Investigation of Market Eciency: An Event Study of Insider Trading in the Stock Exchange of Hong Kong

Elizabeth Wong Stanford University [email protected] May 2002

Abstract Since Great Britains return of Hong Kongs sovereignty to the Chinese government in 1997, attention has been increasingly directed at the degree of market eciency in the post-handover Hong Kong stock market. This study uses a sample of 542 corporate news announcements from January 1994 through December 2000 of Hong Kong and China-aliated rms that are listed on the Stock Exchange of Hong Kong. I examine the eciency of the Hong Kong stock market by investigating the abnormal price and volume performances surrounding the corporate news announcements. Data of U.S. stocks are also used and serve as benchmarks for a comparative analysis of the relative market eciencies. This paper nds that there is very little unusual price and volume behavior for both Hong Kong and U.S. stocks. There exists, however, strong evidence that points towards suspicious insider-trading activities among the Red-Chips and H-share stocks of the China-aliated rms that are listed in Hong Kong, where signicant abnormal returns abound prior to the arrival of good news announcements.

A big thank-you to my advisor, John Shoven. Without his extraordinary guidance, insight, and support, this thesis would not have been possible. I would also like to thank J for his invaluable assistance and encouragement throughout the project. And my eternal love and thanks to my family, to whom I am dedicating this thesis. All errors and opinions expressed herein are my own.

Introduction

Most developed countries around the world have enacted laws prohibiting insiders from trading based on non-public information in order to establish a market environment where security trading is a fair game for all participants. However, the numerous empirical analyses that have focused on insider trading have consistently documented that portfolios which are constructed on the basis of the trading behavior of insiders generate abnormal prots. These abnormal prots, which are the result of insiders prior access to knowledge about public information events, indicate the existence of an inecient market. In an ecient market all available information relevant to the pricing of securities must be rapidly reected in the prices of the securities. The arguments of Fama (1965) form the theoretical foundation for the Ecient Market Hypothesis, which persuasively reasons that in an ecient and active market consisting of many wellinformed investors, equity prices will appropriately reect the eects of information based on present and future expected events. The strong form of the hypothesis asserts that the current market prices fully reect all private (insider) and public information. In other words, insiders should not be able to earn excess returns from privileged asymmetric information. The strong form of the hypothesis represents an absolute standard, and in practice, it is more likely that markets will exhibit only a certain degree of eciency. The previous studies on insider trading display the ability of insiders to make abnormal prots, thus refuting the strong form of the Ecient Market Hypothesis. Among these prior eorts, however, the majority have primarily focused on U.S. capital markets, and thus, their conclusions may not be applicable to security markets

in other parts of the world. In addition, very little research of the same nature has been conducted on Asian markets, especially the nancial markets of Hong Kong. Accordingly, this paper seeks to ll this gap by focusing on the Stock Exchange of Hong Kong (SEHK). Although the SEHK is the largest stock market in Asia outside Japan with a market capitalization in excess of USD615 billion (Fact Book 2000), it has received very little scholarly attention to date. It is thus the primary purpose of this paper to examine the eciency of the Hong Kong stock market by using an event study approach to investigate the abnormal price and volume performance surrounding corporate news announcements of rms listed on the SEHK. More specically, I aim to test for the existence of signicant abnormal changes in equity prices and volumes that take place prior to, during, and after homogeneous news announcements. Since changes in equity prices must fully reect all new available information in an ecient market, any changes prior to a news release directly suggest the existence of insiders taking advantage of their possession of private information and, or, information leakage by insiders, and thus signal a dishonest market. This is an aspect of the securities market environment to which the securities data of both Hong Kong and the U.S. have not been systematically applied. Most prior studies make use solely of the data surrounding the insider trade transaction dates, which are published monthly by their respective securities commissions, and test for abnormal prots around those transactions periods. Such an approach might not reveal illegal insider trading that is intentionally not reported to the securities commissions and that abuses non-public information. I take a dierent approach by directly examining the relationship between public announcements of corporate news and abnormal changes in prices and volume. Such an examination of stock price and trade volume behavior 2

