which generalization and contagion can spread from one firm to other non-affiliated firms and is the channel investigated in this paper. The recent high-profile LIBOR rigging scandal that centered on UBS, Barclays, RBS and a number of other leading global banks has drawn considerable attention. While the damage from the scandal is still being calculated, the LIBOR scandal’s setting provides the opportunity to appraise the penalties imposed by the markets and the degree to which bank miscon1

We construct a panel that consists of thirty banks from three panels: 1) the LIBOR panel, which combines all of the submitter banks for all currencies; 2) Global Systemically Important Banks (GSIB); and 3) G14, the fourteen largest derivatives dealers. Banks like Barclays, Citigroup, and UBS are member banks of all three panels, while others sit in one or two of the three panels.

2

duct can affect other banks. Specifically, the contemporaneous presence of banks accused of LIBOR manipulation and non-accused banks that sit on the LIBOR panel provides us with a research setting in which to test whether commonality of incentives and opportunity to misbehave trigger reputational contagion. The effect of corporate misconduct is often examined in conjunction with stock market reaction (i.e., trading volume, stock price), earnings and risk, cost of capital, firm value, and corporate governance structure. The reputation literature focuses on reputational penalties that derive from financial penalties by regulators. We analyze the security returns of banks that are involved in LIBOR rigging during the period from March 2011 to December 2013 and identify thirty-nine event dates on which banks are accused of or sanctioned for manipulating LIBOR. Relying on a sample of thirty banks involved in fixing the LIBOR and a control sample of thirty banks from eleven countries, we find substantial reputational damage, measured by the banks’ stock returns in a three-day window around the event (i.e., [–1, +1]). We also document a contagion effect of the reputational damage that is passed from banks accused of LIBOR manipulation to non-accused banks that share the same incentives and opportunity to commit the fraudulent act. We also show that the contagion effect is significantly stronger among banks that are large derivatives dealers, who have had the strongest incentives to misbehave. Even though they were not accused of misconduct, they have experienced the strongest repulational penalties. We believe that one merit of this paper rests in the research design, which examines impacts on market value across all identifiable announcements (e.g., Karpoff et al., 2008) related to the LIBOR scandal, rather than only taking the first announcement date, as some event studies on this topic do (e.g., Armour et al., 2017; Nelson et al., 2008). Moreover, we provide insights into a new channel, represented by commonality of incentives and opportunity to commit the fraud, through which contagion can spread from one firm to other non-culpable firms. Our research findings are of significant importance, as the speed and scope of contagion in the banking sector, as demonstrated in many cases throughout history, are faster and larger than they are in other industries. The increased speed and scope of contagion tends to induce a larger scale and scope of failure, larger losses to creditors and depositors, and greater likelihood that the failure will extend beyond the banking sector to affect other industries, even the real economy (Kaufman, 1994). This paper makes four primary contributions to the literature. First, 3

it adds clarity to the literature of reputational penalty by examining reputational damage in banks in eleven developed markets, which is a sizeable scope, comparing to extant studies, most of which are based largely on US data (an exception is Jonsson et al. (2009) which is also in an international context). Second, this paper contributes to the literature on international bank contagion. Prior studies on the topic are done either in a single country or in a group of countries in the same region (e.g., Central and Eastern Europe). Two types of study represent the two extremes of choice of research setting in studying bank contagion: periods of bank failure (e.g., Aharony & Swary, 1983; Schoenmaker, 1996) and normal period (e.g., Brewer & Jackson, 2002; Docking et al., 1997; Furfine, 2003). A moderate setting between the two that presents a common shock to all banks posed by the same event has not yet been explored. Our choice of the LIBOR scandal as the setting in which to examine international bank contagion fills this gap. Third, this paper addresses a number of shortcomings and identification problems in the existing body of literature: Preliminary evidence derived from the existing cross-country studies demonstrates variation across markets, but reputational penalties are often assessed for different events at different times. A study based on a common event that occurs on an international scale has not yet been conducted. Moreover, studies frequently employ conventional methodology from accounting and finance research to address issues in the law and economics disciplines with a focus on coordination between legal sanctions and reputational penalties to reach an optimal social construct and legal system, but these studies do not study the micro level internal to these issues. Therefore, a more thorough enquiry of reputational penalties that uses the research tradition and toolkits from accounting and finance research is needed. Fourth, and most important, we extend the literature on the mechanisms through which contagion can spread from companies that commit a fraudulent act to non-culpable firms. Jonsson et al. (2009: 196) suggest that “two central questions for building theory on the consequences of organizational deviance are what kinds of actions are contagious and what kinds of actors are exposed to contagion.” Although Jonsson et al. (2009) establish that firms do not need to be linked formally to suffer the spillover effect of reputational penalties, Greve et al. (2010) claim that how far generalization will spread remains unclear. This paper contributes to this stream of research by providing evidence that addresses the second question by showing that the spillover effect of reputational penalties is not limited to organizations that 4

share exchange relationships, inter-organizational networks, or similar organizational form but is also channeled through commonality of incentives and opportunity to commit the fraudulent act. This paper of the reputational effect and contagion effect of the LIBOR scandal should also be of interest to practitioners and policy makers. The breadth of the contagion effect in the international banking system shown in this paper reinforces the importance of cross-country banking supervision and risk management. The remainder of this paper is structured as follows: Section 2 provides an overview of LIBOR and the LIBOR scandal. Section 3 reviews the related literature, focusing on ethics and discipline issues in banking and theories of reputation and bank contagion, on which the research hypotheses are developed. Section 4 describes the data and methodology, Section 5 presents the main results and discusses several robustness tests, and Section 6 concludes. 2. LIBOR and the LIBOR scandal LIBOR is a set of rates first published under the auspices of the British Bankers’ Association (BBA) in 1986 that represent the cost of short-term wholesale funds for major banks in the London interbank market. The original form of LIBOR is associated with Greek banker Minos Zombanakis2 , who in 1969 arranged one of the first syndicated loans, an $80 million loan from Manufacturer’s Hanover (now part of JP Morgan) to the Shah of Iran based on the reported funding costs of a set of reference banks3 . By the mid-1980s, the rate-submission banks had also started to borrow heavily using LIBORreferenced contracts, creating an incentive to underreport funding costs. In 1986, BBA took control of the LIBOR to formalize the data collection and administration process. LIBOR, then, is also known as BBA LIBOR. Prior to the transfer of responsibility for administration of LIBOR from BBA to the Intercontinental Exchange Benchmark Administration Ltd. (ICE) on 31 January 2014, LIBOR was calculated and published each business day by Thomson Reuters, to whom major banks submit their cost of borrowing un2 In a telephone interview after the LIBOR scandal, Minos Zombanakis claimed, “I was, more or less...the one who started the whole thing.” 3 A Greek banker spills on the early days of the LIBOR and his first deal with the Shah of Iran. Business Insider, 8 August 2012.

5

secured funds in ten currencies4 for fifteen maturities5 , so 150 LIBOR rates are reported daily. For each currency there is a panel consisting of contributing banks. The submission process requires contributing banks to evaluate the rates at which money may be available in the interbank market. Contributing banks are required to provide a submission based on that particular banks’ cost of borrowing unsecured cash for specific currencies and maturities, by responding the following question: “At what rate could you borrow funds, were you to do so by asking for and then accepting interbank offers in a reasonable market size just prior to 11 a.m.?”6 Thomson Reuters trims the rates by removing the top and bottom 25 percent and then averages the remainder to create the final LIBOR rates, which are published and available to the world around midday London time. This mechanism has a number of shortcomings since it is hypothetical, subjective, and open to abuse when an unethical person submits a rate (McConnell, 2013), so it is ineffective in preventing manipulation. At the heart of the LIBOR prosperity was the boom of the Eurodollar market. The importance of LIBOR grew as a response to the propagation of the Eurodollar7 market in the 1950s and 1960s (Kawaller, 1994; Schenk, 1998). Offshore Eurodollar deposits are a comparatively cheap source of funds for international banks, as these deposits are exempt from deposit insurance and regulatory reserve requirements. In its early days, this “backdoor market” grew secretly for a decade; it was not until 1959 that this strange new money market was inquired (Dormael, 1997). Starting in the 1970s, governments and corporations not only actively borrowed Eurodollar deposits from banks but also started to issue securities in Eurodollars at rates lower than the local borrowing rates. Schenk (1998) argues that the Eurodollar should not be viewed exclusively as a defensive innovation but also an aggressive one as banks took advantage of opportunities for profit through domestic currency swaps and third party lending, and also sought to meet 4

AUD, CAD, CHF, DKK, EUR, GBP, JPY, NZD, SEK, and USD. Overnight, one week, two weeks, one month, two months, three months, four months, five months, six months, seven months, eight months, nine months, ten months, eleven months, and twelve months. 6 The Wheatley Review of LIBOR: Final Report. 7 The Eurodollar is term–deposit denominated in US dollars held by banks outside the US (including foreign branches of US banks) and is free from regulation by the US Federal Reserve. 5

