Insider trading and corporate valuation: evidence
from changes in the regulatory climate
Abstract
This paper investigates the impact of insider trading on corporate valuation through changes in the insider trading regulatory climate. It is expected that more lenient insider trading laws are detrimental for firm value and that stricter insider laws are beneficial for firm value. To substantiate this claim, two Court rulings that have significantly impacted the regulatory climate are used as natural experiments; United States v. Newman and
Salman v. United States. To establish causality, the event study methodology is used. In order to best capture the
effects of these rulings, specific sub-samples are tailored in which insider trading is more and less likely to occur. The cross-sectional cuts involved are informed by theory and include: the presence of blockholders, quality of governance and measures for information asymmetries. The essay concludes that more lenient insider trading laws are indeed detrimental for firm value and that stricter insider trading laws benefit firm value. This effect is especially strong in the sub-sample of firms where insider trading allegedly occurs relatively often. These results strengthen the believe that insider trading is detrimental for firm value.
Author
Anne-Cor Halma
Supervisor
Torsten Jochem
Master Thesis MSc Finance, Corporate Finance track 1 July 2017
Statement of Originality
This document is written by Anne-Cor Ruben Halma who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table&of&Content!
1.! Introduction and rationale!...!4!
2.! Literature review!...!7!
2.1.! The positive effects of insider trading on corporate valuation!...!7!
2.2.! The detrimental effects of insider trading on corporate valuation!...!9!
2.3.! Content and information flow of inside information!...!11!
3.! Methodology!...!12!
3.1.! United States v. Newman – 10 Dec 2014!...!12!
3.2.! Salman v. United States - 6 Dec 2016!...!14!
3.3.! Hypotheses!...!16!
3.3.! Creating two sub-samples!...!17!
3.3.1.! Blockholders!...!18!
3.3.2.! Research and Development expenses!...!19!
3.3.3.! Bid-ask spread!...!20!
3.3.5.! Quality of governance!...!22!
3.4.! Estimation of effects on corporate valuation!...!23!
3.4.1.! Testing for significance – the cross correlation problem!...!25!
4.! Data!...!26!
5.! Results!...!29!
5.1.! United States v. Newman case – 10 Dec 2014!...!29!
5.2.! Salman v. United States - 6 Dec 2016!...!32!
5.3.! Implications for hypotheses!...!35!
5.4.! Caveats and limitations to the methodology!...!36!
6.! Conclusion!...!37!
1. Introduction and rationale
There are few subjects in economic and legal literature that are as extensively debated as insider trading. The first conviction for insider trading in the United States was in 19091 and
in 1934 the Securities and Exchange Act came into force. The Securities and Exchange Act of 1934 was the first Act that adopted provisions to combat insider trading activities2. However, extensive enforcement of insider trading began only in 1960 with the
jurisprudential classic SEC v. Texas Gulf Sulphur3. Upon writing anything on insider trading, it is essential to adequately describe insider trading. A common misconception is that insider trading per definition is illegal. This is not the case, insiders (for example company directors) are permitted to buy and sell shares. These transactions however must be filed with the Securities and Exchange commission4, and cannot be based upon material undisclosed information. Illegal insider trading, on the other hand, is the illegal use of undisclosed information, and can be done by anyone, including non-insiders. Especially in recent times, insider trading enforcement efforts by the SEC have been unmatched in their scope and impact, resulting in very lengthy prison sentences (Strader, 2015)5. In the United States, insider trading is regarded as one of the worst financial crimes. Individuals can face a prison sentences of up to 20 years, and can be fined up to $5 Million for each violation6. Another important distinction, often overlooked by media coverage, is the difference between insider trading and market manipulation. Market manipulation is the practice of, for example, disclosing false information to the market.
Once we established what insider trading refers to, we can go deeper into the earlier mentioned legal and economic debate about insider trading. Briefly put, there is considerable disagreement about the effects of insider trading and insider trading regulation. Some
academics argue that insider trading yields certain benefits, outweighing the disadvantages. For the purpose of this essay, we will limit ourselves to arguments relating to the effect of illegal insider trading on corporate valuation, not addressing very material issues in the
1 Strong v. Repide, 213 S.Ct. 419 (1909).
2 For example, Section 16, however it must be noted that the law of insider trading is effectively the product of
common-law judicial interpretation or the very broad terms in Section 10(b) and Rule 10b-5 of the Act.
3 SEC v. Texas Gulf Sulphur Company, 401 F.2d 833 (2d Cir. 1968).
4 Under Section 16 of the Securities and Exchange Act of 1934 company insiders (executives, directors,
large-shareholders) are required to make filings about their stock holdings with the SEC. The initial filing is on Form 3, the changes in ownership on Form 4 and any transactions that should have been reported earlier on a Form 4 or were eligible for deferred reporting, must be filed through Form 5.
5 Refering to Raj Rajaratnam, infamous hedge fund manager, who was sentenced 11 years in prison. 6 Section 32 (a) Securities Exchange Act of 1934, as amended by the Sarbanes-Oxley Act of 2002.
insider trading debate like, for example, market efficiency, property rights, fairness and market morality.
The effects of illegal insider trading on corporate valuation are not easy to determine. Firstly, since illegal insider trading by definition is not disclosed, it is very hard to estimate illegal insider trading in a specific stock. If we cannot even adequately measure insider trading, establishing a causal relationship between insider trading and corporate valuation seems like an impossible task. However, there are some developments in insider trading regulation that could help us establish causality.
This essay proposes to use two Court rulings that are known to have significantly impacted the insider trading regulation environment. The mean idea is that these two Court rulings led to a change in the amount of (illegal) insider trading, impacting firm value. This alleged impact can be either negative or positive, based on the academic viewpoints described later on. This impact on firm value (stock price) will be assessed by conducting an event study, with these Court rulings as events. However, testing for impact on firm value across a sample that contains almost every publicly traded US firm, arguably is not the most accurate measure of the effect of this regulation. For example, firms that are very well governed are less likely to be affected by more lenient insider trading regulation because insider trading is less likely to occur in these firms. Therefore, this essay uses cross-sectional cuts to divide the sample in sub-samples. These cross-sectional cuts serve to better filter out the effects of more lenient or stricter insider trading regulation on corporate valuation. The-cross sectional cuts are informed by theory and include; the presence of blockholders, quality of governance and measures for information asymmetries. These cross-sectional cuts (requirements) are
cumulative, holding that, for a firm to be included in a sub-sample, it must fulfill all of requirements.
Furthermore, in event studies where the event day is the same for the sample of firms, event studies are prone to cross-sectional correlation among abnormal returns. This essay uses a relatively new econometric test to account for this cross-sectional correlation among abnormal returns.
There are some important limitations and caveats to the methodology. Firstly, it could very well be that the events are too weak to impact stock prices, resulting in insignificant abnormal returns. Alternatively, the events could be anticipated by the market and
incorporated in the stock price before the events took place. Thirdly, in the event windows used, there could be other material events that influence stock prices. These limitations
should be taken into account when assessing the causal claim between insider trading and corporate valuation.
However, taken the above into account, the methodology used in this essay does have the potential to provide modest evidence on the effect of insider trading on corporate
valuation and consequently contributes to the relatively limited literature on the specific subject of insider trading and corporate valuation.