over the pre- and post- announcement periods might provide a clear indication of whether insiders do indeed trade on unreleased information. Another objective of this paper is to gather evidence on the behavior of insider trading protability in the periods both before and after the handover of Hong Kongs sovereignty. Since Great Britains return of Hong Kongs sovereignty to the Chinese government on July 1st, 1997, there has been increasing concern about the prevalence of an emerging market mentality in the territorys nancial markets, and special attention to the question of the degree of market eciency in the posthandover Hong Kong stock market. In particular, throughout the four year period between 1998 to 2001, the media paid a great deal of attention to the rampant rumors circulating within the nancial community about illegal insider trading. The media have also paid attention to the pervasiveness of inappropriate information leakage, which they have suggested is the source of the frequent unexplained leaps and drops in stock prices. Most noticeable amongst this coverage was the March of the Red Chips, which was a sudden surge in the stock prices of Mainland Chinese companies listed in Hong Kong (red-chips).1 This study will thus help us to gain insight into the assessment of the degree to which the Hong Kong stock market is ecient, and particularly, the dierence in degree between pre-handover and post-handover Hong Kong. The last issue concerns the sensitivity of stocks listed on the SEHK. In a study conducted by Fung and Poon (2000), the prices of China-aliated securities traded in SEHK displayed a tendency to be more sensitive to good news than to bad news.
See for example, Betting on the Red Chips, Time 5 May, 1997, and Red-Chip Revelry, BusinessWeek, available at http://www.businessweek.com/1997/26/b3533156.htm, accessed 27 April, 2001.
1

This paper will also examine whether such a phenomenon can be found among the stocks of Hong Kong rms or whether it is only a unique element of the March of the Red Chips. As I aim to provide a more extensive depiction of Hong Kongs stock market eciency, I use the U.S. stock market as a point of comparison throughout the paper. In order to commence my investigation of the SEHK, I would now like to provide some of its background information, with a particular emphasis on the enforcement of insider-trading regulations and the types of shares being traded.

The Stock Exchange of Hong Kong & Insider-Trading Regulations

Although formal stock-trading activities began in Hong Kong as early as 1891, the year the Stock Exchange of Hong Kong was founded, the stock market has only been an important source of capital funds for business enterprises since about 1969. During Hong Kongs industrialization in the early 1950s, most rms started as modest sole-proprietor ventures, which gradually evolved into partnerships or private companies. Firms were incorporated or controlled from abroad, and public corporations were not common until the late 1960s. The establishment of three new stock exchanges (the Far East, the Kam Ngan, and the Kowloon) during the years of 1969 to 1972 was partly stimulated by the sharp upsurge in stock-market activities. Then, in 1986, the unication of the four stock exchanges brought about a new era in share trading in Hong Kong. The negotiation process for this unication took nearly ten years, and the unied Stock

Exchange of Hong Kong Limited did not commence operations until April 1986. After the unication of the stock market and the introduction of computer-assisted trading, the Stock Exchange attracted more business from international investors, and has quickly transformed from an essentially domestic market into a market of international signicance. In Hong Kong, the stock exchange has not been used signicantly for raising capital funds for the manufacturing sector which has been the main driving force behind Hong Kongs economic growth. This is largely due to the small size of the average manufacturing rm, the easily available bank loans for xed-asset investment, and the fact that family-controlled manufacturing rms prefer to borrow from within rather than outside the family. Thus, the trading of equities tends to be dominated by major rms in the business areas of banking, public utilities, trading, and property holding and development. In particular, the listed companies in the property and nance sectors together with various consolidated enterprises account for at least 70% of the total market capitalization. Regulation of the securities industry in Hong Kong gained its main impetus from the stock market boom and collapse of the early 1970s. Prior to 1971, the industry was only loosely regulated under the relevant provisions of the Companies Ordinance and the Stock Exchange Control Ordinance. It was not until 1971 that section 140 of the Securities Ordinance sought to deal specically with insider dealings. Although section 140 was enacted, it was never properly enforced and was eventually repealed by the Securities (Amendment) Ordinance of 1978. During the time period between 1978 and the late 1980s, regulations remained ambiguous. In 1989, the Hong Kong government nally published the Securities (Insider Dealing) Bill which aims to provide a more comprehensive denition of insider dealing and to authorize 5