6

the need of customers. Although the fast-growing Eurodollar markets provide cheap funds to borrowers, the risk arose from the mismatch between the borrowers’ sources of income and the fact that funding remained severe, which triggered a new innovation, the IRS (Interest Rate Swap), to eliminate such risk. By the mid-1980s, the IRS market had reached a point at which some standardization was demanded to facilitate the continuing growth of the market. In 1986, as part of the “Big Bang” (the deregulation of the UK financial markets), BBA developed a mechanism to measure accurately the rates at which banks lend money to each other and contracts are settled, termed LIBOR. The importance of LIBOR continues to grow with London’s status as an international financial center from which more than 20 percent of all international banking lending and more than 30 percent of all foreign exchange transactions are executed8 . The prominence of LIBOR has grown drastically, and it is now deeply rooted in the international financial system, serving primarily as a reference rate and as a benchmark rate. As a reference rate, LIBOR is the basis for valuing financial instruments and establishing the terms of agreement for short-term floating rate contracts like swaps and futures. With notional outstanding values of at least $300 trillion9 , which is equivalent to approximately four and half times the global GDP, LIBOR is one of the single most important global reference rates referenced in transactions. Variable-rate loans from adjustable rate mortgages to credit cards to private student loans are also often tied to LIBOR. As a benchmark rate, LIBOR is used as the performance measure for funding costs and investment returns and is an indicator of the soundness of businesses and financial markets. On June 27, 2012, Barclays, the 300-year-old British bank, admitted to misconduct related to manipulating the daily setting of the LIBOR and the Euro Interbank Offered Rate (EURIBOR10 ) and agreed to a $453 million fine settlement with the UK Financial Services Authority (FSA), the US Com8

BBA LIBOR (2010). Understanding BBA LIBOR A briefing by the British Bankers’ Association. 9 The Wheatley Review of LIBOR: Final Report. 10 EURIBOR is a reference rate overseen by the European Banking Federation. The EUROBOR contributor panel consisted of approximately 42 to 48 banks. Thomson Reuters also acts as an agent for calculation and publishing of this rate. The administration process of this rate is highly similar to the one of LIBOR. However, only the highest and lowest 15 percent of all quotes are trimmed in EURIBOR calculation.

7

modity Futures Trading Commission (CFTC), and the US Department of Justice (DoJ). According to the statement supplied by the DoJ: “From approximately 2005 through 2007, and occasionally thereafter through approximately 2009, certain Barclays swap traders requested that certain Barclays LIBOR and EURIBOR submitters submit LIBOR and EURIBOR contributions that would benefit the traders’ trading positions, rather than rates that complied with the definitions of LIBOR and EURIBOR[...]. The swap traders made these requests via electronic messages, telephone conversations, and in-person conversations. The LIBOR and EURIBOR submitters agreed to accommodate, and accommodated, the swap traders’ requests for favorable LIBOR and EURIBOR submissions on numerous occasions.” Barclays admitted to three types of manipulation: under-reporting, over-reporting, and holding constant. The motives behind the three types of misrepresentations varied over time, from benefitting derivatives trading positions to avoiding the stigma of appearing weak relative to other banks during the financial crisis11 . Evidence from the regulators’ probe confirmed that the manipulation of LIBOR was not a localized event but part of business-as-usual in the global financial markets, a blatantly unethical and occasionally illegal practice that deliberately and systematically manipulated borrowing rates. One of the key issues reflected through this case is the inherent conflict of interest within the banks in question, as until September 1, 2009, UBS derivatives traders had determined the daily benchmark submission12 . The Swiss Financial Market Supervisory Authority’s inquiry into UBS also revealed the problem of inappropriate incentives, as, “on average, the variable part of the remuneration of submitters and traders accounted for 200 to 500 percent of their base salary. Inherent personal incentives therefore existed to benefit proprietary trading positions of the bank during the period under inquiry.”13 The consequence was that traders made requests to both the internal submitter at the bank and external submitters at third-party banks to influence the LIBOR submissions according to their own trading positions in derivatives and the money markets14 . 11

See Monticini and Thornton (2013) for the effect of misrepresentations of LIBOR rates. 12 UBS statement of facts – US Department of Justice. 18 December 2012. 13 FINMA summary report UBS LIBOR. 19 December 2012. 14 In the matter of: The Royal Bank of Scotland PLC and RBS Securities Japan Limited.

8

3. Literature review and hypothesis development 3.1. Bank ethics, market discipline, and regulatory discipline Bank ethics, market discipline, and regulatory discipline (i.e., financial market regulation and supervision) help to shape bank behavior, although the literature on bank behavior usually studies them separately. This section reviews studies in these three areas and draws a link between them as it relates to the issues raised in this paper. Like other areas of finance, banking is often seen as an amoral field that focuses purely on risk and return, yet ethics has played an important role in the evolution of banking (Cowton, 2002). The public perception of a bank’s ethics has a direct effect on its reputation. While good strategy and prudent management play important roles in business success, a good reputation is also a critical factor in ensuring long-term success (Green, 1989). Bank ethics and market discipline can be parallel and complementary in the domain of reputation, as they are both directed toward improving market participants’ perception of a bank’s reputation. However, the main distinction between bank ethics and market discipline rests on how that goal is reached: good ethics are mainly voluntary, while market discipline functions through the reaction (e.g., penalties) of stakeholders. A large body of literature examines whether the market can provide discipline against bank risk taking (Calomiris, 1999; DeYoung et al., 2001; Flannery, 1998; Jagtiani et al., 2002; Levonian, 2001; Morgan & Stiroh, 2001). Market discipline has less to do with the market per se than with the institutional framework, that is, the information, incentives, and control that are used to reduce problems related to the asymmetric information (e.g., moral hazard) that is endemic in banking (Stephanou, 2010). Market discipline is a market-based incentive mechanism that can curb the incentive to take excessive risk by making risk taking more costly. That is, investors in bank liabilities like subordinated debt and uninsured deposits punish banks for greater risk taking by demanding higher yields on those liabilities. Banks that are subject to stronger market discipline tend to limit their risk of default by holding more capital as a buffer against risk (Nier & Baumann, 2006). Bliss and Flannery (2002) interpret market discipline as consisting of two distinct components: market monitoring and market influences. Market monitoring refers to investors’ evaluating changes in a firm’s condition US Commodity Futures Trading Commission. 6 February 2013.

9

and promptly incorporating those assessments into the firm’s security prices, while market influences refer to the process through which external claimants affect a firm’s actions. Regulatory discipline and market discipline are viewed as complementary and self-reinforcing: appropriate regulations can enhance the power of market discipline, while market signals can provide information and incentives for banks and bank supervisors. Market discipline is often seen as a replacement for regulatory discipline, which was reflected in the wave of financial deregulations that has rapidly increased the number of unregulated players (e.g., hedge funds) and instruments (e.g., over-the-counter derivatives), and the dependence of prudential capital regulation on market-based measures of risk (Stephanou, 2010). Market discipline and regulatory discipline jointly determine a bank’s cost of misconduct: markets penalize banks for misconduct by demanding higher costs for financing debt and limiting the types of claims a bank can issue, while regulatory discipline acts through channels like risk-based capital requirements, insurance premiums, examination frequency and intensity, and cease-and-desist orders. Billett et al. (1998) show that bank shareholders perceive that the costs associated with regulatory discipline are less sensitive to risk increases than are the costs associated with market discipline. 3.2. Reputational penalties for corporate misconduct Various consequences of corporate misconduct are documented in the literature. Prior market-based research into corporate misconduct finds substantial losses in the market value of firms that are implicated in misconduct, which legitimates reputational effects as an important device in disciplining firms. The strength of reputational penalties has been examined in a number of different settings, and most of these studies document negative abnormal returns in the selected event period. Karpoff and Lott (1993) examine allegations of fraud by public corporations and suggest that a small portion of the loss in shareholder wealth is explained by criminal and civil sanctions imposed through the courts, while a more substantial portion of the loss is associated with the reputational penalties imposed by the market. They also find some evidence that firms’ earnings drop after a fraud-related announcement, while regulatory violations do not lead to reputational losses. Alexander (1999) finds similar reputational effects on shareholder wealth by investigating the reputational penalties suffered by public corporations that are accused of federal crimes, along with factors like management turnover, 10

shutdown of business units, and announcement of remedial strategies. Based on an investigation of enforcement actions for financial misrepresentation by the US Securities and Exchange Commission (SEC), Karpoff et al. (2008) find that two-thirds of the 41 percent drop in market value resulting from the announcement of misconduct are attributable to reputational losses, evidence that affirms the role of market-based reputational penalties. A considerable amount of the evidence from these studies is conditioned on the contractual relationship between the offending party and the offended party. For instance, Alexander (1999) suggests that losses of shareholder wealth are larger when the offended party is in a contractual relationship with the offending party, rather than a third party. Similarly, Karpoff et al. (1999) argue that the size of reputational penalties differs in defense procurement fraud based on the influence of the contractor, as unranked contractors are penalized heavily for procurement fraud, undergoing both a drop in market value and a subsequent loss in government contract revenue, while such changes are negligible for influential contractors. Although the literature affirms the role of reputation as a disciplinary mechanism for corporate misconduct, Murphy et al. (2009) argue that studies have not provided definitive evidence that links allegation-related wealth losses to changes in financial performance. Instead, they find decreased earnings and increased risk as the explanation for the allegation-related effects on wealth and reputation. Among the types of misconduct (i.e., antitrust violations, bribery, copyright violations, and fraud) that have been studied, fraud has the largest negative effect on firms’ market value. In studying the types of corporate misconduct, Karpoff et al. (2005) find that market reaction to violations of environmental regulations is not, on average, larger than the legal penalties imposed, so losses in share value are attributed to prospective legal sanctions rather than to reputational costs. Another stream of literature focuses on reputational penalties in relation to earnings restatements and announcements of operational loss. Perry and de Fontnouvelle (2005) assess corporations’ stock price reactions to announcements of major operational loss events and find a reputational loss when a firm’s market value declines by more than the announced amount of loss. They find negative reputational effects for losses caused by internal fraud, although external events have no reputational effect. Focusing on operational losses by banks and insurance companies Cummins et al. (2006) obtain results that are consistent with Perry and de Fontnouvelle (2005). Sturm (2013) analyzes reputational damage caused by operational losses in 11