2. Literature review
The legal and economic literature on insider trading is extensive and has a relatively long standing history. The literature is not limited to addressing the implications for corporate valuation, but also covers the implications for, for example, market efficiency and market morality. In this paper, the predominate focus will be on the implications of insider trading for corporate valuation and it will therefore not cover the implications for market efficiency and liquidity (see, e.g., Seyhun, 1986; Leland, 1992; Bhattacharya and Daouk, 2000). Economists continue to be deeply divided on the implication for corporate valuation. Consequently, two camps have emerged through the years: one camp argues that insider trading is associated with greater firm value, and conversely, the other camp argues that insider trading is detrimental for firm value. In addition, it also remains unclear through what channels insider trading affects corporate valuation.
2.1. The positive effects of insider trading on corporate valuation
In 1966 professor Manne raised considerable controversy by asking the question why insider trading should be illegal. One of his arguments opposing the illegality of insider trading was that insider trading actually could prove to be beneficial for firm value. The essence of his argument with respect to corporate valuation constitutes the notion that total wealth will increase if insiders are encouraged by the possibility of insider trading profits to create value-enhancing projects, increasing stock prices. Manne suggested that an entrepreneur’s
contribution to the firm consists of producing new, valuable, information. The compensation the entrepreneur receives should be connected to his contribution in order to make sure the manager is incentivized to produce more of this valuable information. Means of
compensation like salary or stock options are inferior ways of compensating the manager according to Manne, because it is not possible to ascertain the value of the information produced by the manager in advance (Manne, 1966). In other words, Manne suggested that insider trading is an effective way of compensating agents for innovations. One of the scholars that build forward on Manne’s idea that insider trading could be used as a
compensation scheme was Dye (1984), who presented an appealing theoretical model in this context. In the model presented by Dye, allowing the managers to have some discretion in composing their compensation scheme could be beneficial for firm value. Carlton and Fischel (1983) also build forward on Manne’s ideas; they also believed that means of compensation
like salary and options fail to compensate the agent for innovation adequately. The advantage of insider trading as compensation scheme over other forms of compensation, according to Carlton and Fichel, is that that an agent automatically revises his compensation package, without a costly change of the actual compensation contract. By trading (and profiting) on inside information, the compensation of the agent now accounts for the information he produces, which in turn increases his incentive to develop valuable information beneficial for the firm (Bainbridge, 1986).
In the spirit of Manne and Dye others have also addressed the implications of insider trading for corporate valuation via its effect on corporate insiders’ incentives (i.e., agency costs) and adverse selection. For example, with a principal-agent model in mind, Bebchuk and Fershtman (1993, 1994) show that insider trading might increase corporate value7. Bebchuk and Fershtman (1993) consider a situation in which the final output of a firm and the productivity of managerial effort will depend on whether the firm is in a good or in a bad state. They argue that when the state of the firm is not known, the (managerial) contract cannot be made explicitly contingent thereon; implying that a contract that does not allow for insider trading would lead to insiders facing equivalent compensation schemes in good and bad states. However, under a contract that does allow for insider trading, insiders will acquire shares upon receiving good news and will dispose of shares upon receiving bad news. The result of which will be that insiders face different incentive schemes in good and in bad times. Furthermore, they debate whether this effect is desirable, and conclude that this depends on the marginal productivity of managerial effort in good times compared to the productivity in bad times. The essence of their research constitutes the notion that allowing insider trading may improve managers’ effort and consequently, may increase firm value. Bebchuk and Fershtman (1994) also show that insider trading might cause insiders to choose riskier investment projects, because increased volatility of potential results enables insiders to make higher trading profits if they learn these results in advance of the market. The reason they argue this might be beneficial for firm value, is because generally, insiders are relatively risk-averse, pulling them towards a conservative investment policy. When a conservative investment policy is the default policy, the firm benefits from insiders choosing somewhat riskier projects.
7 Interestingly, in 1990 Bebchuck and Fershtman published an article where they took another stance. In that
paper they argued that insider trading may increase managers’ incentive to waste corporate value by making decisions that maximize their trading benefits, instead of the maximizing shareholder value.
Noe (1997), in a theoretical setting, also builds forward on the idea that insider trading profits might be a substitute for effort inducing compensation contracts. He does not quite agree with Bebchuk and Fershtman that insider trading increases insiders’ effort. However, he argues that permitting insiders to trade on inside information is simply the cheaper compensation mechanism, because, simply put, the conventional (effort assuring)
compensation mechanisms may require payments to the manager in excess of reservation levels (Noe, 1997). He also comprehensibly explains that permitting insiders to trade on inside information would lead firms to pay lower salaries, he refers to this as the substitution effect. Consequently, shareholders have an incentive to permit insider trading, since it reduces managerial rents. It should be noted that these potential benefits of insider trading only apply to insider trading done by insiders. Insider trading by persons that do not have any influence on corporate policy (misappropriators) is not beneficial for firm value according to most of the aforementioned academics.
2.2. The detrimental effects of insider trading on corporate valuation
Then there are those academics that argue that insider trading is detrimental for firm value. Manove (1989) argues (in a theoretical setting) that insider traders that act on private information benefit from increased returns at the expense of other shareholders. As a result, Manove argues, insider trading tends to discourage corporate investments and reduces the efficiency of managerial decision-making. As mentioned earlier Bebchuk and Fershtman (1990) argue that insider trading may increase managers’ incentive to waste corporate value by making decisions that maximize their trading benefits, instead of the maximizing
shareholder value. Bebchuk and Jolls (1999), examine the effect of insider trading and other mechanisms of managerial value diversion on shareholder wealth. In their paper they
explicitly question the validity of regarding insider trading as an alternative form of
compensation. They try to answer this question in a theoretical principal-agent model. Their main point is that the view that insider trading can be seen as an alternative form of
compensation, overlooks the cost of such behaviour. Since there are many types of compensation that can provide managers with incentives to enhance shareholder value, replacing such compensation would reduce these incentives.
Several studies tried to address the implications of ownership structures and insider trading. There is an obvious trade-off between ownership structure and insider trading: on the one hand, blockholders are able to monitor the firm better, in theory reducing insider trading,
on the other hand large shareholders want to be compensated for their monitoring efforts and their relatively undiversified portfolios. Since these blockholders generally have better access to information, insider trading could be a very profitable way of getting compensated
(Demsetz, 1986; Bhide, 1993). Prominent scholars who study this connection between ownership dispersion, insider trading and corporate valuation include Schleifer and Vishny (1986) and Bhide (1994). These scholars tend to see the matter as a trade-off. Other scholars seem to have developed a clearer position on the matter. For example; Burkart and Panunzi (2001), La Porta, Lopez-de-Manes, Andrei Schleifer, and Visney (1999) provide evidence that large shareholders are likely to use their influence and additional knowledge to their own benefit (for example through insider trading), at the expense of outside shareholders.