the Insider Dealing Tribunal to impose sanctions on persons identied as having been involved in illegal insider trading. Around the same time, a complete and thorough review of the overall organization of securities trading followed in the shape of the Report of the Securities Review Committee. This report resulted in various reforms. As one of the rst reforms, the establishment of the Securities and Futures Commission (SFC) took charge of the responsibility of regulating trading in securities markets in Hong Kong (McGuinness, 1999). In September 1991, after two years of widespread debate, a series of much more concrete regulations was enacted for the purposes of fair trading. These insider trading laws were largely inherited from the United Kingdom, which in turn has adopted a large body of regulations that were developed by the courts in the United States. In the U.S., the Securities and Exchange Commission (SEC) is charged with the power to enforce insider-trading rules. In order to add transparency to insider dealings, under Section 16(a) of the 1934 Securities Exchange Act, insiders, who are dened to be a companys management (directors and ocers) and substantial shareholders holding 10% or more of the rms outstanding shares, are required to report to the SEC within ten days after the close of each calendar month if any changes in the insiders equity ownership occurred during that month. These insider-trade transactions will then be compiled and published by the SECs monthly Ocial Summary of Security Transactions and Holdings. Moreover, to deter insiders from realizing short-swing prots by taking advantage of their access to private information, Section 16(b) of the Act requires insiders to return all prots made within six months on purchases and subsequent sale to their companies. An insider in the U.S. who illegally trades with the possession of non-public information will be faced with either civil or criminal penalties, or both. Under the provisions laid 6

down in Section 17(a) and 10(b) of the Act, together with Rule 10(b)-5 of the SEC, it is illegal for any insider or quasi-insider to directly or indirectly trade without public disclosure of material non-public information. If they cannot disclose, they must abstain from trading. According to the Insider Trading Sanctions Act of 1984, persons found guilty of insider-trading violations will face a civil penalty of up to three times the prots made (or losses avoided) from the trade transaction. In 1988, Congress increased penalties on insider-trading violations by enacting the Insider Trading and Securities Fraud Enforcement Act. This provision increased the severity of the criminal penalties by making violations punishable by ne or up to ten years of imprisonment or both (Langevoort, 2001). However, such criminalization of insider-trading violations is not practiced in Hong Kong. Hong Kong has two main pieces of legislation on insider trading: (i) the Securities (Disclosure of Interests) Ordinance (SDIO), and (ii) the Securities (Insider Dealing) Ordinance (SIDO), both brought into force during the securities law reform in 1991. Both ordinances were considered to be one of the SFCs important steps in improving transparency and fairness to public investors. The SDIO species how to dene insiders, and what constitutes insider trading, while the SIDO outlines the proscriptions applied to insider dealings and the penalties for failing to comply. Similar to the 1934 Securities and Exchange Act of the U.S., the SDIO denes insiders to include persons connected with a corporation (director, ocers, workers, and substantial shareholders holding 10% or more of the outstanding equity of a company), persons contemplating a takeover, a tippee (a person who has information which he or she knows is relevant in relation to a corporation), and government ocers. As in the U.S., Hong Kong insiders are obliged to report their stock transactions to the SEHK within ve days of their trades, which will be published in the SEHKs 7

Director/Chief Executive Notications Weekly Summary and released to the public. Under the SIDO, insiders with possession of non-public relevant corporate information are proscribed from trading any related securities and counselling or procuring another person to deal in such securities (Ho, 1998). However, as a late comer to insider-trading regulations, Hong Kong faced many criticisms for its overly lax enforcement of insider-trading provisions. In 1996, a survey conducted among Hong Kong fund managers revealed a 60% disapproval rate of the insider-trading regulations, pointing to its weak sanctions and enforcement structure. This perception of market regulatory deciency has been portrayed by the local media as enabling insiders to reap benets from their trades.2 The nancial and legal communities have contended that stricter sanctions should be imposed as Hong Kongs regulatory control over agrant insider trading leaves much to be desired. Under the current securities ordinances, if the Insider Trading Tribunal nds an insider breaching the regulations, it may prohibit the person from direct or indirect involvement in the management of the company for a maximum period of ve years. The Tribunal may also order a ne in an amount not exceeding the amount of any prot gained or loss avoided as a consequence of the illegal insider trading. However, the person found liable may not be convicted for a criminal oense, nor be punished by imprisonment as in the U.S. There is also a concern about the abuse of the ambivalence in the no-prot defence in Hong Kong, which renders an insider not guilty of violation of insider dealing if he or she establishes that (s/)he entered into the transaction otherwise
See for example, Insider buyers and sellers hold key to market, South China Morning Post 2 January, 1995, and Heard in Hong Kong: The Inside Track Can Be Protable, Asian Wall Street Journal 27 February, 2001.
2