European financial firms and contends that the negative stock market reaction is more pronounced in response to announcements of settlement than it is to the initial indications of the loss, and reputational damage is more pronounced for banks that are highly leveraged. Other studies extend the inquiry on this topic into a cross-country setting. Gillet et al. (2010) analyze market reactions to announcements of operational loss announcements by financial firms listed in Europe and the US and filter out the effects of reputational damage connected with internal fraud (i.e., where the loss in market value exceeds the operational loss). Comparing to the US firms, European companies usually demonstrate lower market value, indicating more reputational damage. Such findings provide a setting in which to assess regulatory enforcement and changes in reputation and trust in several geographic areas or societies. Tanimura and Okamoto (2013) note that reputational penalties are larger in Japan than in the US, endorsing the belief that reputation and trust are particularly important in Japanese society. Fiordelisi et al. (2014) suggest that reputational losses in the financial industry are higher in Europe than in North America, while Armour et al. (2017) analyze the reputational losses sustained by financial firms that are penalized by the regulatory body in the UK and find that reputational penalties average nine times the size of the financial penalties. They also note that the size of the penalties leveled in the UK does not necessarily reflect the seriousness of the wrongdoing as perceived by investors and clients; instead, the disclosure of misconduct itself is the primary source of the reputational damage. Our first research question is whether the market penalizes banks for accusations of LIBOR manipulation. Reputational penalties in financial markets are an ethical mechanism that internalizes the cost of corporate misconduct in terms of repeated business transactions (Engelen & van Essen, 2011). We suggest that the stock price—the present value of future cash flows—around the announcement date is a final price generated through the mechanism of multiple underlying channels through which stakeholders—including customers and clients, employees, suppliers, debt holders, and equity holders— penalize the bank for alleged LIBOR manipulation. As discussed above, there is rich evidence that reputational losses occur when corporate misconduct is announced, so we expect that, following an announcement of involvement in the LIBOR scandal, banks experienced reputational penalties imposed by the market, measured by a drop in share value. Therefore, we posit our first hypothesis as follows: 12

Hypothesis 1 (H1): The market penalizes banks in the form of reputational damage upon the announcement of alleged LIBOR manipulation. 3.3. Financial contagion The presence of financial contagion is complex, steeped in the increasingly integrated networks of international financial markets. Results from studies on the presence and magnitude of financial contagion is controversial and diverse in terms of definitions, propagation mechanisms and channels, research settings, and methods. Most tests for financial market contagion are done in the setting of financial crises on the national level (e.g., Bekaert et al., 2014; Kenourgios et al., 2011; Longstaff, 2010), with considerable attention paid to emerging markets (e.g., Baig & Goldfajn, 1999; Chiang et al., 2007; Dimitriou et al., 2013; Iwatsubo & Inagaki, 2007; Khalid & Kawai, 2003). Another avenue of empirical study focuses solely on banking industry, examining whether bad news from one bank or group of banks, such as news about bank failures and financial distress, adversely affects other banks. Aharony and Swary (1983) detect a contagion effect in an analysis of the three largest bank failures in the US but contend that the observed drop in the valuation of solvent banks arises from investors’ response to a negative signal instead of a contagion effect. Schoenmaker (1996) employs a larger sample of bank failures under the US National Banking System from 1880 to 1936 and confirms the presence of contagion risk in banking. Based on a simulation of the impact of the failure of one bank on other banks, Furfine (2003) studies interbank payment flows of 719 US commercial banks that used Fedwire in 1998 and proffers that contagion that arises from direct links between banks does not necessarily present a system-wide threat to the US banking system. Following the typology that classifies US banks into money-center banks and regional banks, Docking et al. (1997) analyze loan-loss announcements over the 1985–1990 period and document a significant negative contagion effect in both types of banks. Slovin et al. (1999) find that dividend reductions by money-center banks have a negative contagion effect on rival banks, but not when the dividend reductions are made by regional banks. Similarly, Bessler and Nohel (2000) study the contagion effect of dividend reductions by US banks and document a contagion effect in both non-announcing money-center banks and big regional banks. Brewer and Jackson (2002) test for a contagion effect following announcements of financial distress by US commercial banks and life insurance companies and argue that the documented interindustry effects on shareholder wealth are not purely contagious in nature 13

but are also attributable to factors like the composition of portfolios, geographic proximity, leverage, and regulatory expectations. The post-2000 period saw many studies on inter-country contagion. This trend suggests that the characteristics of international financial markets, particularly the increasing interconnectedness, are changing. However, evidence from earlier studies that indicate the absence of contagion may still be valid since the architecture of financial markets in the 1980s and 1990s was not as conducive to facilitating financial contagion as it was a decade or more later. Overall, results from these studies suggest escalating bank contagion over time. This discussion suggests that there is ample evidence of financial contagion among both financial and non-financial firms. When the determinants and consequences of contagion are analyzed, the mechanisms underlying the observed spread of reputational penalties to non-affiliated firms must be investigated. The extant literature on the spillover effect of reputational penalties has proposed three main mechanisms through which contagion might occur: 1) economic or exchange ties between firms, 2) inter-organizational networks, and 3) generalization. The first mechanism is the most frequently investigated in the literature. Several studies show that reputational penalties spread from a misbehaving company to other companies that share with it economic or exchange ties. Beatty et al. (1998) show that SEC investigations of underwriters impose a variety of measurable and significant indirect penalties on their clients, particularly IPO clients. Since underwriters’ role is to provide assurances on the degree of information asymmetry between the client and the market, when an underwriter’s reputation for truth-telling is damaged, the market penalizes the underwriter’s current and former clients because of the unexpected increase in the assessment of the information asymmetry surrounding the underwriter’s clients. Consequently, the reputation of an underwriter and its clients are positively related, and the deterioration of the reputation of the former inevitably affects the later (Beatty et al., 1998). Chaney and Philipich (2002) focus on the Enron audit failure and document that the market imposed reputation penalties to the other clients of the Enron’s auditor. Specifically, results reported in Chaney and Philipich (2002) show that, in the three days following Andersen’s admission that a significant number of documents had been shredded, Andersen’s other clients experienced a negative market reaction. Moreover, clients with the audits performed by Andersen’s Houston office, which served Enron directly, suffered a more severe penalty. Jensen (2006) also focuses on the collapse of Arthur Andersen and investi14

gates what affected Andersen clients’ defection. Specifically, Jensen (2006) shows that accountability-induced status anxiety influenced defections by Andersen’s publicly traded clients. Therefore, similar to Beatty et al. (1998) and Chaney and Philipich (2002), Jensen (2006) shows that a fraudulent act can affect non-affiliated companies that share economic and exchange ties with the fraudulent firm. In a similar vein, Akhigbe et al. (2005) investigate the degree to which misleading accounting can affect connected firms that have business relationships with the accused firm. Using Enron as the research setting, Akhigbe et al. (2005) show that adverse news releases about one firm have contagion effects for connected firms, such as suppliers, counterparties to trading, and creditors. In a related study, Gleason et al. (2008) investigate the contagion effects of accounting restatements and find that investors impose penalties on the stock price of the restating company’s peers. The contagion effect also spreads to companies that use the same external auditor as the restating firm. Finally, Bonini and Boraschi (2010) focus on security class-action suits and show that a corporate scandal affects the security issuances and stock price of industry peers. These studies document a negative spillover effect that moves from one company to other organizations that share with it some form of economic or exchange ties, such as clients, suppliers, and industry peers. Therefore, sharing an economic and/or exchange relationship with a firm is one mechanism through which contagion might occur. Kang (2008) examines a second mechanism that can channel contagion: the presence of inter-organizational networks. Specifically, Kang (2008) focuses on director interlocks, which are one of the most widely studied forms of inter-organizational networks. Two companies are connected by a director interlock when a person affiliated with one firm sits on the other firm’s board of directors (Mizruchi, 1996; Kang, 2008). Focusing on financial reporting fraud, Kang (2008) studies whether director interlocks are channels through which the reputational penalties experienced by one firm spill over to other firms. Kang (2008) tests the spillover of reputational penalties using an event study on a sample of 244 associated firms with director interlocks to thirty firms accused of financial reporting fraud in the US. Results are consistent with a contagion effect from accused to associated firms through director interlocks. The novelty in Kang’s (2008) study is the focus on a contagion mechanism that, in contrast to prior literature, focuses on interorganizational networks rather than economic or exchange ties. Finally, Jonsson et al. (2009) introduce a third mechanism for the spillover 15