Furthermore, Maug (2002) finds that, in companies where large shareholders actively engage in monitoring, managers have an incentive to give early warnings about developments to dominant shareholders. He considers this information as a bribe to persuade dominant shareholders to sell their stock and stop their active monitoring. This is because large
shareholders would rather trade on inside information than engage in the costly monitoring of the firm. Consequently, more (or legalized) insider trading would reduce the effectiveness of corporate governance and decrease firm value. A recent paper by Holderness and Edmans (2016) provides a survey of theory and evidence with respect to large shareholders and conclude that there is still considerable uncertainty on the effects of ownership concentration. It must be noted that the papers mentioned above focus predominantly on the the relationship between firm value and ownership concentration, and not necessarily on the relationship between insider trading, ownership concentration, and firm value. However, insider trading can be regarded as a channel through which ownership concentration can affect firm value.
Beny (2001, 2007) does explicitly investigate the corporate valuation implications of insider trading legislation and enforcement at the firm level. She presents an agency model of corporate valuation and insider trading similar to the earlier mentioned principal-agent model. She tries to value insider trading laws by using i.e. Tobin’s Q and controlling for the
incidence of enforcement and firm-level sales growth. She finds that stricter insider trading laws and enforcement are associated with higher firm values, in the sub-sample of firms in which control and ownership are separated. This is in line with the (theoretical) idea that the prohibition of insider trading and the enforcement can mitigate agency costs (Maug, 2002). Beny (2001, 2007) builds forward on Baharya and Daouk (2000), who try to establish a causal relationship between insider trading regulation and the cost of capital. They find that insider trading leads to a higher cost of capital, resulting in lower firm value. Lastly, Masson
and Madhavan (1991) also provide empirical confirmation that (legal) insider trading lowers firm value, but that greater executive stock ownership may raise firm value.
The connection between insider trading and market liquidity is also a factor to
consider, since liquidity is believed to affect corporate value (Amihud and Mendelson, 1986; Fang and Noe, 2009). The latter authors for example show that liquid stocks have better performance measured by market-to-book ratio. If liquidity is beneficial for stock value, the question remains what the affect of insider trading on liquidity is. Unfortunately, there are relatively few empirical studies on the issue of market liquidity and insider trading. Kabir and Vermaelen (1996) examined the effect of the implementation of certain insider trading
restrictions on the Amsterdam Stock Exchange. They find that market liquidity decreased after the introduction of the trading restriction. Cornell and Sirri (1992) report that insider trading does not reduce market liquidity because of the increase in uninformed trading volume. Chakravarty and McConnel (1999) support these findings, using the (in)famous Boesky insider trading case8 as an natural experiment. The limited empirical research seems
to suggest that insider trading does not significantly affect liquidity. As we will see later, it may however be the case that insider trading does occur more in relatively illiquid stock than in liquid stock.
Furthermore, one of the things that become apparent taking the above into account, is that the research on the relationship between insider trading and corporate valuation is relatively out of date. This is likely the result of the absolute consensus among Lawmakers, Policymakers and the public that insider trading is wrong and, consequently, that the need for strict insider trading laws is self-evident. There have however been some interesting recent papers on insider trading (paragraph 2.3), but these papers do not focus on the relationship between insider trading and corporate valuation.
To say the least, as the division of the two camps in the economic literature suggests, the debate over the effects of insider trading on corporate valuation and simultaneously the debate on whether or not and to what extent insider trading should be regulated is indecisive.
2.3. Content and information flow of inside information
In order to better understand how illegal insider trading works, it is useful to know the information flow of inside information and the actual content of this information. Using a
8 One of the biggest and most infamous insider trading case in history. Ivan Boeksy eventually received a prison
sentence of 3.5 years and a US$100 Million fine, after amassing a fortune of more than US200 Million through, amongst other things, insider trading.
hand-picked dataset, Ahern (2017) provides an unique analysis of the social relationships that underlie illegal insider trading. He finds that tips from insider usually originate from
executives and reach buy-side investors that act upon them after three links in the network Furthermore, he finds that inside information flows through close social-proximity in the sphere of family, friends and even neighbours. As to the content of this information, a recent paper by Cziraki, Lyandres and Michaely (2017) shows that inside information used to trade with, usually contains information about future changes in fundamentals related to corporate events.
3. Methodology
In order to answer the question what the effect of insider trading is on firm value we will use two Court rulings that are very likely to have impacted the insider trading regulation climate. The predicted returns that are needed to compute the abnormal returns around these Court rulings will be estimated by the Market Model and the Carhart Four-Factor-Model. We will first proceed making sure that these Court rulings are suitable to base our event study on. After establishing this, we address the criteria for creating the sub-samples. Lastly, we will continue to address the specifications of this event study.
3.1. United States v. Newman9 – 10 Dec 2014
On 10 Dec 2014 a panel made up of three judges of the Court of Appeals vacated the insider trading convictions of hedge funds managers Chiasson and Newman. The District Court had to drop all of the charges against Chaisson and Newman. This Court ruling was very highly anticipated by both legal scholars and practitioners (Anello and Albert, 201310). Very briefly put, the issue the U.S. Court of appeals (Second Circuit) had to resolve was; if, whether to be found guilty of insider trading, the tippee must know that the insider who disclosed the inside information received personal benefit for doing so11.
In order to understand why United States v. Newman (and Salman v. United States in the next paragraph) had such a significant impact on the insider trading regulatory climate, we must first consider the basics. There are two basic ‘theories’ of insider trading. The first
9 United States v. Newman, 773 F.3d 438 (2d Cir. 2014)
theory refers to the classic type of insider trading; a corporate insider (executive, manager) trades on inside information, or discloses (tips) inside information to an outsider (tippee) who trades on this information. The person receiving the inside information (the tippee) may be held liable even in the absence of a fiduciary duty, if he or she knew or should have known that the person proving the information (the tipper) had breached his or her fiduciary duty and if the tipper received personal benefit in return for disclosing the information to the tippee12. In Dirks v. SEC the modern requirements for tippee and tipper liability were established under this classical framework.
Moreover, Courts have expanded the concept of insider trading under the misappropriation theory (the second theory), holding that even temporary insiders (accountants, bankers lawyers) are guilty of insider trading insofar they trade on the information entrusted to them or if they illegitimately disclose this information to other outsiders13.
The Newman case mostly relates to the first theory of insider trading. In the Newman case, hedge fund managers Chiasson and Newman, traded on financial data from various technology companies before this information was made public. The defendants were alleged to have received this inside information from insiders in these technology companies. The (Southern) District Court of New York effectively used the reasoning of the misappropriation (second) theory14 on the Newman case, holding that the tippee only has to know that the tipper broke his fiduciary duty, and not that the tipper received personal benefit in return for doing so. The U.S. Court of appeal (our event) however, sided with Chiasson and Newman; they had to know that the tipper received personal benefits for disclosing the insider
information. One could argue that, after the the Newman ruling, trading on undisclosed information becomes fraudulent (for tippees) only when the insider discloses information for personal gain and the tippee is aware of this. It must be noted that, before the Newman ruling, there was no certainty about this matter.
