Hedging and the Failures of Corporate Governance: Lessons from the Financial Crisis


Published on

Published in: Business, Economy & Finance
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Hedging and the Failures of Corporate Governance: Lessons from the Financial Crisis

  1. 1. Hedging and the Failures ofCorporate Governance: Lessons from the Financial Crisis Rodrigo Zeidan 2012
  2. 2. Hedging and the Failures of Corporate Governance: Lessons from the Financial Crisis Rodrigo Zeidan Fundação Dom Cabral and Nottingham University Business School China rodrigo.zeidan@fdc.org.brAbstract:The paper identifies which failures of corporate governance allowed non-financial companiesaround the world to develop hedging strategies that led to hefty losses in the aftermath of thefinancial crisis. The sample is comprised of 346 companies from 10 international markets, ofwhich 49 companies (and a subsample of 13 distressed companies) lost a combined U$18.9billion. An event study shows that most companies that presented losses in derivativesexperienced negative abnormal returns, including a number of companies in which the effectwas persistent after a year. The results of a probit model indicate that the lack of a formalhedging policy, no monitoring to the CFOs, and considerations of hubris and remunerationcontributed to the mismanagement of hedging policies. For heavily distressed companies,there is evidence that higher ownership concentration implies in a lower probability ofdesigning speculative hedging positions.Keywords: Risk Management; Hedging; Derivatives; Monitoring; Corporate GovernanceStructureJEL Classification: G32; G34; G01 1 Electronic copy available at: http://ssrn.com/abstract=2011297
  3. 3. IntroductionThe purpose of this paper is to relate risk management and corporate governance byanalyzing the case of non-financial companies that posted hefty losses in derivatives tradingduring the financial crisis that started in 2007. Dodd (2009) estimates that for 12 countriesthat include Poland and economies of Asia and Latin America derivatives trading affectedpossibly 50,000 firms, with losses totaling roughly $530 billion. Kamil, Sutton and Walker(2009) present a small subsample of companies in Mexico (6 companies) with total losses ofU$4.7 billion (with an average loss of 23% of total assets) and 3 companies in Brazil withtotal losses of U$5.5 billion - and an average loss of 46% of total assets.There is one simple explanation for these losses: non-financial companies were hedgingfinancial positions – mainly currency exposures – and hence posted only book losses, with acounterpart gain in revenue from the positions being hedged. However, this simpleexplanation is insufficient to explain some relevant consequences from the disclosure of theselosses: many companies filed for bankruptcy; stocks plunged; some companies sued or weresued by the banks that sold the derivatives contracts; accounting rules were changed (inBrazil, India and China); and even the Chinese governments State-owned Assets Supervisionand Administration Commission (SASAC) got involved in trying to allow Chinese state-owned companies involved in derivatives losses to walk away from contracts withinternational banks (Reuters, 2009). The more plausible explanations are that companiespurposefully engaged in speculative positions and/or made mistakes in designing hedgingstrategies involving derivatives to offset foreign-exchange exposure. Irrespective of which 2 Electronic copy available at: http://ssrn.com/abstract=2011297
  4. 4. explanation addresses the question of why companies committed mistakes in derivativestrading, there remains the question of which corporate governance mechanisms failed inallowing shareholders to monitor the behavior of executives who were able to designdestroying value strategies involving derivatives.In this paper I evaluate the failures of corporate governance in monitoring risk managementstrategies in a sample of non-financial companies that posted derivatives losses resulting fromthe financial crisis. First, I show that the disclosure of losses stemming from derivativescontracts by non-financial companies resulted in negative abnormal results which are a clearindicator that the companies were either speculating with derivatives or made mistakes intheir risk management strategies. Then I use a probit cross-sectional model to compare thecorporate governance structure of companies that posted derivatives losses with companiesthat did not by establishing a dependent binary variable in which for the companies thatposted hefty losses it assumes value 1 and for the control group zero. Since we have a clearlydefined event – the losses – the probit model is suitable to analyze which corporategovernance mechanisms failed in preventing executives from implementing value-destroyingfinancial strategies involving derivatives.There is precious little empirical work done in the relationship between corporate governanceand risk management for cross-country companies. Disclosure issues and the difficulty ofarriving at usable data are partially responsible, but the lack of a comprehensive theoreticalframework also hinders empirical analysis. The present paper contributes to the literature inmany ways. First, it presents a comprehensive theoretical framework that explains why 3 Electronic copy available at: http://ssrn.com/abstract=2011297
  5. 5. international companies hedge. It then presents an analysis of a sample of companies thatposted heavy losses in derivatives trading on the wake of the recent financial crisis. Finally, itrelates these losses to failures in the corporate governance mechanisms of the selectedcompanies.The structure of the paper is as follows: the first section describes in the research questionand main goal. The second section presents a theoretical discussion on the link betweencorporate governance and risk management with the goal of developing a theoreticalframework that feed the next section, in which I develop the variables of the econometricmodel and describe the data. The fourth section brings the results and analysis, while the lastsection has some final comments. I. RISK MANAGEMENT AND CORPORATE GOVERNANCE.Tufano (1996) remarks (p.1097) that academics know remarkably little about corporate riskmanagement practices. Even though academia has catched up in the last 15 years [recentmodels include Purnanandam (2008) and Fehle and Tsyplakov (2005)], we are still ignoranton many risk management practices. Because risk strategies are not completely disclosed infinancial statements it is difficult to properly assess the extent of hedging policies and othermeasures of risk management. In the present paper there is a clear event in which riskmanagement strategies are unveiled as a consequence of the financial crisis. The lossesposted by public companies and the effects on stock prices reveal a case of risk management 4