effect of reputational penalties, generalization, which channels contagion from an accused firm to non-affiliated organizations. Jonsson et al. (2009: 195) claim that “individuals observe a single act of deviance, interpret it as potentially harmful or contrary to social norms, and proceed to incorporate the possibility of such events into their general knowledge about organizations.” Focusing on two scandals in the Swedish insurance industry, the authors provide robust evidence that there are three ways through which generalization can be channeled and reputational penalties can be imposed on non-affiliated firms: through a shared form of organization, across forms of organizations with similar characteristics, and across organizations with characteristics made salient by the fraudulent act. In this context, organizational forms are defined as organizations that share industry and operational blueprints (Jonsson et al., 2009). Our second research hypothesis aims to clarify the mechanisms through which contagion can be channeled. In doing so, we focus on firms that have the same incentives and opportunity to commit a fraudulent act as a misbehaving organization has. Specifically, we argue that contagion – through generalization – can spread from the accused firm to organizations that share with it the same opportunity and the same incentive to misbehave. Rationalization, the incentive to commit fraud, and the opportunity to do it are the elements of the so-called “fraud triangle” developed by Cressey (1973). This three-pronged framework suggests that fraud is more likely to occur when there is an incentive to commit the fraud, when weak controls or oversight provide an individual or organization with the opportunity to commit the fraud, and when the perpetrator can rationalize the fraudulent behavior. In the early 1980s, the fraud triangle concept was adapted from criminology for use in accounting (Choo & Tan, 2007), and it has ever since been a useful tool for CPAs who seek to understand and manage fraud risk. The “fraud triangle” framework has also been formally adopted by the auditing profession as part of the Statement of Auditing Standards no.99, which addresses the consideration of fraud in a financial statement audit (Wolfe & Hermanson, 2004). The LIBOR scandal offers a desirable research setting in which to determine whether generalization occurs through commonality of incentives and opportunity. Specifically, we argue that, when one LIBOR bank is accused of manipulation, the reputational penalties spill over to other, non-accused LIBOR banks since they share with the accused bank the incentive and the opportunity to commit the fraud. The commonality of incentives and the 16

opportunity to misbehave is likely to trigger a process of generalization that leads investors to transfer reputational penalties from firms that actually misbehaved to other, innocent (at least in this regard) firms that could have misbehaved, given their set of incentives and their opportunity. Given the above discussion, we posit that: Hypothesis 2 (H2): The reputational damage suffered by banks accused of LIBOR manipulation has a contagion effect on other banks that share the same incentive and opportunity to commit the fraud. 4. Data and methodology 4.1. Data We construct a database for a sample of banks that were accused of manipulating the LIBOR during the period between March 2010 and December 2013. We hand-collected news on the LIBOR scandal from four major news providers: BBC News, Bloomberg, The Financial Times, and The Wall Street Journal. We chose this set of news providers for the following reasons: First, the LIBOR scandal was discovered and published in the UK, and BBC News provided full coverage of the related news in a section dedicated to the updates on this scandal. Second, the other three news providers, Bloomberg, The Financial Times, and The Wall Street Journal, are highly specialized in global coverage of financial news. In addition, The Wall Street Journal was among the first to raise questions about whether global banks were manipulating the process by “low-balling” a key interest rate to avoid looking desperate for cash in the midst of the financial crisis.15 We identify thirty-nine events dates when information regarding the LIBOR investigation was disclosed. For stale news, such as news about market anticipation and news provided by insiders who were familiar with the situation prior to the fine settlement date, we take the earliest available date. Table 1 describes each event dates used in our analysis. [Please insert table 1 about here.] We retrieve market data around the event dates for all banks that belonged to any of three panels: 1) the LIBOR panel, which contains all con15

Study casts doubt on key rate. The Wall Street Journal, 29 May 2008.

17

tributor banks for all currencies; 2) Global Systemically Important Banks (GSIB)16 ; and 3) G14, the fourteen largest derivatives dealers that collectively account for more than 80 per cent of all outstanding interest rate derivatives. As shown in Table 2, most of the contributor banks are GSIB, and those non-GSIB banks are likely to be designated as Local Systemically Important Banks (LSIB) by their local prudential regulators. That is, LIBOR contributor banks are large and systemically important. On the other hand, all of the fourteen largest derivatives dealer banks are LIBOR contributor banks, indicating an incentive behind the rate manipulation. Some banks, such as Barclays, Citigroup, and UBS, are members of all the three panels, while some banks sit on one or two of the three panels. In fact, the combination of information in Table 1 and Table 2 shows a tendency toward a positive correlation between the number of panels of which a bank is a member and the level of its involvement in the LIBOR scandal. We also pooled the three panels into one set, named “LIBOR banks,” as presented in Table 2. Some of the LIBOR banks were involved in the LIBOR scandal, while others have not been accused of misconduct. Nonetheless, all banks that sit on any of the three panels had the incentive and the opportunity to be involved in the LIBOR manipulation, given that they were highly interconnected: more than half of IRS trades are between financial institutions rather than between a bank and a customer, and even this trading is concentrated in a small number of large banks (Fleming et al., 2012). We use this sample of LIBOR banks to test whether i) the market punished LIBOR banks involved in the LIBOR scandal and ii) non-accused LI16

In November 2011, the Financial Stability Board (FSB) published an integrated set of policy measures to address the systemic and moral-hazard risks connected with systemically important financial institutions (SIFIs). In that publication, the FSB identified as global SIFIs (G-SIFIs) an initial group of systematically important banks (GSIBs) using a methodology developed by the Basel Committee on Banking Supervision (BCBS). The group of G-SIFIs is updated annually based on new data and published by the FSB each November. Since the November 2012 update, the GSIBs have been allocated to buckets corresponding to the higher loss-absorbency requirements that they would be required to hold. GSIBs are subject to requirements for group-wide resolution planning and regular resolvability assessments. In addition, the resolvability of each GSIB is reviewed in a high-level FSB Resolvability Assessment Process by senior policy-makers in the firm’s Crisis Management Groups. GSIBs are also subject to higher supervisory expectations for risk-management functions, risk data aggregation capabilities, risk governance, and internal controls.

18

BOR banks received reputational penalties because they shared with the culpable banks the same incentives and opportunity to misbehave. We refer to the former effect as a reputational effect and to the latter effect as a contagion effect. Following Kaufman (1994), Docking et al. (1997), and Gropp and Moerman (2004), we use a working definition of contagion as negative stock returns experienced by banks that arise from the disclosure of misconduct by another bank or group of banks. Our definition of financial contagion is close to the spirit of Lang and Stulz (1992) and Kaufman (1994), who regard negative information externalities (spillovers) as contagion. Our identification strategy relies on comparing market returns around the event dates for LIBOR banks (both accused banks and non-accused banks) and a control group of banks that were external to any LIBOR banks. Therefore, banks in the control group were neither involved in the scandal nor had the incentives and opportunity to misbehave since they were not included in any LIBOR panels. Our two research hypotheses predict a negative and significant market reaction around event dates for accused and non-accused LIBOR banks and a non-significant market reaction around event dates for non-LIBOR banks. [Please insert table 2 about here.] We obtain daily stock price data from March 2010 to December 2013 from Compustat Security Daily. In order to capture an accurate effect of the event on the banking industry, we construct a market portfolio of the financial services sector (SIC code ranging from 6000 to 6500) in each country to calculate the capitalization-weighted market return, rather than using stock indices as a proxy for market return. In addition, we obtain each country’s 10-year bond yield rates from Thomson Reuters to proxy for the treasury interest rate. 4.2. Methodology The measurement of reputational penalties in the literature is largely based on the assumption that stock price represents investors’ expectations concerning the value of the company based on the mechanism of supply and demand, which consists of multiple channels through which stakeholders impose reputational penalties on the company. In this mechanism, reputational penalties imposed by stakeholders–such as the loss of customers, business 19

partners, employers, and suppliers that results in increased costs, and the potential rise in the cost of capital demanded by debt holders and equity holders to reflect the company’s increased risk–are incorporated in a drop in stock price upon the announcement of corporate misconduct. However, reputational penalties are measured in varying ways, among which is to take the residual in market value loss after deducting the amount of either legal sanctions or the announced loss. Some studies, such as Murphy et al. (2009), consider the sum of the loss in market value, drop in earnings, and the rise to reflect the reputational penalty. We adopt the multifactor model from Beatty et al. (1996), in which bank-specific returns are regressed on the market return, on a variable that captures interest rate changes, and on event-related dummy variables. The event-related dummy variables provide for mean shifts in returns on event days. We assume that the following models hold for each bank in our sample. For ease of presentation, firm and time subscripts are omitted.