While this seems rather specific, it is common understanding among legal
practitioners and scholars that this Court ruling delivered a significant blow to the DOJ’s insider trading prosecution campaign led by the US Attorney’s Office and will impact the prosecution of insider trading cases significantly (Stockman, 2014). In the aftermath of the Court ruling two camps emerged: prosecutors who complained that the ruling will tie their
12 Dirks v SEC, 463 US 646, 659-660 (1983). 13 United States v O’Hagan, 521 US 642 (1997).
hands in pursuing Wall Street crime, and criminal defence Attorneys who believed that this Court ruling finally put a stop to what they see as an illegitimate overreach of the United States government15 (Epstein, 2015). Under the assumption that less strict insider trading laws lead to more insider trading, this event, at least in terms of content, is likely to provide a useful event to conduct an event study on. Apart from content, the outcome of the Court ruling needs to be unknown in advance. In other words, the event should not be anticipated by the market. Is it not likely that the market expected this ruling, the probability of the Court ruling in favour of the defendants was not skewed towards their favour and the Court had not issued any pre-announcement of a forthcoming ruling (Anello and Albert, 2013). Moreover, a wide variety of articles published in law journals on the Court ruling also strengthens the believe that the outcome of this Court ruling was not expected in advance (Strader, 2014; Stockman, 2014; Epstein, 2015).
Furthermore, a (empirical) robustness check was carried out. The robustness check includes the abnormal returns of day T-1 for both sub samples of firms. The T-test performed on the abnormal returns of 9 Dec 2014 yields no significant results, implying that, if the Court ruling had any effect, at least in the day before the announcement, there was no information leakage. It should be noted that this check is relatively weak, since anticipation and/or leakage is likely to be incorporated in the stock price before T-1.
3.2. Salman v. United States16 - 6 Dec 2016
On 6 Dec 2016 the Supreme Court ruled – as opposed to the Newman case - in favour of prosecutors in a major insider trading case. The matter and the key issue was not much
different from the Newman case. This Supreme Court ruling was at least as highly anticipated as the Newman case (Becker, 2016; Nagy 2016). The case was about the ‘gift’ of confidential information from insiders to relatives. The tipper in the Salman case was a banker employed at Citigroup who disclosed inside information about M&A transactions to his brother. The tipper knew that his brother was using this confidential inside information to his advantage by trading on the securities markets. The brother also disclosed this inside information to
Salman, also a family member, who also traded on the inside information knowing that this
15 Especially referring to a Southern District decision on the required state of mind for tippee liability; Whitman
v. United States 574 US 904 (S.D.N.Y. 2014).
inside information was confidential and that the information originated from Citigroup17. In Salman’s appeal to the Ninth Circuit, he argued that his conviction for insider trading was not legally substantiated, since the banker did not receive any personal benefit for disclosing the information18. The Ninth Circuit did not side with Salman, and ruled that the tipper ‘gifting’ inside information to a family friend or relative is sufficient for tippee-liability. Effectively, the Ninth Circuit breaks with the ruling in Newman, where the Court ruled that the tipper who disclosed information to a tippee is required to have received personal benefit for doing so. Consequently, the Supreme Court had to resolve this jurisprudential split in Salman v. United States.
In the Salman Supreme Court ruling, the Court upheld the interpretation of the Ninth Cicruit, ruling that the Dirks v. SEC standard for insider trading (as explained in the previous paragraph) relating to tippees, still holds19. They resolve the jurisprudential split by
answering the question of what constitutes a ‘personal benefit’ to the insider. The Court argues that giving confidential information to a friend or relative, constitutes personal benefit. Therefore, in order to be liable as a tippee, the tipper does not need to have received tangible personal benefit; the benefit for the tipper is in the ‘gift’ of information to his friends or relatives. In other words, the Court assumes that giving a tip to a friend or a relative is beneficial for the insider20.
While this ruling also seems very specific, it will result in a significant increase in tippee/tipper enforcement (Becker, 2016; Nagy 2016). After Newman, the prosecution in the Southern District of New York chilled. The Supreme Court’s decision in Salman reopens the doors closed by Newman, making it a lot easier to prosecute tippee liability cases. As
research by Ahern (2017) suggests, insider trading by tippees accounts for a substantial part of all the illegal insider trading. Consequently, it is not hard to see why the Salman case significantly impacts the insider trading climate. At the very core, Salman made insider trading easier to enforce and therefore might have a reverse effect on corporate valuation compared to the Newman case, since it is assumed that easier enforcement of insider trading laws deters insider trading. All of the above indicates that Salman, at least in terms of content, is a good candidate for an event study.
While the content of the Salman case makes it a good candidate for an event study, there are some concerns with respect to the degree of ‘unexpectedness’. In an article
17 Id. at 1080-1088. 18 Id. at 1090-1995.
19 Id. at 664 (Dirks v. Sec quote). 20 Id. at 1092
published before the actual Court ruling, the author (who is a prominent legal scholar) argues that he is relatively certain that the Supreme Court will side with the Ninth District (Nagy, 2016). However, not all legal scholars seem agree on this (Adams, 2016). It remains very hard to determine to what degree a Court ruling is unexpected or surprising. However, we can cautiously conclude that the Salman case, while at least as highly anticipated, is not as
surprising as the Newman case. Making it a less ideal (but still a feasible) candidate for an event study.
Furthermore, a (empirical) robustness check was carried out. The robustness check includes a test for early portfolio price movements in the day(s) prior to the Court ruling. More specifically, we test for significant abnormal returns across the whole sample for Monday 5 Dec 2016. The T-test statistic obtained from the test is insignificant, implying that at least in the day before the Court ruling, there was no information leakage.
3.3. Hypotheses
As mentioned before, the main goal of this essay is to establish a causal relationship between insider trading and corporate valuation by using two Court rulings as natural experiments. Under the assumption that more lenient or stricter insider trading laws and enforcement lead to more respectively less insider trading; the following empirical hypotheses apply;
Hypothesis 1: More lenient insider trading laws and enforcement [more insider trading]
decreases firm value, and stricter insider trading laws [less insider trading] increases firm value.
The first hypothesis applies to master sample (almost all US publicly traded companies) of firms. These are US publicly traded firms, not funds or other inactively traded firms. Taking all of the literature in chapter 2 into account, it seems like there is more empirical evidence that insider trading, on average, is detrimental for firm value. Consequently, if the events are ‘strong enough’ to influence the stock market all a whole, it is expected that this hypothesis holds. It should however be noted that just by looking at the stock price on these specific event days – without specifically tailored sub-samples – cannot lead to a very robust conclusion.
Hypothesis 2: In firms where insider trading is more likely to occur, more lenient insider
trading laws and enforcement have a stronger negative effect on firm value compared to firms where insider trading is less likely to occur.
Hypothesis 3: In firms where insider trading is more likely to occur, stricter insider trading
laws and enforcement have a stronger positive effect on firm value compared to firms where insider trading is less likely to occur.
The second and third hypotheses take into account the sub-samples that are created via the cross-sectional cuts in the next paragraph. The second hypothesis applies to the Newman case and the third to the Salman case. The cross-sectional cuts that the sub-samples are based on are informed by theory and should be able to better capture the effect (establish causality) of the rulings.
Hypothesis 4 (null): The Newman and the Salman case, however impactful on the regulatory
insider trading climate, have not significantly impacted the US stock market.
If the event studies relating to the two rulings do not have significant abnormal returns for both the whole sample and the sub-samples, we can conclude that these rulings, however impactful they may have been in a legal sense, did not significantly impact the stock market environment. This might be the case if the events are not strong enough to affect the stock market or if the events were anticipated by the market.