  6. 6. gone wrong and I exploit it by relating it to the corporate governance structures of theaffected companies in different markets around the world.As for why companies use risk management strategies in the case of foreign-exchangeexposure, the main driver for a value-enhancing hedging policy is that in efficient marketsdiversification transfer risks from the companies to the market. However, as Stulz (1996) andBartram (2000) show, to really generate value hedging policies have to deal with one of thefollowing possible gains: reductions in bankruptcy and distress costs, reductions in expectedtax payments, reductions in expected payments to stakeholders and/or reductions in costs ofraising funds. Moreover, when we add the perspective of management, canonical agencytheory models assume that egotistical agents maximize utility and thus corporate governanceshould align this utility maximizing behavior with the interest of shareholders. Tufano (1996)shows that two variables related to executives matter in determining if companies hedge ornot: the amount of shares owned by managers and the nature of the managerial compensationcontract. The author shows empirically that managers maximize their utilities through riskmanagement in two ways: if managerial wealth is affected by share prices companies hedgesubstantially, with the converse being true – if management owns a small stake companieshedge little; and if executive compensation involve options or similar features then managersare more risk-prone and thus hedge less. More recent models, like Purnanandam (2008) andFehle and Tsyplakov (2005), expand the risk management theoretical literature by includingmore sophisticated hypotheses, but the main results remain the same – risk managementshould enhance value by dealing with one of the four characteristics present in Stulz (1996), 5
  7. 7. but is constrained by managers’ incentives given by executives’ ownership of shares and thestructure of managerial compensation.Even though the combination of corporate finance theory and agency theory explains mostpatterns in risk management practices, some recent literature has also been investigating theimpact of the institutional context on the behavior of management in designing hedgingpolicies. The main idea from behavioral finance is that managers are also affected by theinstitutional environment in which executives are embedded (Wiseman et al, 2012, is anexample of a theoretical contribution that highlights the institutional impact on riskmanagement theory). This means that legal issues, stakeholders’ proactivity, corporategovernance and even managers’ innate characteristics also influence risk managementpractices. For instance, Zhang (2009) argues that new transparency regulation in theAmerican market (SFAS 133) may have discouraged firms speculative use of derivativeinstruments; Bremer et al (2009) show that the provision of job security as a proxy foremployee interests has a significant effect on the likelihood of CFO dismissal and affect therisk-taking behavior of CFOs, thus affecting possible hedging strategies; Indjejikian andMatějka (2008) argue that firms mitigate earnings management or other misreportingpractices in part by deemphasizing CFO incentive compensation; Magnan et al (2008) findthat hubris (characterized by exaggerated self-confidence, arrogance and oblivion to realityby executives) may be a critical factor to understand corporate financial frauds; and Dionneand Triki (2005) find that better educated directors relate positively to better riskmanagement strategies and enhance corporate value. Regarding overconfidence as a possibleexplanation for the miscalculation of risks by executives, Ben-David et al (2006) show how 6
  8. 8. CFOs build overconfidence by, among other factors, a focus on a recent series of pastsuccesses. They also show that CFOs draw their “worst case scenario” from recentrealizations of the market and the firm, but upper confidence bounds are affected only by thepast 3-month stock market movement. The theoretical framework that I follow is the onepresent in figure 1 below. On it, risk management and hedging policies are the outcome of acombination of mechanisms based on finance theory, management incentives and the broaderinstitutional framework. PLEASE INSERT FIGURE 1 HEREIn the present framework corporate governance is one of the main features of the institutionalframework that help shape companies’ hedging strategies. There are two ways in which thecorporate governance structure of a company reflects on hedging: monitoring and efficiency.The monitoring aspect is clear because an effective corporate governance structure isdesigned to prevent strategies that are against the interest of shareholders, usually preventingspeculative positions with financial instruments that are supposed to be used to hedge.Specifically, non-executive directors are supposed to control risk-taking behavior or at leastalign it to the interest of shareholders. This point relates to efficiency seeking by allocatingcompany resources to create value. However, in the case of risk management, speculativepositions create value by exploiting private information in the hands of the company, butthere are two conditions for this to be carried out: internal transparency and privateinformation. 7
  9. 9. The empirical evidence on derivatives usage, however, shows unequivocally the relevance ofsuch instruments for non-financial companies. Earlier studies [e.g. Bodnar et al. (1995);Phillips (1995); Berkman and Bradbury (1996); Berkman et al. (1997); Howton and Perfect(1998); Bodnar et al. (1998); De Ceuster et al. (2000); Mallin et al. (2001)] analyze thepatterns of derivatives usage and its determinants, showing widespread usage of derivativesby non-financial companies and its relevance to corporate risk management. More recently,Bartram et al. (2009) show that, in their sample of 7,292 non-financial companies from 48countries, 59.8% of the companies use derivatives in general, while of those mostly usecurrency derivatives, followed by interest rate derivatives and commodity price derivatives.Also, other than the size of the local derivatives market no other country-specific factor issignificant, while the companies hedge for risk management purposes and not speculation.Regarding the risk management theory based on corporate finance, Bartram et al (2009) findthat companies are in line with the financial distress hypotheses, and tests indicate thatderivatives users have significantly higher leverage and income tax credits as well as lowerliquidity (as measured by quick ratios and coverage ratios). However, the authors also arriveat some evidence that runs counter to corporate finance theory - specifically, more profitablefirms and firms with fewer growth opportunities (market-to-book ratios) tend to hedge more.There is no explicit theory that relates corporate governance and risk management since thetheoretical background uses theories from corporate finance and agency theory, as previouslyobserved, but Dionne and Triki (2005) present some interesting empirical results regardingCFOs personal characteristics. They establish a relationship between corporate hedging and 8