R = α1 + β1 Rm + γ1 ∆T reasury + δ1 Event + θ1 ADJF IN E + ε

(1)

R = α2 + β2 Rm + γ2 ∆T reasury + δ2 N onevent + ε

(2)

R = α3 + β3 Rm + γ3 ∆T reasury + δ3 N onLIBOR + ε

(3)

where R is the daily stock return for an individual bank in the sample, Rm is the capitalization-weighted daily market return, ∆T reasury is the daily change in the 10-year government bond yield, and ADJF IN E is the fine settlement scaled by the pre-settlement market capitalization. Model 1 uses a three-day event window to test for the reputational effect (H1) for each of the events listed in Table 1. That is, upon the date of the announcement of a bank’s alleged manipulation, Event takes a value of 1 on the day before, the day of, and the day after the event date. If an event takes place on a weekend, the event date is adjusted to the next trading day. If two event windows overlap, we merge the two event windows by extending from the day prior to the first event to the day following the second event. The estimation of Model 1 is based on 21,151 daily observations, constituted of only Event banks (i.e. accused LIBOR banks). H1 predicts a significant and negative sign on δ1 . Model 2 tests for the contagion effect (H2). N onevent takes a value of 1 when a bank was a LIBOR bank and was not accused of manipulation 20

when one or more LIBOR banks were accused on the event date. In other words, these nonevent LIBOR banks were not yet officially involved in the scandal (i.e., public/press announcements) when other LIBOR banks were accused to have been involved. The estimation of Model 2 is based on 14,624 daily observations17 . The coefficient of interest, δ2 , represents the contagion effect. H2 predicts that δ2 is significantly negative. In order to rule out the possibility that the negative stock returns in Models 1 and 2 are industrywide instead of specific to LIBOR banks, we estimate Model 3, in which the dummy variable N onLIBOR takes a value of 1 when a bank is not a LIBOR bank and it is an event date for one or more LIBOR banks. The estimation of Model 3, which is based on 862,840 daily observations, constituted of all banks that are external to LIBOR banks. The banks we use to estimate Model 3 were neither involved in the scandal nor had the incentive or opportunity to misbehave since they were not on the LIBOR panel. Consequently, if the market imposed a reputational penalty on the banks that were involved in the scandal, and the reputational penalties spilled over to non-accused banks that shared the incentive and opportunity to commit the fraud, we expect δ3 to be not statistically significant at any conventional level. Results from Model 3 help to mitigate the concern that the market reaction around event dates is driven by market-wide factors rather than by the unfolding of the LIBOR scandal. Moreover, a non-significant coefficient on N onLIBOR in Model 3 supports that the contagion effect is due to the commonality of incentives and opportunity to commit the fraudulent act, rather than being common to all banks. Finally, to corroborate the support for our research hypothesis that deals with the contagion effect, we estimate Model 2 by adding an interaction term in order to determine the role of incentive as a mechanism that channels contagion. Specifically, if commonality of the incentive to commit fraud channels generalization and contagion, we should observe a stronger contagion effect among the LIBOR banks. Therefore, we test whether contagion is stronger among the most active derivative dealers that have the most to gain from a potential manipulation of the LIBOR. The regression takes the following form: 17

The drop in the number of observations from Model 1 to Model 2 is due to banks are considered N onevent banks only until they are officially announced as part of the LIBOR scandal.

21

R = α4 + β4 Rm + γ4 ∆T reasury + δ4 N onevent + θ4 N onevent × T opDealers + µ4 T opDealers + ε (4) where T opDealers is a dummy that equals one for the fourteen largest derivatives dealers, and zero otherwise. This group of banks, commonly referred to as the G14, could arguably gain the most from manipulating the LIBOR. This classification is widely used in the business press and by organizations like the Bank for International Settlements (BIS). The ISDA Market Survey for Mid-Year 2010 pointed out that G14 banks held 82 percent of the total notional amount of derivatives that were outstanding by mid-2010. In Model 4, the coefficient of interest, θ4 , investigates whether the contagion effect is stronger among large derivative dealers that have stronger incentive to misbehave. We expect a negative and significant coefficient on θ4 . This additional analysis is particularly well suited to our research setting because the regulators’ probe highlights the key role of certain Barclays swap traders in the LIBOR scandal, who requested that certain Barclays submitters submit LIBOR and EURIBOR rates that would benefit the traders’ trading positions, rather than rates that complied with the definitions of LIBOR and EURIBOR. 4.3. Descriptive statistics We describe our sample in three ways. First, we report the sample composition and coverage by country (Table 3). Our sample consists of banks from eleven countries. The sample that we use to calculate the daily market return comprises 2,860 banks. The final sample consists of 1,262 unique banks: 30 are LIBOR banks and 1,232 banks that are non-LIBOR banks with available data from Compustat Security Daily during the sample period. Second, Table 4 reports descriptive statistics for 21,151 daily observations, and Table 5 reports the strength and direction of correlations between variables. [Please insert tables 3, 4 and 5 about here.]

5. Results 5.1. Initial emergence of the LIBOR scandal Table 6 reports the significant events surrounding the initial emergence of the LIBOR scandal. On Thursday, March 24, 2011, Barclays emerged as 22

a key focus of the investigation by US and UK regulators. The following day (Day +1), Barclays’ shareholders experienced a raw return of –0.31 percent, while the financial services industry’s capitalization-weighted daily return in the UK was 0.96 percent. The capital market’s reaction to the initial emergence of the LIBOR probe was weak, particularly compared to the reaction to the first LIBOR settlement on Wednesday, June 27, 2012, when Barclays admitted to misconduct and reached fine settlements with both UK and US regulators that amounted to $450 million. Barclays’ shareholders experienced a raw return of –15.5 percent, equivalent to a £120.27 million loss in firm value, on the following day (Day +1), while the financial services industry’s capitalization-weighted daily return in the UK was –0.64 percent. The difference between the two event dates suggests that the market did not penalize the bank until the misconduct was affirmed with the financial penalty imposed. An alternative explanation is that the financial penalty issued by the regulators was higher than the market expected. The average raw return of the remaining non-accused LIBOR banks on June 28, 2012 (Day +1), was –0.42 percent, providing preliminary evidence for the contagion effect of the penalty on Barclays’ misconduct. [Please insert table 6 about here.]

5.2. Reputational penalties and the contagion effect Table 7 shows the results from the tests for H1 and H2. When interpreting the results in the table, note that the single-event results reflect an average return during the event window. The overall sample’s abnormal return can be obtained by multiplying each of these reported figures (i.e., the coefficients of Event, N onevent, and N onLIBOR) by the number of days in the event window (three days in this paper). Column 1 shows that banks that either were accused of or admitted to misconduct experienced an average –0.6 percent cumulative abnormal return (CAR) during the three-day event window. The coefficient of Event in Column 1 is negative and statistically significant, indicating that banks accused of misconduct experienced negative abnormal returns during the periods surrounding the announcement of alleged manipulation, which supports H1. Column 2 reports the market reaction of non-accused LIBOR banks during the period when one or more LIBOR banks were accused of misconduct. Nonevent banks experienced an average –0.3 percent CAR during the three-day period. The coefficient of 23

N onevent is significantly negative, suggesting that the LIBOR banks that were not accused of misconduct experienced negative abnormal returns during the period surrounding the announcement of manipulation, which we interpret as the contagion effect of the reputational damage channeled by the commonality of incentives and opportunity to commit the fraud. The coefficient of N onLIBOR in Column 3 is not significant, so the reputational penalties and the contagion effect of the reputational penalties are specific to LIBOR banks, either accused of manipulation or characterized by a communality of incentives and opportunity to commit the fraudulent act. The insignificant coefficient of δ3 in Table 7 mitigates the possibility that our results are due to industry-wide shocks that occurred around the event dates. Moreover, it shows that the spillover of reputational penalties affected only banks with common incentives and opportunity to commit fraud, and not the entire sample of banks. Overall, our results in Table 7 support both H1 and H2. [Please insert table 7 about here.] Our theoretical framework posits that contagion is channeled by the commonality of incentives and opportunity to commit a fraudulent act. Consequently, we should expect that contagion is stronger among banks that have stronger incentives and opportunities to misbehave. Table 8 presents results from estimating Model 4, in which we interact the dummy variables N onevent with T opDealers to determine whether the contagion effect is stronger among large derivatives dealers that have inherently greater incentive to commit the fraud. As expected, the coefficient on the interaction term is negative and statistically significant, providing support for the notion that the commonality of incentives to misbehave channels the contagion effect. [Please insert table 8 about here.]

5.3. Robustness In this section, we perform additional tests to examine the robustness of the results presented in the previous sections. A first concern relates to the possibility that reputational penalties spread to non-accused LIBOR banks, not because of the commonality of incentive and opportunity to commit the fraudulent act, but because they were the most similar banks in terms of 24

size. This alternative explanation is consistent with results in Jonsson et al. (2009), but they contrast with our H2. In order to address this issue, we create two benchmark groups, each matched on the banks’ sizes in terms of total assets: a sample of thirty banks created by one-to-one matching on size with LIBOR banks in each country and a sample of thirty banks generated by one-to-one matching on size with LIBOR banks in the eleven countries. Results from using these size-matched samples to estimate Model 3 are reported in Table 9, Panel A. Regardless of which benchmark group we use, the results are consistent with our previous findings: the coefficient of is insignificant across all the regressions performed, reinforcing the notion that the reputational penalties and their contagion effect are restricted to LIBOR banks and do not extend to other banks that are simply similar in size. An additional concern that might affect our analysis relates to the inclusion of multiple events for each bank. This feature of our research design allows us to increase the number of events analyzed in order to improve the power of our tests. Even so, this design choice might introduce a bias in the analysis. The first step that we take to address this concern is to exclude all overlapping events, retaining only the first event. Results reported in Panel B of Table 9 are still consistent with our main inferences, confirming the support of H1 and H2. Second, using an alternative approach to address the issue, we maintain all the event dates in the sample but cluster the standard errors at both the firm and the event level, so all observations generated from subsequent events are clustered together. By doing so, we correct for the interdependence of observations of subsequent events. Results from this alternative specification, reported in Panel C of Table 9, provide support for H1 and H2. [Please insert table 9 about here.]