3.3. Creating two sub-samples
In order to identify firms that are likely to be affected by these Court rulings and best capture the effects of these rulings, samples must be created. The goal is to end up with two sub-samples; one sample contains firms that are unlikely to have their stock price be affected by the change in insider trading regulation, and the other sample must contain firms that are likely to be affected by the change in insider trading regulation. The basic underlying reasoning is as follows: more lenient insider trading laws lead to more insider trading and more insider trading is either beneficial or detrimental for firm value. The sub-sample that contains firms where insider trading is more common, is expected to be influenced the most, having the biggest abnormal returns.
In order to use the methodology above, we need to know the determinants of insider trading in order to determine in what sub-sample a certain firm belongs. In other words, the cross-sectional cuts need to be related to the determinants of insider trading, inspired by either theory or empirics. The sub-sample where the effects are expected to be stronger includes only firms that make all of the cross-sectional cuts (the requirements for the sub-samples are cumulative), resulting in relatively small sub-sub-samples. The cumulative requirements serve to best capture the effects of the events.
One very important determinant of insider trading is information asymmetry. Generally speaking, in firms with great information asymmetry, insiders will be in a better position to benefit from insider trading. In determining the cross-sectional cuts, information asymmetry will be a recurrent theme. More specifically, the likelihood of (rational) informed insider trading happening can be thought to be driven by two components of information asymmetry; the precision of the insider’s informational advantage and the uncertainty about the fundamental value of a stock. Consequently, the above-mentioned cross sectional cuts will party be determinant by (empirically proven) proxies for information asymmetry.
3.3.1. Blockholders
Academics like La Porta, Lopez-de-Manes, Andrei Schleifer, and Visney (1999) and more specifically Maug (2002) have shown that large shareholders might use their position and influence to their advantage. The rational behind this is that large shareholders want to be compensated for their monitoring efforts and lack of diversification (Bhide, 1994). One way of benefiting from their position is through insider trading. Large shareholders, through their controlling position and corresponding informational advantage, have more access to insider information than other shareholders and could use this information to their advantage. This can be done in many ways. We have seen that insider trading commonly occurs through tipping. Large shareholders, having greater access to insider information, could leak information to friends, relatives, or even investors (for example hedge funds).
On the other hand concentrated ownership might improve monitoring and therefore increase firm value through better governance, since large shareholders can be regarded as more incentivized and better able to monitor the firm (Demsetz, 1986; Schleifer and Vishny, 1986). However, taking into account the empirical research above (and chapter 2) it seems like the empirical evidence showing that large shareholders are likely to use their position to their advantage is more up-to-date and generally more convincing, although the exact
implication for corporate valuation remain uncertain (Holderness and Edmans, 2016). Therefore, it seems right so assume that if insider trading laws became more lenient, the negative effects on corporate valuation are stronger in the sub-sample where large
shareholders are present, since they can now more easily compensate themselves through insider trading at the expense of other shareholders (also in line with Beny, 2001, 2007). Since the presence of blockholders is very common (96%) among the average US firm (Holderness 2016; Edmans, 2009), it makes sense to use a different, limited, definition of a blockholder. If such a limited definition was not used, the cross-sectional cut would lead to two samples that are very different in size. The cross-sectional cut used here will be based upon the presence of institutional ownership of at least 5% around the event date, therefore, executive ownership (inside ownership) will not be included here. This will result in an almost symmetric split of the master sample, since approximately 51% of the firms have a institutional shareholders that holds at least 5% of the company’s stock.
3.3.2. Research and Development expenses
Aboody and Lev (2000) show that in firms that invest heavily in research and development, insider trading gains are substantially larger than in firms that do not invest heavily in research and development. The ratio behind this finding is straightforward; the famous SEC
vs. Texas Gulf Sulphur case in 1960 already illustrates this. In this ruling, the SEC brought a
suit against Texas Gulf Sulphur Co after insiders in this company bought shares of Texas Gulf while they secretly had positive information regarding mining activities carried out by the company. Mining companies like Texas Gulf are R&D-heavy, partly because they have to conduct surveys regarding mining exploration. In this case, a specific area was considered promising by the survey. The analysis showed that the area was especially rich in minerals. This was the non-public material information the insiders traded on. The information asymmetry in the Texas Gulf case follows directly from an R&D expense (the survey). The same can be argued for, for example, companies in the pharmacy sector. The development of a promising medicine can be regarded as non-public information known with insiders, but unknown with outsiders. Also, the accounting rules with respect to R&D are much different than the accounting rules for other financial assets. Most financial assets require quarterly or annual reporting in financial statements. As opposed to these financial assets R&D does not require periodic updating on the change in productivity and R&D value. Even major R&D events like a drug successfully passing clinical trials, are not periodically reporting to
investors (Aboody and Lev, 2000). Consequently, the firm-specific amount spend on R&D expenses seems a very good determinant for the likelihood of insider trading occurring, since R&D expenses substantially increase information asymmetry. In order for a firm to be placed in the sub-sample where insider trading is more likely to occur, a firm must be in the upper half of firms with respect to the amount of R&D expenditure.
3.3.3. Bid-ask spread
In financial markets theory and empirics, the bid-ask spread is thought to be the product of information asymmetry and ‘informed trading’ (which in itself is also a product of
information asymmetry). A large bid-ask spread reflect either a large portion of informed traders, a lot of information asymmetry or a combination of both. In theory Information asymmetry leads to more informed traders, including insider traders (Aboody and Lev, 2000). To show this the ‘Glosten-Milgrom model’ is used, as thought of by Glosten and Milgrom (1985).
Suppose that there are traders that trade on inside information or have another information advantage over the market, the so called ‘informed traders’. These informed traders place an order at time t with probability π. An informed trader is defined as someone with advance knowledge (inside information) of the fundamental value of a certain stock v. With the probability 1- π the order comes from an uninformed trader (a regular investor), who places either a buy or a sell order with probability 0.5 each. VH indicates a high fundamental value of the stock and VL a low fundamental value and θt and 1- θt are the probabilities that
the dealers assign to the occurrence of a VH respectively VL. The dealer sets the bid and ask prices according to what the dealer thinks the fundamental value of the stock is:
!" = $ %"&'+ 1 − % " &+
The dealers’ view on value of the security are adjusted according to the order flow they receive and information from other channels. With probability π the dealer receives an informed order, giving him information about the fundamental price of the stock.
Consequently, the dealer will adjust the bid/ask price according to this new information:
,"= $ !"- = .(&|1
"23, 5" = 1) 7" = $ !"8 = .(&|1
Where !"- is the dealers’ estimate of the security’s fundamental value if they receive a buy order (5"= 1) and !"8 is the dealers’ estimate of the security’s fundamental value upon receiving a sell order ( 5"= −1). Buy orders by informed traders are good news; it signals that the true value of the security is high. The expected profits of trading with an informed trader are:
1
2:*(,38&')
However, dealers are competitive resulting in expected returns of zero:
,3 = !<$+ $: &'8!<$ = $ !<$+ :
2(&'8&+)
The same goes for the bid price since it is symmetric (73 = $ !<$+=>(&'8&+)). ?s can be seen from the equation above, the ask price now includes a mark-up over the initial estimate of the fundamental value of the security !<$. Consequently, in this widely used (simplified) Glosten-Milgrom model the bid-ask spread, being the difference between ,"and 7"is:
@"= ,3873 = $:$(&'8&+)
The bid-ask spread can be seen as compensation required by dealers to cover the loss they incur when trading with persons trading on insider information (informed traders). The spread is a function of the amount of informed traders : and the volatility of the security’s value. The spread increases with the proportion of informed traders and the volatility of the security’s value. Therefore, we can assume that a large bid-ask spread in a specific stock can be an indication that (illegal) insider trading is present, either because there are informed traders active (and among them traders that trade on inside information) or because the stock is very volatile, which increases the likelihood of insiders trading on inside information since their informational advantage now potentially yields more proceedings (Kyle 1985). The firms that have the highest (upper half) mean bid-ask spread will be placed in the sub-sample of firms where insider trading is more likely to occur. The lower half will be placed in the other sub-sample.