than with a view to the making of a prot or the avoiding of a loss (whether for himself or another) by the use of relevant information.3 As expressed by Ho (2000), this defence is a curious indulgence. A more disturbing fact is the constitution of the Hong Kong consultative government system where 70% of the legislators are executive directors or shareholders, and only 26% have no commercial concerns.4 In an eort to solicit support from the Hong Kong public (particularly the elite) without granting the Hong Kong people the right to vote, the British administration had established a consultative system aimed to have representative members drawn from a wide range of persons from dierent sectors of the community to participate in policy-making. Unfortunately, the majority of these consultative members were drawn from industrial, nancial, and real estate circles, and a curious situation arose as directors and shareholders of corporations were invited to sit in legislative meetings where a large body of information is shared and could be used to further an insiders information advantage. Adding to the intrigue are the China-aliated companies that are listed on the SEHK. There are two types of China-aliated stocks: the Red-Chips and the Hshares. Red-Chips are stocks of Hong Kong registered and listed rms in which Chinese interests control more than 35% of issued stock. They are usually established by a ministry or a provincial government in China, and their senior management includes important gures in the mainland (McGuinness, 1999). Ever since the Red-Chip frenzy and boom commenced in 1992, the Red-Chips have taken on a talismanic mystique among Hong Kong investors. A parallel phenomenon was the rise of the H-shares, which are stocks of Chinese central government-owned rms
3 4

SIDO, s. 10(3). Ming Pao 8 April 1997, 1.

that remained incorporated in China but obtained permission from the government to list on the SEHK. Both the Red-Chips and the H-shares have been the targets of local media reports about suspicious insider dealings and possible price manipulations. Stories abounded that reporters were given insider-trading tips in exchange for favorable press coverage.5 The Hong Kong Securities and Futures Commission (SFC) is aware of the China-aliated stocks possible insider information abuse.6 However, the SFC might not be in a position to regulate them because, while they are listed in Hong Kong, the parents of both the Red-Chips and H-shares are regulated from Beijing, where poor disclosure, lack of transparency, and a relationship-based system are the hallmarks of the Chinese rms entities.

Review of Previous Research

Unlike the abundance of past literature that has examined insider trading in the U.S. capital markets, there has been very little research attention devoted to the investigation of insider-trading practices in the Hong Kong stock market. The little research that has been done has only provided a general analysis of the stock exchanges overall eciency. The following literature review thus primarily considers prior studies done on insider trading in the U.S. capital markets. In particular, this review pays attention to the protability of abnormal trading, the question of whether outsiders, by mimicking insider trades, can earn abnormal prots, and the attendant moral debate on insider-trading practices.
The Wild West of the East, The Economist, 10 July, 1997, available at http://www.economist.com/PrinterFriendly.cfm?Story ID=370561, accessed 27 April, 2001. 6 Red-Chip Revelry, BusinessWeek, available at http://www.businessweek.com/1997/26/b3533156.htm, accessed 27 April, 2001.
5

10

3.1

Protability of Insider Trading

Although insiders are legally constrained by the securities commission from trading on private information, the control mechanism is imperfect. A statistical study such as this one can document the existence of abnormal trading in a stock market, but, because it is very dicult for regulators to dierentiate between insider trading that infringes against the regulations and insider trading that does not, the enforcement activities tend to be ineective. Insider trading comprises many dierent activities. Insiders, like other investors, may trade securities because of changes in wealth, preferences, and consumption opportunities. Unlike other investors, however, insiders frequently have access to private information and may trade to realize the benets of that information. Since the use of private information is not directly observable, researchers have focused on the issue of private information by examining the protability of trading strategies based on insider trading. As noted earlier, the majority of the previous studies were based on data drawn from the U.S. capital markets. Scholars have analyzed information presented in SECs monthly Ocial Summary of Securities Transactions and Holdings to assess whether 1) inside traders earn abnormal prots from the use of non-public information (a test of the strong form of market eciency), and whether 2) non-insider investors can prot from the information published in the Ocial Summary (a test of the semi-strong form of market eciency). To date, there have been no conclusive ndings on either of these issues. In investigating abnormal prots obtained by insiders, the classic approach has been to form portfolios of rms on the basis of the number of insiders who buy and sell in a given month and to run statistical tests on the portfolios returns using