  10. 10. the background and education of the board and the audit committee members. The authorsfind that financially educated directors seem to encourage corporate hedging whilefinancially active directors and those with an accounting background play no active role inhedging strategies. Dionne and Triki (2005) also find a positive relation between hedging andfirm performance, suggesting that shareholders are better off with financially educateddirectors on their boards and audit committees. II. HYPOTHESES, DATA AND METHODOLOGY.II.A - Data Description.Hundreds of companies posted losses with derivatives during the financial crises. However,in most countries there is no duty to report risk management strategies, and most companieswith losses are not listed companies. Also, many losses were rolled over a number of periods,were negotiated with banks or information was never made public. Because of suchconstraints, data for the present paper were hand collected following three criteria: lossesshould have been public, thus posted in the media through newspapers, websites, magazinesetc; financial statements should reveal those losses; finally, data on the corporate governancestructure of the companies (described in section 3.2) should have been able to be constructed.The idea behind using news media as a source of information has a long tradition in empiricalresearch, leading to the development of the methodology of event studies, which is used toanalyze abnormal stock returns given some new publicly available information. 9
  11. 11. The final sample has 49 companies in 10 financial markets (Australia, Brazil, China, HongKong, India, Indonesia, Japan, South Korea, Mexico and Poland). Companies did not postlosses in the same period due to different accounting systems and the timing of the financialcrisis, hence I use for each company and the other companies in the same market the financialyear in which losses are revealed. The earliest period in the sample is June of 2008 for Indiancompanies and the latest March of 2009 for some Chinese companies. I treat this window as asingle event for the purpose of relating it to the corporate governance structure of thecompanies at this and earlier periods and for the event study. The 49 companies in the samplelost a combined U$18.9 billion. Table I below presents figures for the absolute losses and theratio between the losses and revenues, as well as market capitalization. We can see that lossesaffected small and large companies and it did not discriminate by developing or developedmarkets or by regions – the sample include companies from five continents and the range ofrevenue is U$26 million (C-Motech from Korea) to U$33.7 billion (China Railway Group).Even though the sample is heavily skewed towards industrial and exporting companies thatpresent foreign exchange exposure, it also presents companies that were hedging other risks,such as oil prices. PLEASE INSERT TABLE I HERELosses as a percentage of revenue range from 0.6% to 88.1%, while as a share of marketcapitalization from 0.9% to 651%, which shows that companies are impacted in differentways. Because of this I create a subsample composed of companies that experienced majordistress. Major distress is defined as bankruptcy, acquisition by another company, major asset 10
  12. 12. sales (Win4Net had to sell its headquarters building) or a restructuring of derivative contractswith banks to avoid a default on the contracts. This subsample includes 14 companies (APNProperty, Aracruz, Sadia, Citic Pacific, Win4Net, C-Motech, Taesan, Baiksan, Kalbe Farma,Controladora Mexicana, Gruma, Vitro, Ropczyce and Odlewnie Polskie).II.B - Event Study.The idea behind the event study is to verify if the losses by the sample companies can beregarded as value destroying. If the risk management strategies were sound companies shouldonly suffer accounting losses and there should be no effect on stock prices. Here I follow thestandard methodology summarized in MacKinlay (1997). The main caveat is that there is noprecise way to define the exact period for the analysis since it is based on the date in whichthe media reported the event. For each company the event starts at time 0. As usual I presentresults for 7, 60 and 360-day windows. The abnormal rate of return (ARit) is the difference ^between the actual return (ri,t+e) and the forecast rate of return ( r i,t+e): ^ARite  ri ,t e  r i ,t e (1)The average and variation equations are based on Corhay and Tournai (1996), a GARCHmodel that accounts for time-varying volatility effects. Variables  t and hit1 arerespectively the information set at time t for firm i and the conditional variance. Equationsare: 11
  13. 13. rit  Ci  rmt   h ri ,t h   it (2)  it 1 ~ N (0, hit .d ), hit  i   ik  2 i ,t k   ij hi ,t  j (3)Results are in table II. The patterns we can see are: for distressed companies 11 of 14companies show significant negative abnormal returns in the first day, an effect that ispersistent for 10 companies one year later. For the non-distressed companies 21 of 35companies present abnormal negative returns in the first day, but only 7 companies stillpresent negative returns after one year. The sample companies use value-destroying hedgingstrategies, a result that was persistent over a period of time. PLEASE INSERT TABLE II HEREII.C - Hypotheses.The hypotheses for the econometric testing follow the theoretical framework presented infigure 1. Since the main objective is to relate corporate governance and risk management Ibuild hypotheses regarding how the governance structure of a company can impact the designof hedging strategies. Relating the theoretical framework to workable hypotheses isconditional on building variables that are constructible from financial statements and publicinformation on the corporate governance structure. I use previous research and focus on therole of the CFO and the underlying structure that allow the design of misplaced hedgingstrategies. Unfortunately, unlike Dionne and Triki (2005), I have no data on the backgroundof executives, but a comprehensive study of the companies’ governance structure yields other 12
  14. 14. qualitative indicators of the relationship between governance, agency theory and riskmanagement. The following hypotheses are tested in the econometric model:H1: No monitoring (MON): Lack of proper CFO monitoring increases the probability ofleveraged positions in derivatives.In many companies the CFO is directly responsible for designing and monitoring the riskmanagement strategy, presenting the results to the board. No good monitoring of the CFO’sstrategies can have a real impact, as exemplified by the Brazilian company Sadia. Thecompany redesigned its governance structure right after disclosing its losses in 2008. Figure 2shows the old and the redesigned governance structure regarding its risk managementstrategies. As we can see, the old structure gave too much discretionary power to the CFO,because he was responsible to monitor the strategy and present the no compliance report tothe rest of the Board. Data for MON is binary and come from the financial statements of thecompanies. It assumes value 1 for companies in which the CFO is responsible for managingand answering for the risk management strategy and 0 otherwise. PLEASE INSERT FIGURE 2 HERE.H2: Disclosure (DIS): Disclosure leads to better monitoring by shareholders and lessprobability of leveraged positions in derivatives. 13