6. Conclusion This paper provides evidence that generalization occurs through shared incentives and opportunities to commit fraud. Specifically, we show that negative reputational penalties are imposed by the market to non-culpable firms that, simply by sitting on the LIBOR panel, share with fraudulent companies the opportunity and the incentive to misbehave. Therefore, this paper provides evidence that contagion can spread from the misbehaving 25

firms to innocent, non-affiliated firms, even if the innocent company is not in an exchange relationship with the misbehaving firm and does not share with it an inter-organizational network. The results on large derivative dealers demonstrate the ability of shared incentives and opportunities to channel contagion to non-affiliated firms. Therefore, it has been sufficient for the market to recognize a commonality of incentives and opportunity to commit the fraud to transfer reputational penalties from accused firms to non-accused firms. This paper documents reputational penalties and international bank contagion in the setting of a financial scandal, investigating how organizational actions that are contrary to social norms lead to reputational damage for organizations that share incentives and opportunity to engage in wrongdoing with the offending organization, even if they do not so engage. This paper is relevant to policy makers who seek to influence banks’ risk-taking behavior. Solutions to unethical behavior in banking and banking regulation in general require deep and far-ranging actions. Some countries have included ethical guidelines in legislation, while others have not. As bank ethics are still often voluntary, the task of disciplining banks falls to market discipline and legal enforcement. Sound and ethical banking demands the joint effort of banks in establishing ethical behavior as a foundation for the industry, the market and regulatory disciplines to guide and supervise bank behavior, and an effective legal environment that provides contextual support for the function of these two devices. The important role of bank ethics and regulatory discipline is reinforced in the absence of an implicit government guarantee, which absence dulls market discipline by reducing investors’ incentives to monitor and price the risk taking of large banks.

26

7. References Aharony, J., & Swary, I. (1983). Contagion effects of bank failures: Evidence from capital markets. Journal of Business, 56 , 305–322. Akhigbe, A., Madura, J., & Martin, A. D. (2005). Accounting contagion: The case of Enron. Journal of Economics & Finance, 29 , 187–202. Alexander, C. R. (1999). On the nature of the reputational penalty for corporate crime: Evidence. Journal of Law & Economics, 42 , 489–526. Armour, J., Mayer, C., & Polo, A. (2017). Regulatory sanctions and reputational damage in financial markets. Journal of Financial & Quantitative Analysis, 52 , 1429–1448. Baig, T., & Goldfajn, I. (1999). Financial market contagion in the Asian crisis. IMF Staff Papers, 46 , 167–195. Beatty, A., Chamberlain, S., & Magliolo, J. (1996). An empirical analysis of the economic implications of fair value accounting for investment securities. Journal of Accounting & Economics, 22 , 43–77. Beatty, R. P., Bunsis, H., & Hand, J. R. M. (1998). The indirect economic penalties in SEC investigations of underwriters. Journal of Financial Economics, 50 , 151–186. Bekaert, G., Ehrmann, M., Fratzscher, M., & Mehl, A. (2014). The global crisis and equity market contagion. Journal of Finance, 69 , 2597–2649. Bessler, W., & Nohel, T. (2000). Asymmetric information, dividend reductions, and contagion effects in bank stock returns. Journal of Banking & Finance, 24 , 1831–1848. Billett, M. T., Garfinkel, J. A., & O’Neal, E. S. (1998). The cost of market versus regulatory discipline in banking. Journal of Financial Economics, 48 , 333–358. Bliss, R. R., & Flannery, M. J. (2002). Market discipline in the governance of U.S. bank holding companies: Monitoring vs. influencing. Review of Finance, 6 , 361–396. Bonini, S., & Boraschi, D. (2010). Corporate scandals and capital structure. Journal of Business Ethics, 95 , 241–269. Brewer, E., & Jackson, W. E. (2002). Inter-industry contagion and the competitive effects of financial distress announcements: Evidence from commercial banks and life insurance companies. Federal Reserve Bank of Chicago Working Paper. Calomiris, C. W. (1999). Building an incentive-compatible safety net. Journal of Banking & Finance, 23 , 1499–1519. Chaney, P., & Philipich, K. (2002). Shredded reputation: The cost of audit failure. Journal of Accounting Research, 40 , 1221–1245.

27

Chiang, T. C., Jeon, B. N., & Li, H. (2007). Dynamic correlation analysis of financial contagion: Evidence from Asian markets. Journal of International Money & Finance, 26 , 1206–1228. Choo, F., & Tan, K. (2007). An “American dream” theory of corporate executive fraud. Accounting Forum, 31 , 203–215. Cowton, C. J. (2002). Integrity, responsibility and affinity: Three aspects of ethics in banking. Business Ethics: A European Review , 11 , 393–400. Cressey, D. R. (1973). Other People’s Money: Study in the Social Psychology of Embezzlement. Montclair, NJ: Patterson Smith. Cummins, J. D., Lewis, C. M., & Wei, R. (2006). The market value impact of operational loss events for US banks and insurers. Journal of Banking & Finance, 30 , 2605–2634. DeYoung, R., Flannery, M. J., Lang, W. W., & Sorescu, S. M. (2001). The information content of bank exam ratings and subordinated debt prices. Journal of Money, Credit & Banking, 33 , 900–925. Dimitriou, D., Kenourgios, D., & Simosa, T. (2013). Global financial crisis and emerging stock market contagion: A multivariate FIAPARCH–DCC approach. International Review of Financial Analysis, 30 , 46–56. Docking, D. S., Hirschey, M., & Jones, E. (1997). Information and contagion effects of bank loan-loss reserve announcements. Journal of Financial Economics, 43 , 219–239. Dormael, A. V. (1997). The Power of Money. London, UK: Palgrave Macmillan. Engelen, P.-J., & van Essen, M. (2011). Responsible investment in times of turmoil. chapter Reputational Penalties in Financial Markets: An Ethical Mechanism? Netherlands: Springer. Fiordelisi, F., Soana, M.-G., & Schwizer, P. (2014). Reputational losses and operational risk in banking. European Journal of Finance, 20 , 105–124. Flannery, M. J. (1998). Using market information in prudential bank supervision: A review of the U.S. empirical evidence. Journal of Money, Credit & Banking, 30 , 273–305. Fleming, M., Jackson, J., Li, A., Sarkar, A., & Zobel, P. (2012). An analysis of OTC interest rate derivatives transactions: Implications for public reporting. Federal Reserve Bank of New York Staff Reports. Furfine, C. H. (2003). Interbank exposures: Quantifying the risk of contagion. Journal of Money, Credit & Banking, 35 , 111–128. Gillet, R., Hbner, G., & Plunus, S. (2010). Operational risk and reputation in the financial industry. Journal of Banking & Finance, 34 , 224–235.

28

Gleason, C. A., Jenkins, N. T., & Johnson, W. B. (2008). The contagion effects of accounting restatements. The Accounting Review , 83 , 83–110. Green, C. F. (1989). Business ethics in banking. Journal of Business Ethics, 8 , 631–634. Greve, H. R., Palmer, D., & Pozner, J.-E. (2010). Organizations gone wild: The causes, processes, and consequences of organizational misconduct. Academy of Management Annals, 4 , 53–107. Groppa, R., & Moerman, G. (2004). Measurement of contagion in banks equity prices. Journal of International Money & Finance, 23 , 405–459. Iwatsubo, K., & Inagaki, K. (2007). Measuring financial market contagion using duallytraded stocks of Asian firms. Journal of Asian Economics, 18 , 217–236. Jagtiani, J., Kaufman, G., & Lemieux, C. (2002). The effect of credit risk on bank and bank holding company bond yields: Evidence from the post–FDICIA period. Journal of Financial Research, 25 , 559–575. Jensen, M. (2006). Should we stay or should we go? Accountability, status anxiety, and client defections. Administrative Science Quarterly, 51 , 97–128. Jonsson, S., Greve, H. R., & Fujiwara-Greve, T. (2009). Undeserved loss: The spread of legitimacy loss to innocent organizations in response to reported corporate deviance. Administrative Science Quarterly, 54 , 195–228. Kang, E. (2008). Director interlocks and spillover effects of reputational penalties from financial reporting fraud. Academy of Management Journal , 51 , 537–555. Karpoff, J. M., Lee, D. S., & Martin, G. S. (2008). The cost to firms of cooking the books. Journal of Financial & Quantitative Analysis, 43 , 581–611. Karpoff, J. M., Lee, D. S., & Vendrzyk, V. P. (1999). Defense procurement fraud, penalties, and contractor influence. Journal of Political Economy, 107 , 809–842. Karpoff, J. M., & Lott, J. R. (1993). The reputational penalty firms bear from committing criminal fraud. Journal of Law & Economics, 36 , 757–802. Karpoff, J. M., Lott, J. R., & Wehrly, E. W. (2005). The reputational penalties for environmental violations: Empirical evidence. Journal of Law & Economics, 48 , 653– 675. Kaufman, G. G. (1994). Bank contagion: A review of the theory and evidence. Journal of Financial Services Research, 8 , 123–150. Kawaller, I. G. (1994). Comparing Eurodollar strips to interest rate swaps. Journal of Derivatives, 2 , 67–79.