11

the Capital Asset Pricing Model (CAPM). One portfolio is formed when buyers exceed sellers, and another portfolio when sellers exceed buyers. The returns of these two portfolios are compared in subsequent months, and signicant dierences between them are attributed to the abnormal returns earned by insiders. In one of the pioneer studies on the issue of insider trading, Lorie and Neiderhoer (1968) investigate stock performances following months in which there are at least two more buyers than sellers or at least two more sellers than buyers among insiders of a company. They discover that a security experiencing an intensive buying month is more likely to advance than to decline relative to the market in the six months subsequent to the event. And, conversely, a security experiencing an intensive selling month is more likely to decline than to advance relative to the market in the six months subsequent to the event. In conrming the results of Lorie and Neiderhoers investigation, Pratt and DeVere (1970) and Jae (1974) report signicant abnormal returns earned by insiders, especially in situations when buyers outnumber sellers by three or more in one month or sellers outnumber buyers by three or more. In using the classic approach, Finnerty (1976) tries to test the strong form of the ecient market hypothesis by determining whether insiders average returns from their market transactions are above that of the market in general. In doing so, he points out the shortcomings of the previous studies. Finnerty identies their selection bias in choosing samples in favor of performances that would have a higher probability in exceeding the returns earned by an average insider and also the lack of data availability, because no precise stock prices of insider trades were reported to the Securities and Exchange Commission before 1965 (previous studies utilized data before 1965). In his study, therefore, Finnerty chooses to use the entire population of insiders reported in the 12

Ocial Summary after 1965 to evaluate the performance of an average insider. He concludes that insiders can outperform the market by beneting from asymmetrical information advantages about their corporations. Consistent with the prior studies on insider trading in the U.S. stock market in the 1960s, Finnertys nding implies a refutation of the strong form of the ecient market hypothesis. Nevertheless, Seyhun (1986) questions the appropriateness of the above mentioned studies use of the CAPM when testing for abnormal returns on securities. The author cites works by Banz (1981) and Reinganum (1981) that show that the CAPM based residuals are on average positive for small rms and negative for large rms. Such systematic bias in CAPM residuals can lead to inaccuracy in estimating abnormal returns in insider-trading studies. To avoid the biases introduced by the CAPM, Seyhun uses the market-model in measuring the abnormal returns of individual stocks. His results indicate that insiders are able not only to earn abnormal prots but also to predict abnormal future stock price changes. Evidence presented in the study documents that insiders purchase stock before an abnormal rise in stock prices and sell stock before an abnormal decline in stock prices. In addition, it shows that insiders who are expected to be more knowledgeable about the overall aairs of the rm, such as the chairman of the board of directors or ocer-directors, are more successful predictors of future abnormal stock price changes than ocers or shareholders alone. His results also suggest that insiders can discern the dierences in the value of their information and trade greater volumes of stock to exploit more valuable information. In a later study, Seyhun (1992) strengthens his initial ndings by documenting a strong relation between past aggregate insider trading and future excess stock returns. This later study specically reports that during the period 1975 to 1989 13

up to 60% of the variation in twelve-month-ahead excess stock returns can be forecast using the previous twelve-month aggregate insider trading. Furthermore, the study shows that the predictability of stock returns increases with the length of the forecasting horizon, the number of months of past insider trading, and the market sensitivity of stocks. As in Seyhuns studies, many previous works have generally assumed that insider trading activity is based on the exploitation of private information prior to public announcements. Several articles have examined aspects of insider trading other than the direct relationship with subsequent returns. Certain patterns of insider trading would be consistent with the use of specic private information, such as purchasing before the public release of good news and selling before the release of bad news. Penman (1982) investigates one insider-trading pattern by focusing on management forecasts of annual earnings. He uses the sign of price change on the days preceding and including the forecast announcement to determine the expected direction of insider-trading activity. Portfolios are chosen on the basis of these price changes, and net insider buying and selling behaviors are examined for four months before and after the announcement. He concludes that there appears to be greater selling by insiders before forecasts associated with either a price decrease or a price increase, arming the hypothesis that protable trading is associated with the release of public information. Other studies investigating insider trading relative to specic corporate announcements include Elliot and Richardson (1984), Givoly and Palmon (1985), and Oppenheimer and Dielman (1985). Unlike Penmans ndings, these studies conclude that at best only a small proportion of insider trades may be related to rm-specic announcements. Elliot, Morse, and Richardson attempt to test whether insiders can 14