  15. 15. Financial disclosure rules are different worldwide. In the United States the standard forfinancial reporting of derivatives, SFAS 133 Accounting for Derivative Instruments andHedging Activities, went into effect in 2001 and presents a series of compulsory financialdisclosures such as differentiating between hedging and speculation and requiring derivativecontracts to be marked-to-market and recorded as assets or liabilities on the balance sheet.Derivatives used in speculation are marked-to-market with gains or losses realized in thecurrent periods income. For many countries, such as Brazil, India, and China, disclosure onderivatives dealings was optional before the crisis, with most companies choosing to shareminimal or no information. In Brazil, the regulatory agency requested, in October 17th 2008and weeks after the first news regarding the hefty losses of companies such as Sadia andAracruz, that public companies disclose more information on derivatives. Since then furtherregulations have been enacted to tighten the disclosure of such information, with CVM(Comissão de Valores Mobiliários) requesting account restatements of some companiesregarding derivatives trading in 2010. I build a qualitative binary variable regarding thequality of disclosure of derivatives information. Data come from the notes of the financialstatements of the companies.H3: Shares in the American Market (USA): Compliance with SEC rules leads to bettermonitoring by shareholders and less probability of leveraged positions in derivatives.If a company has shares in the American market it has to disclose its derivatives dealings,which provides even more opportunities for shareholders’ monitoring. USA is a binary 14
  16. 16. variable, assuming value 1 when the company does not have shares in the American market.Data come from the companies’ websites.H4: Concentration (CON): Higher ownership concentration enhances monitoring anddecreases the probability of leveraged positions in derivatives.Ownership concentration is a standard variable in empirical corporate governance studies.Even though dispersed capital enhances value, it creates monitoring problems in relation torisk management strategies. CON is here defined as the proportion of shares in the hands ofthe three major shareholders, hence it is a continuous variable in the (0,1) interval. Data comefrom Datastream and Compustat Global.H5: Institutional Investors (IIN): Institutional investors in the Board enhance monitoringand decrease the probability of leveraged positions in derivatives.Participation of institutional investors on the board should help monitoring of financialmanagement. Data for the binary variable come from the financial statements or companies’websites.H6: Formal Hedging Policy (FHP): lack of a formal hedging policy increase the probabilityof leveraged positions in derivatives. 15
  17. 17. Many companies specify a formal hedging policy – Air China, for instance, hedge at most50% of its oil costs. Such policies should be a determent for speculative positions inderivatives. Data for the binary variable come from the financial statements or companies’websites. Variable assumes value 1 when the company has a policy in place and 0 otherwise.H7: Trend of Major Source of Risk (TRE): a clear trend on the risk being hedged increasethe probability of a leveraged position in derivatives.Hubris is the most difficult proposition to test for regarding agency theory, since it is almostimpossible to prove intent. I provide two indirect measures that account for the possibility ofhubris through the creation of an environment conductive to overconfidence, following Ben-David et al (2006). The first is a medium-term trend in the risk being hedged. Years ofcurrency appreciation, for instance, may build confidence in leveraged positions inderivatives to exploit the continuation of the trend. The variable is designed as the lineartrend of the underlying risk in the last 3 years, which means that for Brazilian companies it isthe linear trend of the Real, for Korean companies the Won, and for companies that werehedging oil prices it is the trend of oil (Brent) prices.H8: Recent Financial Results (FIN): recent financial gains increase the probability of aleveraged position in derivatives.The second variable pertaining to an indirect indication of hubris is a qualitative binaryvariable that represents the last three quarter of financial results. It indicates if the company 16
  18. 18. posted better than market-average financial results, and assumes value 1 only if the result ofthe company beat the market in all three previous quarters. The idea is that, following Ben-David et al (2006), positive reinforcement breeds overconfidence. Data come from financialstatements.H9: Remuneration (REM): a remuneration package for the CFO that is based on short-term incentives increase the probability of a leveraged position in derivatives.The variable relate to the design of the remuneration of the CFO. If remuneration is tied tomedium-term performance, it assumes value 0, otherwise it assumes value 1. The rationale isthat short-term incentives are tied to excessive risk-taking. Data come from companies’websites and corporate governance reports.H10: Management Stake (MAN): Higher management stake increase the probability of aleveraged position in derivatives.Following Tufano (1996), MAN is a continuous variable that is the average of themanagement stake in the previous three years.Variables relating to independent directors and CEO duality, which are usually a mainstay ofcorporate governance studies, were dropped because of poor predictability in the model. It issafe to argue that those variable show a poor relation to risk management strategies, sinceneither independent directors nor CEOs as Chairman of the Board would, in principle, be able 17
  19. 19. to curtail or give incentives of excesses in risk management strategies just by beingindependent or by acquiring a dual role in a company. Even though independent directors andCEOs who are not Chairmen of the Board are usually figures that result in better governancein the literature, in the present case we cannot see how they impact in a better riskmanagement monitoring role. The failures in risk management seem to be institutional innature in the present case. If incompetence plays a role, it is not one borne out of the role ofindependent directors and CEOs as Chairmen, but it is the result of Directors, including theCFOs, who weren’t able to curtail excesses in risk management strategies regardless of theirinherent role in the companies.II.D - Probit cross-sectional model.The dependent variable of the model that tests the relationship between corporate governanceand risk management is a binary variable which assumes value 1 for the companies thatposted derivatives losses and value 0 for companies that did not. The selection process ofcompanies for the test is very simple: public companies from the same market and industry asthe companies that posted losses, restricted by data availability. As an example, for theBrazilian market the selection comprises 30 companies from petrochemical, steel, food, pulp,and textile industries. For the 10 international markets the total of selected companies,including the sample, is 346. Since the dependent variable is binary, the resulting model is aprobit cross-sectional model. The general model is given by equation 1, in which C is thevector of controls and D the vector of sector dummies: 18