29

Kenourgios, D., Samitas, A., & Paltalidis, N. (2011). Financial crises and stock market contagion in a multivariate time-varying asymmetric framework. Journal of International Financial Markets, Institutions & Money, 21 , 92–106. Khalid, A. M., & Kawai, M. (2003). Was financial market contagion the source of economic crisis in Asia?: Evidence using a multivariate VAR model. Journal of Asian Economics, 14 , 131–156. Klein, B., & Leffler, K. B. (1981). The role of market forces in assuring contractual performance. Journal of Political Economy, 89 , 615–641. Lang, L. H., & Stulz, R. M. (1992). Contagion and competitive intra-industry effects of bankruptcy announcements: An empirical analysis. Journal of Financial Economics, 32 , 45–60. Levonian, M. (2001). Subordinated debt and quality of market discipline in banking. Federal Reserve Bank of San Francesco Working Paper. Longstaff, F. A. (2010). The subprime credit crisis and contagion in financial markets. Journal of Financial Economics, 97 , 436–450. McConnell, P. (2013). Systemic operational risk: The LIBOR manipulation scandal. Journal of Operational Risk , 8 , 59–99. Morgan, D. P., & Stiroh, K. J. (2001). Market discipline of banks: The asset test. Journal of Financial Services Research, 20 , 195–208. Murphy, D. L., Shrieves, R. E., & Tibbs, S. L. (2009). Understanding the penalties associated with corporate misconduct: An empirical examination of earnings and risk. Journal of Financial & Quantitative Analysis, 44 , 55–83. Nelson, K. K., Price, R. A., & Rountree, B. R. (2008). The market reaction to Arthur Andersen’s role in the Enron scandal: Loss of reputation or confounding effects? Journal of Accounting & Economics, 46 , 279–293. Nier, E., & Baumann, U. (2006). Market discipline, disclosure and moral hazard in banking. Journal of Financial Intermediation, 15 , 332–361. Perry, J., & de Fontnouvelle, P. (2005). Measuring reputational risk: The market reaction to operational loss announcements. Federal Reserve Bank of Boston Working Paper. Schenk, C. R. (1998). The origins of the Eurodollar market in London: 1955–1963. Explorations in Economic History, 35 , 221–238. Schoenmaker, D. (1996). Contagion risk in banking. LSE Financial Markets Group Discussion Paper.

30

Slovin, M. B., Sushka, M. E., & Polonchek, J. A. (1999). An analysis of contagion and competitive effects at commercial banks. Journal of Financial Economics, 54 , 197–225. Stephanou, C. (2010). Rethinking market discipline in banking : Lessons from the financial crisis. World Bank Policy Research Working Papers. Sturm, P. (2013). Operational and reputational risk in the European banking industry: The market reaction to operational risk events. Journal of Economic Behavior & Organization, 85 , 191–206. Tanimura, J. K., & Okamoto, M. G. (2013). Reputational penalties in Japan: Evidence from corporate scandals. Asian Economic Journal , 27 , 39–57. Wolfe, D. T., & Hermanson, D. R. (2004). The Fraud Diamond: Considering the four elements of fraud. The CPA Journal , 74 , 38–43.

31

Table 1: List of Events Date 24/03/2011 26/07/2011 09/12/2011 03/02/2012

10/02/2012 20/03/2012 27/06/2012 03/07/2012 05/07/2012 18/07/2012 31/07/2012 05/08/2012∗ 09/08/2012 16/08/2012 23/08/2012 07/09/2012 10/09/2012 15/10/2012

26/10/2012

29/10/2012? 15/11/2012 03/12/2012 11/12/2012 13/12/2012 14/12/2012 19/12/2012 20/12/2012 25/01/2013

Event Barclays has emerged as a key focus of the investigation by the US and UK regulators. Bank of America and Citibank have also received subpoenas. UBS confirmed the LIBOR investigation has widened scope to Yen rates. Citigroup and UBS face TIBOR penalties. Swiss authorities have launched a probe into 12 banks (Bank of Tokyo-Mitsubishi, Citigroup, Credit Sussie, Deutsche Bank, HSBC, JP Morgan, Mizuho Corporate Bank, Rabobank, RBS, Societe Generale, Sumitomo Mitsui Banking Corporation, and UBS) over claims they have been fixing their interbank lending rates. Citigroup was forced to write off $50m after two traders accused of attempting to influence global lending rates left the bank. Deutsche Bank gets data request in LIBOR probe. Barclays admitted to misconduct. The UK’s FSA and the US Department of Justice and the CFTC imposed fines worth $450m in total. Crown Office confirms investigation into Scottish banking sector. (Lloyds Banking Group, RBS) RBS withdrawn from TIBOR panel. Moody’s and S&P have lowered their outlook on Barclays from stable to negative amid the LIBOR rigging scandal. Investigations are focusing on Barclays, whose traders were the ringleaders of a circle that included Credit Agricole, HSBC, Deutsche Bank, and Societe Generale. Deutsche Bank has confirmed that a limited number of staff were involved in the LIBOR scandal. Credit Agricole , Deutsche Bank, HSBC, Rabobank, and Societe Generale are linked to the LIBOR rigging probe. Bank of Tokyo Mitsubishi has become the latest lender to face questions in the widening LIBOR rigging scandal. Barclays, Citigroup, Deutsche Bank, HSBC, JP Morgan, RBS, and UBS are to be questioned in the US for alleged LIBOR rigging. A former Singapore-based trader at RBS has opened a new window into how attempts were allegedly made to manipulate LIBOR. RBS in talks to settle LIBOR allegations that would cost it £200–300m. Trial begins of former UBS trader. A group of US homeowners is suing Barclays, Bank of America, JP Morgan, UBS, RBS, Citigroup, Rabobank, Credit Suisse, Deutsche Bank, HSBC, Lloyds Banking Group, and Royal Bank of Canada, claiming they are liable for their mortgage rates being artificially higher because of illegal LIBOR rigging. Subpoenas have been sent to: Bank of America, Bank of Tokyo Mitsubishi, Credit Suisse, Lloyds Banking Group, Rabobank, Royal Bank of Canada, Societe Generale, Norinchukin Bank, and WestLB. The banks joined the probe to increase the number of banks under investigation by the two state prosecutors to 16. First LIBOR damages trial set to proceed, a case brought by a care home operator against Barclays to go ahead. Canadian regulators investigating a half-dozen global banks in LIBOR manipulation probe have publicly rebuked RBS. UBS is in global talks to reach a settlement of more than $450m over the alleged manipulation of LIBOR. Three men have been arrested in connection with investigations into the LIBOR rigging. Hayes (UBS, Citigroup) and Two brokers (RP Martin). UBS faces $1bn fine over LIBOR allegation. UBS staff face LIBOR probe in the UK. UBS has agreed to pay $1.5bn to US, UK and Swiss regulators for attempting to manipulate the LIBOR inter-bank lending rate. Former UBS trader who faces criminal charges in the probe has been linked to traders at RBS, JP Morgan, Deutsche Bank and Citigroup. Ex-Barclays chiefs named in LIBOR case.

32

Date 06/02/2013 19/03/2013

11/04/2013 17/06/2013 18/06/2013 18/09/2013∗ 23/09/2013

21/10/2013

31/10/2013 08/11/2013 04/12/2013

Table 1 List of Events continued Event RBS has been fined $610m by UK and US authorities for its part in the LIBOR rigging scandal. Japanese banks accused of TIBOR fixing. Freddie Mac has sued more than a dozen banks (Bank of America, JP Morgan, UBS, Citigroup, Credit Sussie, and Deutsche Bank) and the British Bankers’ Association. UBS joins exodus from EURIBOR panel. Yen LIBOR probe focus on RBS. Former UBS and Citigroup trader Hayes has been charged by the Serious Fraud Office in connection with its investigation into the LIBOR rigging scandal. HSBC probed by Hong Kong regulator over HIBOR. A Japanese investment banking unit of UBS was ordered to pay a $100m criminal fine after pleading guilty to manipulate LIBOR. The US credit union regulator has filed an anti-trust lawsuit against 13 banks (UBS, RBS, Barclays, Societe Generale, Credit Suisse, JP Morgan, Lloyds Banking Group, WestLB, Raiffeisen Bank, Norinchukin Bank, Bank of Tokyo Mitsubishi, and Royal Bank of Canada) as part of the LIBOR rigging scandal. Former employees of Rabobank, RBS, Deutsche Bank, UBS, and ICAP were among 22 names that the UK Serious Fraud Office included as alleged co-conspirators on a draft indictment against Hayes, a former trader at both UBS and Citigroup who is facing criminal charges stemming from a probe into alleged LIBOR rigging. Fannie Mae sues 9 banks for $800m over LIBOR: Barclays, Deutsche Bank, Citigroup, Bank of America, UBS, RBS, Credit Sussie, JP Morgan, and Rabobank. Barclays and Deutsche Bank to face LIBOR claims in civil cases. The European Commission has fined six banks (RBS, Deutsche Bank, Societe Generale, JP Morgan, Citigroup, and RP Martin).

This table describes 39 events dates on which banks were accused of or sanctioned for manipulating LIBOR. ∗ Event day is on a weekend. ? Hurricane Sandy shuts down the stock market in the US.

33

34

Abbey National (Santander) Bank of America Bank of New York Mellon Bank of Nova Scotia Bank of Tokyo-Mitsubishi UFJ Barclays BBVA BNP Paribas Canadian Imperial Bank of Commerce Citigroup Credit Agricole Credit Suisse Deutsche Bank Goldman Sachs HSBC ING JP Morgan Chase Lloyds Banking Group Mizuho Bank Morgan Stanley Nordea Bank Royal Bank of Canada Royal Bank of Scotland Societe Generale Standard Chartered State Street Sumitomo Mitsui Banking Corporation UBS UniCredit Wells Fargo 2 1 1 1 1 2 1 1

1 2 1

3 2 2 3 2 4 1 4

2 3 1 3

2 1

GSIB∗

X X

X X

X X

X X X

X X X

X X

X

X X X X

X X X X

X

X X X X X X

X X

X X X

USD X

LIBOR X X

X X

X

X

X X X X

X

X

X

X

G14?