3.3.5. Quality of governance
Theory dictates that well-governed firms suffer less from typical agency problems like insider trading through i.e. better monitoring. Therefore, creating sub-samples based on governance characteristics would probably yield differences in abnormal returns around our events. There are no specific corporate governance characteristics that are closely linked to the specifics of the Court rulings, so the most sensible thing to do would be to create a corporate governance index as a proxy for the balance of power between shareholders and managers to generate two different samples.
In order to create such an index, we have several options. We could choose to recreate the Gompers, Ishii and Metrick (GMI)-index around the time of our events, which consists of 24 governance rules to proxy for manager/shareholder imbalance. Another option is to use the takeover provisions identified by Bebchuk and Ferrel (2009) as most relevant. The latter is chosen mainly because its (relatively up-to-date) empirical substantiation in their paper and its easiness to compute. Like GMI they construct a corporate governance index (the Entrenchment index, or E-index), this E-index is based on six takeover provisions Bebchuk and Ferral argue are the most influential in corporate governance. They rate firms (0-6) depending on how many takeover provisions a specific firm has. These takeover provisions include four provisions that prevent a majority of shareholders from having their way and two takeover ‘readiness’ provisions that boards put in place to be ready for a hostile takeover. The first type of provisions includes: staggered boards, supermajority requirements, requirements for charter amendments, and requirements for by law amendments). The second type of provisions include golden parachutes and poison pills. These provisions are readily available in the Institutional Shareholders Services (ISS) (formerly RiskMetrics) database, on a “yes” or “no” basis. According to theory, the amount of provisions a firms has, serves as a proxy for governance (Bebchuk and Ferell, 2009). The higher a firm score on this index, the more power managers have compared to shareholders. In firms that score high, the effects of the change in insider trading regulation allegedly are the strongest, since in firms where the balance of power between shareholders and managers is more skewed towards the managers, principal-agent problems like insider trading are more likely to occur (for example Dai and Kang, 2016). Consequently, ranking 3-6 is a requirement for firms to enter the sub-sample where the effects are allegedly the strongest, and firms that rank 0-2 is a requirement to enter the sub-sample that is expected to be relatively unaffected. Moreover, these cuts are also chosen because firms that rank 3-6 constitute around 50% of all the firms in the master
sample and firms that rank 1-2 constitute the other half. This is important, because else the sub-samples would diverge to much in terms of observations.
Lastly, it should be noted that such an index does not precisely reflect the relative impacts of each of the the different provisions.
3.4. Estimation of effects on corporate valuation
The effects on corporate valuation are measured by comparing the average daily abnormal returns of a portfolio of firms that should be more effected by the Court ruling to firms that should not be effected as much. In order to calculate the abnormal returns we first need to estimate the expected return for the event date. We do this by using the Market Model:
AB" = $ CB +$DBAE" + FB"
Where CB and DB are unknown parameters to be estimated for each security I, AE" is the market return and AB" is the firm specific return. The Market Model effectively considers the firm’s CAPM risk by multiplying AE" with the firm specific DB. We estimate the model parameters by OLS regression, which in turn are used for the calculation of abnormal returns in the event window. The abnormal return is defined as the (estimated) error term of the model:
?AB" = $ FB"
Multiple event windows will be used: [0], [0,1], [0,2], These event windows are short term, but it is expected that, if there is a significant effect, this will show relatively quick in the stock prices. The estimation window will be based on 150 trading days, ending 30 days before the event. After obtaining the abnormal return in the event windows, we cumulate the abnormal returns into the cumulative abnormal returns (CAR). The cumulative average of the CAR’s over a certain period of trading days in the event window will be the CAAR. Buy and hold returns could also be used, but since this is regarded as a short-run event study, it is argued by Barber and Lyon (1997) that the usage of CAR instead of BHAR is likely to be the better choice. The last step would be to test (in the main sample and in the sub-samples, for both events) whether the abnormal returns are significantly different from zero on a statistical basis where the null (our forth) hypothesis is:
G< = . H?AB = $0
If this null hypothesis would prove to be true (no significant abnormal returns) for both events and for both the master sample and the sub-samples, our forth hypothesis would hold. Holding that the Newman and the Salman case, however impactful on the regulatory insider trading climate, have not significantly impacted the US stock market. The first hypothesis would prove to be true if there are significant negative abnormal returns in the master sample on or around the Newman ruling, and significant positive abnormal returns on or around the Salman ruling. The second hypothesis would prove to be true if there are relatively stronger significant negative abnormal returns around the Newman ruling for firms where insider trading allegedly occurs more. Lastly, the third hypothesis would prove to be true if there are relatively stronger significant positive abnormal returns around the Salman ruling for firms where insider trading allegedly occurs more.
Although the Market Model is used in +- 80% of the event studies and is arguably the better choice for event studies with relatively short return windows (Fama, 1998; Holler, 2014), another expected return model will be used to check the results of the Market Model. According to theory, using another expected return model, should not drastically change the outcome (Brown and Warner 1980, 1985; Ahern, 2009). The Carhart four-factor model will be the expected return model used to check the Market Model. This model essentially expends the Fama-French Three Factor model by adding an additional factor, namely momentum. This model follows the following specification:
ABJ8RLJ = $ αN+$βN(RPJ8RLJ) + sNSMBJ+ hNHMLJ$+ uNUMDJ$εNJ
Where SMBJ is “Small Minus Big”, which captures the excess return of small over big stocks (measured by Markett Cap), HMLJ$ is “High Minus Low”, which captures the excess return of stocks with high market-to-book ratio over stocks with low market-to-book ratio and UMDJ is “Up Minus Down” (the Carhart momentum factor; Carhart, 1997)21. Like in the Market Model, the abnormal returns are defined as the (estimated) error term of the model (ARNJ = $εNJ).