  20. 20. 1 MON   2 DIS   3USA   4 CON   5 IIN   6 FHP   7TRE  Yi  1 0 (1)  8 FIN   9 REM  10 MAN   C   D   i The resulting control vector (after iterations of the model with other financial variables) iscomposed of:Family-Owned: binary variable, resulted from hand-collected data;Age: continuous variable with data coming from the companies’ websites;Leverage: continuous variable based on the average financial debt ratio for the last threeyears. Data came from financial statements.As for dummies, I used market dummies in the first round of estimation, one for eachinternational market, but none improved the models. I dropped it in presenting the finalresults.The error term should capture all variation that is not explained by the selected variables. It isimpossible to perfectly model risk management decisions. In the present context we cannotexpect that the constructed variables can capture all the decision making process regardingthe losses in derivatives. The cognitive process that leads to decision making is truly multi-dimensional and the present variables can at most capture the incentives that may lead to thedecision to overhedge or speculate with derivatives. Ideally one should have very descriptivedata on the risk management department and the CFO characteristics, plus its relationshipwith the Board. Since such data do not exist or is unavailable, we should not expect an 19
  21. 21. overtly fit model to the available data, even though the theoretical framework presentlydeveloped should be appealing. III – RESULTS AND ANALYSIS.I divide the results in two, one for the whole sample of 49 companies and one for thedistressed sample composed of 13 companies. For both samples the dependent variableassumes value 1 for affected companies that lost in derivatives and value 0 for the othercompanies in the sample. The main diagnostic test is the cross-dependence (CD) test basedon Pesaran (2004) and Hsiao et al (2007) that indicates independence across the sample (it isbased on the Lagrange Multiplier, and the null hypothesis is for cross-section independence -H0 : R = IN).The probit econometric model is run using STATA 10.0 and is based on equation 1. Somesensitivity analyses are also performed. In particular, many other controls are used in firsttrials of the model, such as liquidity, price/earnings ratio and other financial variables. Allfinancial variables other than leverage result in poorer modeling performance and aredropped. Market dummies are also dropped due to poor performance. Table II presents theeconometric results and table III the marginal effects of the probit model. Controls areomitted for brevity, but no control other than financial leverage (and only for distressedcompanies and even so, marginally) is significant. 20
  22. 22. PLEASE INSERT TABLE III HERE. PLEASE INSERT TABLE IV HERE.Results for the whole sample show that corporate governance plays an important role in thedesign of risk management strategies (in the present case, the dependent variable when itassumes values 1 represents mismanagement of derivatives), as does the trend of the sourceof risk and the remuneration incentives of the CFO. For the whole sample three hypothesesrelating to corporate governance are statiscally significant: lack of formal monitoringstructures (MON); shares in the American market (USA); and formal hedging policy (FHP).All post the expected sign, with lax monitoring resulting in higher probability of leveragewith derivatives, and shares in the American market and formal hedging policy acting asdeterrents to mismanagement of such instruments. The marginal probabilities are notparticularly high, but are significant nevertheless. No monitoring structure enhances theprobability of mismanagement of derivatives, in the present model, in 2.3%, while shares inthe American market and a formal hedging policy decrease the probability by 1.9% and1.2%, respectively. Hypothesis 2 - disclosure does not present a significant impact, and itsexplanation is that most companies in a single market usually follow the same disclosurerules and thus we can conclude that voluntary disclosure is lacking in the markets analyzed.Hypothesis 7, relating to the trend of source of risk variable posts an interest result. It is anindicative of an incentive to overconfidence by building strategies based on previous results.It was certainly used as an explanation in the media by executives – in an interview the CFOof Companhia Siderúrgica Nacional of Brazil dismissed the hefty losses by arguing that thederivatives strategies have netted sufficient gains in the past to make it worthwhile, even if 21
  23. 23. shares dropped heavily after the company’s losses were announced. As for hypothesis 9,remuneration, it clearly shows that there is an incentive for CFOs to hedge more if theircompensation is tied to short-term performance, as the probability of hedging increases withthis incentive.Results change somewhat when we consider the sample with distressed companies.Monitoring and shares in the American market are not significant anymore whileconcentration and management stake play a role in the case of companies who suffered themost with derivatives losses. Hypothesis 4, concentration, does not present the expected sign.It is expected that market monitoring through dispersed shares enhances the probability ofgiving the correct hedging incentives, but the results show that higher concentrationdecreases the probability of major distress in derivatives dealings by 1.5%. In the marketsthat comprise the sample there is a culture of ultimate owners with high levels of control,which results in more incentives to monitor high leverage. In fact, the same reason that makesacademics argue that dispersed companies are in general more efficient may have resulted, inthe specific case of speculation with derivatives, in a more risk-taking position by managers,in contrast with the usually more cautious and centralized approach when a company has asingle or a small group of owners. Also, the result corroborate Tufano (1996), since bothreasons raised by the author, represented in the present sample by remuneration andmanagement stake, are statistically significant. 22