X

X X X

X X X

X

X X

X

X X

X

EUR

X

X X X

X X X

X

X

X X

X

X X

X

GBP

X X

X X

X X X

X

X

X

X X

JPY

X

X X

X X

X

X X

X

X X

CHF

X X X

X

X

X

X

X

X

CAD

X

X X

X

X

X

AUD

X

X X

X

X

X

NZD

? Hurricane Sandy shuts down the stock market in the US.

This table displays the interconnection between the three panels: GSIB, G14, and LIBOR. We construct a panel that consists of thirty banks from the three panels and name it “LIBOR Banks.” ∗ Global Systemically Important Banks; The bucket number developed by the Financial Stability Board (FSB) to measure the systemic importance of a bank from 1 (lowest) to 5 (highest).

Origin ESP USA USA CAN JPN GBR ESP FRA CAN USA FRA CHE DEU USA GBR NLD USA GBR JPN USA SWE CAN GBR FRA GBR USA JPN CHE ITA USA

Contributor Bank

Table 2: LIBOR Banks

X

X X

X

X

X

DKK

X

X X

X

X

X

SEK

Table 3: Sample Composition and Coverage by Country Country

Number of Banks

Number of Banks∗

Number of LIBOR Banks?

Canada France Germany Italy Japan Netherlands Spain Sweden Switzerland UK US

261 44 94 45 198 13 25 15 54 156 1,955

4 40 89 43 187 11 22 13 52 142 659

3 3 1 1 3 1 2 1 2 5 8

Total

2,860

1,262

30

∗ Banks that are used to construct the market portfolio for market return calculation ? Banks in the final sample

Table 4: Descriptive Statistics

R Rm ∆T reasury ADJF IN E

N

Mean

Standard Deviation

21,151 21,151 21,151 21,151

0.00 0.00 0.00 0.00

0.02 0.01 0.02 0.00

p25

p50

p75

–0.01 0.00 0.01 –0.01 0.00 0.01 –0.01 0.00 0.01 0.00 0.00 0.00

This table presents descriptive statistics for 21,151 observations. The sample period is March 2011–December 2013.

Table 5: Correlation Between Variables

R Event Rm ∆T reasury ADJF IN E

Event

Rm

∆T reasury

0.001 1 0.222*** 0.003*** 1 0.099*** 0.002 0.299*** 1 –0.000 0.131*** 0.000 0.002**

This table reports the strength and direction of correlations between the study variables. ***, ** represents statistically significant at the 1 and 5% level, respectively.

35

36

–1 0

1

Event Day –1 0

1

23 March 2011 24 March 2011

25 March 2011

Date

26 June 2012 27 June 2012

28 June 2012

£1.66

£1.92 £1.96

Closing Price

Barclays

£2.90

£2.87 £2.91

Closing Price

Barclays

–15.50%

–0.95% 1.90%

Daily Return

–0.31%

–0.43% 1.27%

Daily Return

0.21%

0.05% 1.38%

–0.42%

–0.61% 2.22%

–0.59%

–0.06% 0.81%

Mean Daily Re- FTSE250 turn (n=27) Return

Nonevent Banks

–0.18%

–0.03% 0.69%

Mean Daily Re- FTSE250 turn (n=27) Return

Nonevent Banks

Daily

Daily

–0.64%

–0.17% 0.85%

Financial Industry Daily Return

0.96%

–0.75% –0.43%

Financial Industry Daily Return

Barclays admitted to misconduct. The UKs FSA and the US Department of Justice and the CFTC imposed fines worth $450m in total.

News

Barclays has emerged as a key focus of the investigation by the US and UK regulators.

News

This table reports news and stock returns associated with the two key events: Barclays’ emergence as the key focus of the LIBOR investigation (March 24, 2011) and the first bank to have received financial penalty from the regulators (June 27, 2012). The sample of “Nonevent LIBOR Banks” consists of LIBOR banks that have not been accused of having participated in LIBOR manipulation on the event date. Thus, we have two different sample sizes for the two event dates reported in this table. The “Financial Services Industry Daily Return” is the return to a market-capitalization-weighted portfolio consisting of all the UK common stocks in Compustat with SIC code ranging from 6000 to 6500.

Event Day

Date

Table 6: The Initial Emergence of the LIBOR Scandal

Table 7: Reputational Penalties and the Contagion Effect

(1) Event N onevent N onLIBOR Rm ∆T reasury ADJF IN E Constant

Observations R-squared

(2)

–0.002** [–2.528] – – – – 1.087*** [18.872] 0.134*** [4.442] 0.056 [0.178] –0.000 [–1.296] 21,151 0.563

(3)

– – – – –0.001** – [–2.633] – – –0.000 – [–0.443] 1.076*** 0.398*** [17.592] [30.915] 0.133*** 0.040*** [3.414] [9.615] – – – – –0.000 0.000*** [–1.268] [9.729] 14,624 0.593

862,840 0.046

This table reports the coefficient of estimates from the regression Model 1 to 3. Column 1 shows that banks that either were accused of or admitted to misconduct experienced an average –0.6 percent cumulative abnormal return (CAR) during the three-day event window. The coefficient of in Column 1 is negative and statistically significant, indicating that banks accused of misconduct experienced negative abnormal returns during the periods surrounding the announcement of alleged manipulation. Column 2 reports the market reaction of nonevent LIBOR banks during the period when one or more LIBOR banks were accused of misconduct. Nonevent banks experienced an average –0.3 percent CAR during the three-day period. The coefficient of is significantly negative, suggesting that the LIBOR banks that were not accused of misconduct experienced negative abnormal returns during the period surrounding the announcement of manipulation, which we interpret as the contagion effect of the reputational damage. The coefficient of in Column 3 is not significant, so the reputational damage and the contagion effect of the reputational damage are specific to LIBOR banks. ***, ** represents statistically significant at the 1 and 5% level, respectively. [Robust t-statistics in brackets]

37

Table 8: Top Derivatives Dealers

(4) N onevent N onevent × T opDealers T opDealers Rm ∆T reasury Constant

N onevent + (N onevent × T opDealers)

Observations R-squared

0.000 [–1.377] –0.001** [–2.157] 0.000** [2.150] 1.076*** [17.591] 0.133*** [3.414] –0.000** [–2.146] –0.002*** [–3.140] 14,624 0.593

This table reports the coefficients of estimates from the regression Model 4. The coefficient on the interaction term between N onevent and T opDealers is negative and statistically significant, thus indicating that the contagion effect is stronger among large derivatives dealers. Since large derivative dealers had the strongest incentives to engage in LIBOR manipulation, results reinforce the intuition that contagion is channeled by a communality of incentives to commit the fraud. ***, ** represents statistically significant at the 1 and 5% level, respectively. [Robust t-statistics in brackets]

38

Table 9: Robustness

Panel A

N onLIBOR (n=30) N onLIBOR (n=30) Rm ∆T reasury Constant

Observations R-squared

(1)

(2)

0.000 [0.322] – – 0.694*** [8.129] 0.080** [2.639] –0.000 [–0.174]

– – 0.000 [0.730] 0.782*** [9.609] 0.067** [2.144] 0.000 [0.682]

22,365 0.293

22,445 0.249

(1)

(2)

Panel B

Event N onevent N onLIBOR Rm ∆T reasury ADJF IN E Constant

Observations R-squared

–0.002** [–2.118] – – – – 1.087*** [18.899] 0.134*** [4.449] 0.065 [0.208] –0.000 [–1.208] 21,073 0.563

39

(3)

– – – – –0.001** – [–2.379] – – –0.000 – [–0.863] 1.076*** 0.398*** [17.556] [30.930] 0.133*** 0.040*** [3.434] [9.629] – – – – –0.000 0.000*** [–1.266] [9.723] 14,552 0.593

857,196 0.046

Panel C (1) Event N onevent N onLIBOR Rm Treasury ADJF IN E Constant

Observations R-squared

(2)

(3)

–0.002 – – [–1.875] – – – –0.001** – – [–2.321] – – – –0.000 – – [–0.406] 1.087*** 1.076*** 0.398*** [20.381] [19.068] [33.847] 0.134*** 0.133*** 0.040*** [4.834] [3.905] [10.737] 0.056 – – [0.170] – – –0.000 –0.000 0.000*** [–1.237] [–1.305] [9.749] 21,151 0.563

14,624 0.593

862,840 0.046

The tables report results from several robustness tests. Panel A shows coefficients of estimates from the regression Model 3 based on the following benchmarking criteria corresponding each column: (1) a sample of 30 banks created by one-to-one matching on size with LIBOR banks in each country; and (2) a sample of 30 banks generated by one-to-one matching on size with LIBOR banks in the eleven countries. Panel B reports results from excluding all overlapping events and keeping only the first event. Finally, Panel C reports results from clustering the standard errors both at the firm and event level. ***, **,* represents statistically significant at the 1, 5, and 10 % level, respectively. [Robust t-statistics in brackets]

40