21 Data from Kenneth French’s website, which covers all NYSE, AMEX and NASDAQ firms and easily pairs
Lastly, it should be noted that, although the estimated error term of the model in theory does have to potential to capture the effects of the Court rulings, the abnormal returns found could very well be the result of other market forces. Running the tests for the sub-samples serves to more precisely capture the effects of the Court rulings
3.4.1. Testing for significance – the cross correlation problem
When the event day is the same for the sample of firms, event studies are prone to cross-sectional correlation among abnormal returns. More specifically the event-date clustered analysis may introduce a downward bias in the standard deviation and consequently lead to overstating the T-statistic, which leads to over-rejection of the null hypothesis (Collins and Dent 1984; Kothari and Warner, 2007). To address this problem, we can deploy several econometric strategies. The traditional approach (Jaffe, 1984) would be to account for this bias be using the so called portfolio method. In this method the firm’s returns are aggregated into an equally weighted portfolio and consequently, the abnormal returns of this portfolio is estimated. Alternatively, the multivariate regression method with generalized least squares can be used (Kothari and Warner, 2007). However, this requires an adequate estimate of the variance-covariance matrix, which is impossible in finite samples. Also, because every covariance plays a crucial role in the generalized least squares through matrix inversion, estimation errors in individual covariance introduce more sampling error than they account for (Kolari and Pynnönon, 2010). Furthermore, Chandra and Balachandran (1990) argue that the generalized least square method is highly sensitive to model misspecification (Kolari and Pynnönon, 2010). Lastly, some non-parametric tests have proven to be useful. For example, the Rank-test by Corrado (1989) and Corrado and Zivney (1992) and the general sign test by Cowan (1992) They are useful because of their non-parametric trait; they do not assume that the data has a particular probability distribution. From these non-parametric tests, the
cumulated ranks test was inferred (Cambell and Wasley, 1993, 1996).
However, Kolari and Pynnönon (2010) show that the most powerful way of dealing with the cross-correlation problem would be to use an adjusted T-test. Regardless of the cross-correlation it would be wise to account for event-induced volatility in testing the event effect. The starting point would therefore be the scaled abnormal returns model by Boehmer, Musumeci and Poulsen (BMP):
T^= $ A n
s
where s is the cross sectional standard deviation of the event-day scaled abnormal returns and$A is the mean abnormal return. Furthermore, to account for cross-correlation we use the follow test:
T`^ = $ T^
1 − r 1 + (n − 1)r$
where 3-(c23)b32b is the correlation factor and r is the average of the sample
cross-correlations of the estimation-period residuals. It should be noted that the initial BMP test and the adjusted BMP test yield the same results if the average return cross-correlation is zero (Kothari and Pynnönon, 2010).
4. Data
Stock data is obtained from the CRSP-database and balance sheet data and general
information about, for example, R&D expenses, are obtained from Compustat. CRSP and Compustat are matched by CUSIP22. Data needed for the cross-sectional cut based on
institutional ownership is obtained from the SEC Edgar database (13-f Filings) and linked by the CIK firm identifier. The total shares outstanding around that time are obtained from the CRSP database, so that the percentage of institutional ownership can be computed. Data on the quality of governance (ratings) originates from the Institutional Shareholder Services (Formerly RiskMetrics) database and can linked with CRSP/Compustat by Npermno. From all of the provisions readily available in the database, the six provisions identified by Bebchuk and Ferell (2009) as most relevant, are downloaded for the timespan around our events. As a result, the the E-index is effectively recreated for the time around our events. Furthermore, data needed on the bid-ask spread is also obtained from the CRSP-database. Lastly, data needed to construct the abnormal returns with the Carhart four-factor model as
benchmark model is provided by French’s website23. The master sample for the Newman event contains (3890) unique firms that are traded in the US and the master sample for the Salman event contains (4090) unique firms. The difference between the amount of unique firms on both events stems from new listings in the period between the Newman and the Salman events. Firms that have a market cap below 10Mill, multiple share listings, inactively traded firms, investment funds, unit trusts and other listings that only serve as holding
company are excluded. Please note that in the table below, only the sample statistics for the Newman event are presented. The sample statistics for the Salman event are not reported since they do not differ materially.
Table 1: Sample Statistics Master sample Newman event
In panel A the summary statistics for the master sample are presented (N=3890). Mkt cap is in millions. R&D is denoted as percentage of revenue. Panel B refers to the institutional ownership data. Mean ownership refers to the mean percentage of ownership by institutional investors. Panel C refers to the
recreated E-index around the event date.
Panel A: Firm Characteristics
(1) (2) (3) (4) (5)
Characteristics Mean Median SD Min Max
Mkt Cap Bid-Ask R&D expenses 1,709.77 0.0971 4.75% 153.91 0.0237 1.93% 4,984 0.7087 5.98% 10.09 0.0033 0% 717000 469.1 4399.89%
Panel B: Institutional ownership Panel C: Entrenchment
As to the sub-samples and the corresponding cross-sectional cuts it must be noted that the cumulative amount of firms in the sub-samples does not equal to master sample. Firstly, this is because the sub-samples are created using the determinants of insider trading as cumulative criteria. This means that in order for a firm to be included in the sub-sample where insider trading is more likely to occur, it must fulfil all of the requirements as set forth
23There are some discrepancies in the observations used in the Market Model versus observations used in the
Carhart model. This is because the data on Kenneth French’s website does not 100% correspond with our dataset. Mean ownership Median ownership SD ownership Firms >5% 13.1% 6.5% 14.1% 51% 0 1 2 3 4 5 6 11.1% 12.7% 24.7% 24.9% 18.9% 4.5% 1.0% 100%
in paragraph 3.3). Secondly, this is because the cross-sectional cuts are based on data that are not available for all of the firms. For example, not every firm covered in CRSP/Compustat is covered in the ISS-database, not every firm adequately reported their R&D expenses and there is some error margin in merging the CRSP/Compustat with the 13-f flings. The sub-sample where the change in insider trading regulation and enforcement should have a
stronger effect (Sample I) contains (99) unique firms for the Newman event (and 105 for the Salman event) and the sub-sample that should be relatively unaffected (Sample II) contains (150) unique firms for the Newman event (and 137 for the Salman event). It is also worth noting that the firms in Sample I are significantly bigger than firms in Sample II. This is mainly because of the requirement based on R&D. Firms that report little or even zero R&D-expenses are usually significantly smaller than firms that have high R&D R&D-expenses.
5. Results
5.1. United States v. Newman case – 10 Dec 2014
First, the daily returns of the firm across the master sample will be regressed on the market return. As described above, the regression estimates will be used to generate out-of-sample predictions for the event window around the event. Consequently, the difference between the predicted return and the actual returns are the abnormal returns. The sum of the abnormal returns constitutes the Cumulative Abnormal Returns, which will be subject to testing. For the expected return models both the Market Model and the Carhart four-factor model will be used.