  24. 24. III.A - Implications for the Regulation of Financial Markets.The results are especially significant if we think in terms of regulation of stock markets.Ultimately shareholders and stakeholders shared the burden in the losses by the companies.The results of the two models show that skewed incentives for the managers coupled withsome lax governance structure (especially a formal hedging policy and no monitoring)contributed to the companies’ mismanagement of the hedging policies. There are twoimportant implications: we can argue that it is the responsibility of shareholders to effectivelymonitor the companies’ strategies and hence the issue was not one of a market failure or lackof regulation; or we can argue that for incipient and developing markets the evolution ofregulation should embody rules and codes that prevent the possibility of the design ofpossible deleterious strategies. Since we find no evidence of mismanagement, in the contextof the financial crisis, in the American or other developed markets, it may be assumed thatmarket failures in developing countries contributed somewhat to the effects experienced bythese companies. If we take the example of Brazil, in which the regulatory agency (CVM –Comissão de Valores Mobiliários) requested ex-post that companies disclose their derivativesposition, we can see a reaction to an acknowledgement that transparency played a role. Eventhough I found no evidence that transparency was statiscally an issue – after all, companiesthat did not lose with derivatives were not compelled to disclose their position – we can gobeyond the simple transparency prescription to address issues of the design of corporategovernance structures. As figure 2 shows, the checks and balances of risk managementstrategies were more important than simple disclosure issues. Not only investors can learnfrom what happened with these companies, but the regulatory agencies now have subsidies to 23
  25. 25. request formal structures that comply with a situation in which companies should not beallowed to unwittingly speculate with derivatives. In developing markets corporategovernance structures matter more than in mature markets – hence the development offeatures like the New Market (Novo Mercado) in Brazil. It should be the purpose of theregulator to help shareholders prevent hubris and incompetence from hurting not onlycompanies, but as we can see from the hefty losses, whole markets.Can markets alone prevent developments like the positions in derivatives that bankruptedsome companies and yielded major losses to others? If markets are truly efficient theleveraged position of companies is already part of the shareholders’ portfolio. However, weknow that earlier derivatives lessons, like the one from Metallgesellschaft (MG) - whichposted losses of over U$1 billion in the mid-90s, have not prevented companies around theworld to leverage their positions, intentionally or not, in these instruments. In developingmarkets regulation is supposed to foster a better business environment, especially becauseinformation asymmetry and lack of liquidity prevent full market efficiency. In the case ofnon-financial companies, regulations like the one in Brazil which now requires betterdisclosure of derivatives positions, is a step forward in that direction. However, regulatorsshould also recognize that disclosure alone is insufficient. As we can see from the results,hubris may have played a role, as did skewed incentives to management throughremuneration and management stakes. Moreover, the simple lack of a formal hedging policy(Air China had one in place which allowed management to hedge at most 50% of oil costs)also contributed to these losses. In fact, not all the losses were derived from speculativepositions, but in all cases it caught management and investors unaware, and such failures 24
  26. 26. should be avoided in the future. Better governance regulation imply, for instance, changes inthe relationship between monitoring of risk management strategies, while leaving forshareholders to devise correct incentives for CEOs in relation to risk-taking positions. IV. FINAL COMMENTS.The main goal of the paper was to establish a relationship between corporate governance andrisk management by focusing on the case of companies that posted heavy losses inderivatives during the financial crisis. I built a sample of 49 companies from 10 internationalmarkets, from Latin America, Europe, Asia and Oceania. The combined losses of thesecompanies were a combined U$18.9 billion. Moreover, 13 of these companies went intobankruptcy or suffered heavy restructuring, being acquired by other companies and such.For the purpose of relating corporate governance and risk management first I built atheoretical framework that encompasses regular finance models with agency theory andbehavioral finance. To test hypotheses concerning this theoretical framework I hand-collecteddata on many qualitative indicators of corporate governance, proxies for hubris and othermanagement characteristics such as the remuneration scheme and the stake of management inthe companies. The empirical testing through a probit cross-sectional model reveals that somecorporate governance characteristics, such as the inexistence of formal hedging policy, laxmonitoring of the risk management department and dispersed ownership concentration arerelevant to the mismanagement of derivatives instruments. Results also show that variables 25
  27. 27. relating to overconfidence and incentives to executives are also relevant. The main impact ofthis study is related to the regulation of stock markets – it clearly shows that the failures inrisk management are due to reasons not only related to transparency. In fact, the firstresponse by market regulators to the distress of many companies was change disclosurepolicies regarding derivatives instruments. While better disclosure policies is probablyeffective in allowing shareholders better monitoring, we can clearly see from the results thatthe losses experienced by the selected companies resulted from more complex issues thanthat. Since the analyzed markets are mostly in developing countries, with relevant issues likeimperfect and asymmetric information being part of the environment, regulators should lookinto designing policies that correct incentives for proper risk management by non-financialcompanies.Several avenues of research remain. This study only analyzed a sample of 49 companies in across-sectional study. Future research should complement the present one by focusing on theevolution of risk management strategies. Are companies devising better governancestructures to curb future losses? The finance literature is full of examples of companies thatposted hefty derivative losses, and still companies around the world lost an estimated U$500billion during the financial crisis. As other markets develop, new companies enter and historyis forgotten there is no guarantee that a new cycle of derivatives losses will not happen.Social welfare should not be affected by events like this, but in many developing marketswidespread losses lead to less liquidity and hamper the development of capital markets. 26