Table 2:
Abnormal returns and test statistics for the Newman event of 10 Dec 2014 with N = 3890. Event window [-1,0] is included as a (weak) robustness test. Benchmark is the Market Model. CAAR are the Cumulative Average
Abnormal returns on the event date, T-Test refers to the standard T-Test and BMP-test refers to the cross-correlation adjusted T-Test, the adjusted BMP-test (see methodology)
(1) (2) (3)
Event window CAAR T-Test BMP-test [-1,0] [0,0] 0.0011 (0.0071) 0.87 (1.99)** 1.13 (1.79)* [0,1] (0.0049) (1.69)* (1.61) [0,2] (0.0019) (1.13) (1.3) *** p<0.01, ** p<0.05, * p<0.1 Table 3:
Abnormal returns and test statistics for the Newman event of 10 Dec 2014 with N = 3782. Event window [-1,0] is included as a (weak) robustness test. Benchmark is the Carhart Four-Factor model. CAAR are the Cumulative
Average Abnormal returns on the event date, T-Test refers to the standard T-Test and BMP refers to the cross-correlation adjusted T-Test, the adjusted BMP-test (see methodology)
(1) (2) (3)
Event window CAAR T-Test BMP-test [-1,0] [0,0] (0.0000) (0.0067) 0.69 (2.01)** 0.03 (2.00)** [0,1] (0.0002) (1.43) (1.42) [0,2] (0.0001) (0.89) (0.90) *** p<0.01, ** p<0.05, * p<0.1
Testing the significance of the abnormal returns (Market Model as benchmark) of the first event (United States v. Newman 10 Dec 2014) with an event window of [0,0] across the master sample, results in cumulative average abnormal returns of 0.7% at a Tstatistic of -1.99, making it significant at the 5% level. Event window [0,0] should be the main event window of interest, since the informational content of the Court ruling is expected to be incorporated by the market relatively quickly. However, wider windows are also included (up to 2 days after the event [0,2]). Accounting for cross-correlation by using the adjusted T-test including the cross-correlation correction factor, the T-statistic denotes -1.79, now making it significant at the 10% level. In line with the relevant literature on cross-correlation (Collins and Dent 1984 and Kothari and Warner 2007) the T-statistic with the normal T-test
significantly overstates the T-statistic compared to the test that accounts for cross-correlation. It should be noted that if the window is wider, the abnormal returns turn out to be less
significant. Since the market is expected to incorporate the informational content of the ruling relatively quick (on day [0]), this is to be expected. When using the Carhart Four-factor model as benchmark for the expected returns, the results are somewhat weaker and less significant. The above indicates that, in the sample without cross-sectional cuts (almost all publicly traded US firms), the event apparently yields significant negative stock returns on the day of the event.
Consequently, it can be argued that having more lenient insider trading laws is
detrimental for the average US firm. There is of course the problem of causality, but since the event is truly exogenous, the cumulative abnormal returns across the firms turn out to be statistically significant and the ruling significantly impacted the regulatory environment, it is not unreasonably to assume that at least some of the (negative) abnormal returns are due to this event (see paragraph 5.4 for further limitations). Furthermore, using event windows that are longer decreases the effect and significance.
When it comes to the sub-samples with the cross-sectional cuts based on i) R&D expenses ii), bid-ask spread, iii) governance index, and (iv) institutional blockholders it turns out that in the sub-sample where insider trading is more likely to occur, the results are also significant. Testing the significance of the abnormal returns on and around the first event (Market Model) with an event window of [0,0] across this sub-sample results in cumulative average abnormal returns of -1.09% with accompanying T-statistic of -2.01 making it significant at the 5% level. Using the Carhart Four-Factor-Model as expected return model does not yield very different results. When adjusting for cross-correlation by with the adjusted BMP test, the T-stat decreases somewhat. Testing the abnormal returns in the
sub-sample where insider trading is relatively uncommon with the same event window results in a non-significant T-statistic of 1.10. These results are in line with academics like Manove and Beny that argue that insider trading is detrimental for firm value (see paragraph 2.2).
Unfortunately, with the current methodology, we cannot determine through what channels more insider trading leads to significant negative abnormal returns.
Table 4:
Abnormal returns and test statistics for the Newman event of 10 Dec 2014 with N = 99 for Sample I, and N= 150 for Sample II. Note that, as can be read in chapter 3, in order for a firm to be placed in Sample I it must fulfil the following criteria: upper half R&D expenses, corporate governance E-index ratings of 3-6, upper half
mean bid-ask spread and lastly, the presence of at least 1 institutional investor that holds more than 5% of the shares. Event window [-1,0] is included as a (weak) robustness test. Benchmark is the Market Model. CAAR I refers to the CAAR in sub-sample I, the sub-sample that should be affected more strongly. CAAR II refer to the
other sub-sample that is expected to be relatively unaffected. The same goes for T-Test I&II and BMP-Test I&II.
(1) (2) (3) (4) (5) (6)
Event window CAAR I T-Test I BMP-Test I CAAR II T-Test II BMP-Test II [-1,0] [0,0] (0.0009) (0.0109) (0.99) (2.01)** (0.78) (1.97)** 0.0000 0.0092 1.10 1.64* 1.11 1.39 [0,1] (0.0049) (1.69)* (1.61) 0.0004 0.98 0.97 [0,2] (0.0019) (1.13) (1.39) (0.0001) (0.71) (0.69) Observations 99 99 99 150 150 150 *** p<0.01, ** p<0.05, * p<0.1 Table 5:
Abnormal returns and test statistics for the Newman event of 10 Dec 2014 with N = 103 for Sample I, and N= 149 for Sample II. Event window [-1,0] is included as a (weak) robustness test. Cross-sectional cut is based on the same factors as referred to in table 3. Benchmark is the Carhart Four-Factor model. CAAR I refers to the CAAR in sample I, the sample that should be affected more strongly. CAAR II refer to the other sub-sample that is expected to be relatively unaffected. The same goes for T-Test I&II and BMP-Test I&II.
(1) (2) (3) (4) (5) (6)
Event window CAAR I T-Test I BMP-Test I CAAR II T-Test II BMP-Test II [-1,0] [0,0] (0.0000) (0.0111) (0.52) (2.21)** (0.42) (1.87)* 0.0001 (0.0032) 0.03 (1.56) 0.01 (1.65)* [0,1] (0.0055) (1.70)* (1.50) (0.0001) (0.67) (0.41) [0,2] 0.0000 0.12 0.09 0.0000 1.28 0.90 Observations 103 103 103 149 149 149 *** p<0.01, ** p<0.05, * p<0.1
5.2. Salman v. United States - 6 Dec 2016
Using the same methodology as in paragraph 5.1 for the master sample, the overall results are less significant than for the Newman case. As discussed in chapter 3, this Court ruling
significantly impacted the regulatory climate, effectively countering the Newman ruling. After the Salman ruling regulatory enforcement by the SEC became significantly easier. However, as mentioned in chapter 3, some legal scholar anticipated this Court ruling. It could therefore very well be that the price effect of the ruling was incorporated in the stock prices prior to the ruling. Event window [0,0] is the main window of interest, but wider event windows are also included.
Testing the significance of the abnormal returns across the whole sample for this event yields mildly significant and not very strong results; a CAAR of 0.56% with a
corresponding T-statistic of 1.58 and adjusted for cross-correlation 1.53. Therefore, we can cautiously conclude that this event did not significantly effect the stock prices of the average publicly traded US firm. Using the Carhart Four-Factor model yields roughly similar results as the Market Model.
Table 6:
Abnormal returns and test statistics for the Salman event of 6 Dec 2016 with N = 4090. Event window [-1,0] is included as a (weak) robustness test. Benchmark is the Market Model. CAAR are the Cumulative Average Abnormal returns on the event date, T-Test refers to the standard T-Test and BMP refers to the cross-correlation
adjusted T-Test, the adjusted BMP-test (see methodology)
(1) (2) (3)
Event window CAAR T-Test BMP-test [-1,0] [0,0] 0.0000 0.0056 0.51 1.58 0.82 1.59 [0,1] 0.0018 0.04 0.00 [0,2] (0.0000) (1.00) (0.00) *** p<0.01, ** p<0.05, * p<0.1