  28. 28. REFERENCES.Bartram, S.M. (2000) Corporate Risk Management as a Lever for Shareholder ValueCreation. Financial Markets, Institutions, and Instruments, 9(5). 279-324.Bartram, S.M., G. Brown and F.R. Fehle (2009) International evidence on financialderivatives usage, Financial Management, 38(1), 185-206.Ben-David, I., J.R. Graham and C.R. Harvey (2006) Managerial Overconfidence andCorporate Policies. Working Paper. Duke University.Berkman, H. and M.E. Bradbury (1996) Empirical Evidence on the Corporate Use ofDerivatives. Financial Management, 25(2), 5–13.Berkman, H., M.E. Bradbury and S. Magan (1997) An International Comparison ofDerivatives Use. Financial Management, 26(4), 69–73.Bodnar, G. M., G.S. Hayt and R.C. Marston (1995) Wharton survey of derivatives usage byUS non-financial firms. Financial Management, 24(2), 104–114.Bodnar, G. M., G.S. Hayt and R.C. Marston (1998). 1998 Wharton survey of derivativesusage by US non-financial firms. Financial Management, 27(4), 70–91.Bodnar, G. M., A. de Jong and V. Macrae (2003) The Impact of Institutional Differences onDerivatives Usage: a Comparative Study of US and Dutch Firms. European FinancialManagement, 9(3), 271–297.Bremer, D., J.P. Lüdtke, A. Richter and U. Schäffer (2009) Who Disciplines the CFO? AnAssessment of Stakeholder Power in Corporate Governance, European Business SchoolResearch Paper Series 09-05. 27
  29. 29. Corhay, A. and Tournai, R. (1996) Conditional Heteroskedasticity Adjusted Market Modeland an Event Study. The Quarterly Review of Economics and Finance, 36(4), pp. 529-538.De Ceuster, M.J.K., E. Durinck, E. Laveren and J. Lodewyckx (2000) A survey into the useof derivatives by large non-financial firms operating in Belgium. European FinancialManagement, 6(3), 301–318.Dionne, G. and T. Triki (2005) Risk management and corporate governance: The importanceof independence and financial knowledge for the board and the audit committee, Workingpaper, HEC Montreal.Fehle, F. and S. Tsyplakov (2005) Dynamic Risk Management: Theory and Evidence,Journal of Financial Economics, 78(1), 3-47Howton, S.D. and S.B. Perfect (1998) Currency and Interest-Rate Derivatives Use in USFirms. Financial Management, 27(4), 111–121.MacKinlay, A.C. (1997) Event Studies in Economics and Finance, Journal of EconomicLiterature 35(1), 13-39.Magnan, M., D. Cormier and P. Lapointe-Antunes (2008) Like Moths Attracted to Flames:Managerial Hubris and Financial Reporting Frauds, cahier de recherche, Chaired’information financière et organisationnelle (ESG-UQAM).Mallin, C., K. Ow-Yong and M. Reynolds (2001) Derivatives usage in UK non-financiallisted companies. The European Journal of Finance, 7(1),63–91.Phillips, A.L. (1995) 1995 Derivatives Practices and Instruments Survey. FinancialManagement, 24(2), 115–125. 28
  30. 30. Purnanandam, A. (2008) Financial distress and corporate risk management: Theory andevidence, Journal of Financial Economics, 87, 706-739.Reuters (2009) Beijings derivative default stance rattles banks, August 31,http://www.reuters.com/article/2009/08/31/china-derivatives-idUSSP4732742090831.Tufano, P. (1996) Who Manages Risk? An Empirical Examination of the Risk ManagementPractices in the Gold Mining Industry. Journal of Finance 51(4), 1097-1137.Wang, Z., Salin, V., Hooker, N., and Leatham, D. (2002) Stock Market Reaction to FoodRecalls: a GARCH Application, Applied Economic Letters, 9, pp. 979-987.Wiseman, R.M., G. Cueva-Rodríguez and L.R. Gomez-Mejia (2012) Towards a SocialTheory of Agency, Journal of Management Studies, 49(1), 202-222.Zhang, H. (2009) Effect of Derivative Accounting Rules on Corporate Risk-ManagementBehavior, Journal of Accounting and Economics, 47(3), 244-264. 29
  31. 31. Figure 1 – Theoretical Framework that impact the design of Hedging Strategies. Hedging Strategies 30
  32. 32. Figure 2 - Old and New Risk Management Structure of Sadia.Figure 2 shows the change in structure of Sadia after it lost U$1.29 in currency derivatives in 2008. The oldstructure allowed too much discretion to the CFO, while the new structure reveals better governance for riskmanagement. Sadia was acquired by its main competitor, Perdigão, in May of 2009. The resulting company iscalled Brasil Foods. In 2006 Sadia attempted a hostile takeover of Perdigão but was not successful. Source: adapted from Sadia (2008). 31
  33. 33. Table I – Companies and Losses with Derivatives in the Financial CrisisTable I reports data on the 49 companies in the sample with losses in derivatives in the year 2008. Columns givethe companies’ names, country of origin, the book value of losses, and the ratios of those losses, annual revenueand market capitalization. Market capitalization is at the immediate date before the reported losses. Losses Losses/ Losses/ Company Country (U$ million) Revenue Market Cap APN EUROPEAN RETAIL PROPERTY Australia 86.14 88.1% 651.0% WESTFIELD TRUST Australia 504.79 35.6% 2.3% BRASKEM Brazil 737.75 7.5% 10.9% SADIA Brazil 1,290.52 20.8% 24.3% COMPANHIA SIDERÚRGICA NACIONAL Brazil 863.15 11.3% 10.8% EMBRAER SA Brazil 103.02 1.6% 2.4% ARACRUZ Brazil 1,145.91 32.2% 34.5% VICUNHA TEXTIL SA Brazil 81.20 13.3% 21.8% AIR CHINA China 281.13 3.7% 1.8% CHINA COSCO HOLDINGS China 338.36 1.8% 3.4% CHINA EASTERN AIRLINES China 857.84 14.2% 11.5% CHINA RAILWAY CONSTRUCTION China 180.92 0.6% 2.1% CHINA RAILWAY GROUP China 595.74 1.8% 6.1% SHENZHEN NANSHAN POWER CO China 30.56 6.7% 6.8% CHINA HAISHENG JUICE HLDG Hong Kong 24.54 11.3% 14.7% CITIC PACIFIC Hong Kong 2,084.07 35.0% 27.5% AUROBINDO PHARMA India 66.68 9.4% 6.1% HCL TECHNOLOGIES India 68.41 2.9% 0.9% KPIT CUMMINS INFOSYSTEMS India 20.58 11.3% 5.4% RAJSHREE SUGARS & CHEMICALS India 3.27 3.7% 12.9% SABERO ORGANICS GUJARAT India 3.19 3.7% 3.1% SUNDARAM MULTI PAP India 5.65 19.0% 8.2% ZEE ENTERTAINMENT ENTERPRISE India 20.34 4.1% 2.5% ANEKA TAMBANG Indonesia 27.74 3.0% 1.4% ELNUSA Indonesia 3.22 1.2% 1.5% KALBE FARMA Indonesia 22.56 2.4% 1.1% TIMAH Indonesia 12.57 1.6% 1.0% AJINOMOTO Japan 372.58 3.2% 5.8% ARIAKE JAPAN Japan 8.01 3.8% 1.6% SAIZERIYA Japan 160.45 18.8% 20.5% BAIKSAN Korea 25.45 15.6% 22.3% C-MOTECH Korea 5.37 20.0% 26.8% DAEWOO SHIPBUILDING & MARINE Korea 1,773.05 15.9% 28.2% HAN KWANG Korea 3.68 10.0% 19.2% MONAMI Korea 18.51 7.5% 10.2% TAESAN LCD Korea 685.67 81.3% 94.2% UJU ELECTRONICS Korea 10.90 13.3% 5.4% WIN4NET Korea 39.28 56.0% 71.9% ALFA Mexico 537.29 5.2% 6.8% CEMEX Mexico 1,480.93 7.3% 21.9% CONTROLADORA COML MEXIC Mexico 2,187.34 45.7% 50.8% GRUMA Mexico 1,321.16 32.9% 10.8% GRUPO INDUSTRIAL SALTILLO Mexico 215.65 24.4% 45.3% GRUPO POSADAS Mexico 128.58 20.8% 18.9% VITRO Mexico 368.03 14.1% 20.5% APATOR Poland 11.96 7.7% 4.2% ODLEWNIE POLSKIE Poland 47.39 77.3% 46.6% ZAKLADY MAGNEZYTOWE ROPCZYCE Poland 23.91 12.7% 9.0% ZELMER Poland 10.91 5.4% 4.5% 32
  34. 34. Table II -Abnormal Daily Stock Market Performance after Announcement of Derivative Losses.Table II reports cumulative abnormal return (CARt) up to the specified day t in event time. Column 1 has eachcompany, followed by the market, the date that the media announced the derivative loss of each company. Column 4is the stock return on the day of announcement. Event time is days relative to the media announcement of derivativelosses. Abnormal returns are computed given the market model parameters which are estimated with the GARCHmodel rit  Ci  rmt   h ri ,t h   it through the period [-190; -10] in event time, where rit is the continuouslycompounded local return on date t in country i and rmt is the continuously compounded daily market return on indexon day t. The sample period runs from 90 trading days before the date in the third column to the dates in columns 5 to8. Values in bold are statistically significant at 5%. Table is divided in two categories: major distressed companies andnon-distressed companies. Major distress is defined as bankruptcy, acquisition by another company, major asset salesor a restructuring of derivative contracts with banks to avoid a default on the contracts. Company Market Date Day 1 t=0 t=7 t=60 t=360 Major Distress APN EUROPEAN RETAIL PROPERTY Australia 8/26/2008 -5.35% -1.948 -4.030 -2.307 -16.999 SADIA Brazil 9/25/2008 -35.50% -32.439 -41.945 -38.778 -60.477 ARACRUZ Brazil 9/29/2008 -19.60% -17.009 -42.741 -55.820 -74.124 CITIC PACIFIC Hong Kong 10/21/2008 -55.10% -54.972 -45.952 -35.381 -19.103 C-MOTECH Korea 6/2/2008 -9.50% -5.599 -1.966 -2.297 -14.492 BAIKSAN Korea 7/7/2008 -8.28% -7.414 -2.675 -7.244 5.901 TAESAN LCD Korea 8/18/2008 -14.88% -14.337 -9.992 -11.930 -16.558 WIN4NET Korea 8/14/2008 -14.85% -12.894 -8.550 3.522 4.424 KALBE FARMA Indonesia 10/10/2008 -5.54% -1.796 2.890 7.267 13.921 CONTROLADORA COML MEXIC Mexico 10/8/2008 -75.39% -74.325 -72.157 -56.709 -31.301 GRUMA* Mexico 10/13/2008 -55.09% -53.260 -51.672 -48.449 -24.593 VITRO Mexico 10/10/2008 -20.13% -16.988 -15.483 -1.932 -2.075 ODLEWNIE POLSKIE Poland 10/15/2008 -5.03% -1.472 -12.025 -36.975 -35.190 ZAKLADY MAGNEZYT. ROPCZYCE Poland 11/26/2008 -8.57% -8.565 -22.680 -34.576 -19.400 Non-Distress WESTFIELD TRUST Australia 8/26/2008 -3.36% -1.371 -2.298 -5.031 -2.030 BRASKEM Brazil 10/3/2008 -6.30% -3.302 -0.874 -1.134 -1.811 COMPANHIA SIDERÚRGICA NAC. Brazil 10/1/2008 -5.50% -1.894 2.229 1.344 0.277 EMBRAER SA Brazil 11/4/2008 -3.40% -1.382 0.380 1.097 5.235 VICUNHA TEXTIL SA Brazil 11/11/2008 -11.11% -10.574 -6.992 -9.859 -12.954 AIR CHINA China 10/16/2008 -10.01% -8.704 -1.620 -1.204 -2.059 CHINA COSCO HOLDINGS China 12/12/2008 -5.20% -5.178 -0.719 0.524 -1.287 CHINA EASTERN AIRLINES China 10/24/2008 -5.27% -2.151 2.185 1.353 0.323 CHINA RAILWAY CONSTRUCTION China 10/22/2008 -9.97% -6.169 -1.262 -0.276 0.155 CHINA RAILWAY GROUP China 10/10/2008 -3.23% 0.101 1.724 3.175 8.523 SHENZHEN NANSHAN POWER CO China 10/24/2008 -3.16% 0.944 3.235 -0.622 0.861 CHINA HAISHENG JUICE HLDG Hong Kong 10/22/2008 -14.26% -13.266 -8.402 -12.109 -22.797 AUROBINDO PHARMA India 3/4/2008 -3.73% -0.806 0.964 0.006 -0.006 HCL TECHNOLOGIES India 7/11/2008 -5.62% -1.719 1.050 0.163 -0.113 KPIT CUMMINS INFOSYSTEMS India 4/28/2008 -12.41% -9.111 -5.782 -1.232 -7.201 RAJSHREE SUGARS & CHEMICALS India 3/17/2008 -8.74% -6.872 -4.159 -1.825 0.224 SABERO ORGANICS GUJARAT India 1/7/2008 -5.00% -0.044 0.775 2.117 9.442 SUNDARAM MULTI PAP India 1/16/2008 -4.51% 0.299 1.358 1.762 2.812 ZEE ENTERTAINMENT ENTERPRISE India 6/10/2008 -3.53% -1.808 0.490 0.338 0.129 ANEKA TAMBANG Indonesia 10/17/2008 -7.85% -4.250 -1.705 -3.523 -11.033 ELNUSA Indonesia 10/8/2008 -13.03% -9.257 -5.957 -7.796 -12.609 TIMAH Indonesia 10/6/2008 -31.28% -29.182 -25.012 -13.640 -5.518 AJINOMOTO Japan 10/16/2008 -10.88% -10.801 -4.835 -1.573 -0.595 ARIAKE JAPAN Japan 10/22/2008 -9.95% -6.825 -2.696 -1.365 2.386 SAIZERIYA Japan 11/27/2008 -15.56% -14.192 -10.135 -2.068 -1.388 DAEWOO SHIPBUILDING & MARINE Korea 10/16/2008 -12.64% -8.421 -4.570 1.703 -2.973 HAN KWANG Korea 7/2/2008 -21.84% -20.952 -9.745 2.335 7.747 MONAMI Korea 7/9/2008 -19.70% -16.210 -12.135 -3.711 -1.399 UJU ELECTRONICS Korea 7/7/2008 -6.70% -2.664 1.429 -1.142 2.970 ALFA Mexico 10/6/2008 -9.59% -5.765 -3.643 2.777 5.445 CEMEX Mexico 10/1/2008 -7.10% -6.925 -6.070 -2.678 -5.042 GRUPO INDUSTRIAL SALTILLO Mexico 10/8/2008 -13.03% -9.903 -7.503 -21.964 -6.631 GRUPO POSADAS Mexico 10/14/2008 -13.33% -10.616 -10.207 -12.298 -1.338 APATOR Poland 10/16/2008 -8.79% -4.853 -1.223 -0.275 0.151 ZELMER Poland 10/14/2008 -5.53% -5.565 2.680 -0.770 0.557*The Mexican Stock Exchange froze trading on Gruma until October 30. 33
  35. 35. Table III – Results from the Probit ModelTable III relates corporate governance and other variables to risk management through a probit panel datamodel. Dependent variable is 1 for companies that posted losses in derivatives for the year 2008, and 0 for othercompanies. MON represents lack of monitoring; DIS relates to disclosure; USA shares in the American market;CON ownership concentration; IIN participation of institutional investors; FHP formal hedging policy; TREtrend of major source of risk; FIN recent financial results; REM remuneration incentives for the CFO; and MANmanagement stake. Last two variables follow Tufano (1996). Total number of companies is 346, and the lastcolumn represent only distressed companies, i.e., companies in which derivative losses resulted in bankruptcy,acquisition by another company of major restructuration. Ommited controls are: age of each company, averagefinancial debt ratio and a binary variable for family-owned company. Whole Sample Distressed Companies β σ Β σ H1 No monitoring 0.429* 0.08 0.232 0.07 H2 Disclosure 0.218 0.44 0.561 0.91 H3 Shares in the US market -0.812* 0.35 -0.270 0.67 H4 Concentration -0.054 0.24 -0.343* 0.15 H5 Institutional Investors 0.044 0.42 0.157 0.87 H6 Formal hedging policy -0.558* 0.22 -0.780* 0.30 H7 Trend of source of risk 0.754* 0.26 0.785* 0.32 H8 Recent financial results 0.234 0.62 0.059 0.64 H9 Remuneration 0.861* 0.22 0.338* 0.14 H10 Management stake -0.261 0.60 0.042* 0.09 N obs 346 310 LR X2 97.56 89.35 Prob > X2 0.000 0.001 Log likelih. -239 -208 Pseudo-R2 0.34 0.32 Cross-dep (H0 : R = IN) 0.015 0.018 * variable significant at 5%. 34
  36. 36. Table IV – Marginal Effect After the Probit ModelTable IV is the marginal effect of the probit mode that relates corporate governance and other variables to riskmanagement. Dependent variable is 1 for companies that posted losses in derivatives for the year 2008, and 0for other companies. MON represents lack of monitoring; DIS relates to disclosure; USA shares in the Americanmarket; CON ownership concentration; IIN participation of institutional investors; FHP formal hedging policy;TRE trend of major source of risk; FIN recent financial results; REM remuneration incentives for the CFO; andMAN management stake. Last two variables follow Tufano (1996). Total number of companies is 346, and thelast column represent only distressed companies, i.e., companies in which derivative losses resulted inbankruptcy, acquisition by another company of major restructuration. Ommited controls are: age of eachcompany, average financial debt ratio and a binary variable for family-owned company. Whole Sample Distressed Companies Β σ Β σ H1 No monitoring 0.023* 0.010 0.002 0.010 H2 Disclosure 0.002 0.019 0.006 0.009 H3 Shares in the US market -0.019* 0.008 -0.002 0.007 H4 Concentration -0.004 0.004 -0.015* 0.005 H5 Institutional Investors 0.003 0.002 0.001 0.019 H6 Formal hedging policy -0.012* 0.005 -0.020* 0.007 H7 Trend of source of risk 0.036* 0.012 0.021* 0.003 H8 Recent financial results 0.002 0.003 0.009 0.021 H9 Remuneration 0.037* 0.012 0.016* 0.005 H10 Management stake -0.001 0.001 0.014* 0.007 Mg effect 0.1987 0.2140 N obs 346 310 * variable significant at 5%. 35
  37. 37. | 37 |