earn abnormal prots from advance knowledge of new announcements prior to their release to the public. Specically, the authors study abnormal returns generated before and after the public disclosure of information concerning dividends, earnings, bond ratings, mergers, and bankruptcies. It is a general perception that insiders can benet from their private access to companies information and large prots can be made through the exploitation of such information. In their study, Elliot, Morse, and Richardson cite previous works by Lorie and Neiderhoer (1968), Pratt and DeVere (1970), Jae (1974), and Finnerty (1976), all of which performed indirect tests by focusing on the abnormal returns obtained by insiders on the basis of their buying and selling decisions as published by the SEC. Such prior studies have supported the ability of insiders to earn abnormal returns, but the authors question those studies general assumptions that insider trading activity has been based on private information. We do not know to what extent these returns are associated with specic news announcements. In their journal article, Elliot, Morse, and Richardson take a dierent approach by directly examining the relationship between insider trading and public news announcements. They reason that if insiders are in fact exploiting their early access to private information, then insider buying activities should precede the issuance of good news and the opposite should hold true for announcements of bad news. In addition, since corporate insiders cannot make any trade transactions on holding periods of less than six months, if they take a speculative position prior to an announcement, their positions should be reversed following the public announcement subject to the six-month constraint. Elliot, Morse, and Richardson then perform various binomial tests to investigate insider trading that surrounds four good-news events (large earning increases, large dividend increases, bond rating increases, and merger announcements) and 15

four bad-news events (large earning decreases, large dividend decreases, bond rating decreases, and bankruptcies). They test the distributional characteristic of insider trading by comparing its frequencies twelve months before each announcement, the month of the announcement, and twelve months after each announcement. The signicance tests results, however, do not suggest a consistency in speculative prots through the use of private information. In addition to testing whether insider buying is greater than, less than, or equal to insider selling, the authors also examine extreme insider trading activity before information events. Extreme insider trading is stipulated to be when the number of buyers exceeds the number of sellers by three or more for a given month and vice versa. If insiders take speculative positions in using private information, we should observe periods of intense buying to precede good-news events, and the opposite for bad-news events. But, once again, there is no consistency among the results attained by the authors. Elliot, Morse, and Richardson thus make use of a multivariate test to examine whether the information events might jointly explain extreme insider trading. However, the coecients from this logit model are once again proven to be statistically insignicant. Thus, the results of this investigation are only weakly consistent with the common belief of a strong link between insider abnormal prots and news announcements as most insider trading is found to be independent of such information events. This is a surprising conclusion. Elliot, Morse, and Richardson try to explain why only a small portion of insider trading can be justied by use of insider information by noting that many uses of private information may be subtle. Insiders who hold onto stocks before good announcements may go unnoticed, but they are still eectively exploiting their private knowledge. Nonetheless, it must be pointed out 16

that Elliot, Morse, and Richardsons choice of sample, especially their decision to exclude the data of small rms, might have signicantly biased their ndings. The evidence presented in a later study by Givoly and Palmon (1985) may help in clearing up the puzzling results obtained by Elliot, Morse, and Richardson. Givoly and Palman wish to determine the extent to which the abnormal returns on insider transactions found by previous studies are caused by price changes due to the subsequent disclosure of news and by information revealed by the trade itself. They nd that prots from insider trading are not associated with the disclosure of specic news about the company. Their evidence suggests that the abnormal returns to insider transactions endure well beyond the typical period of market reaction to the disclosure of a specic news event. Furthermore, they show that a signicant abnormal return is produced in the wake of the trades themselves, lending support to the hypothesis that outside investors accept the superior knowledge of insiders and follow their footsteps. In interpreting the evidence, one must, however, bear in mind that Givoly and Palmon use securities listed on the American Stock Exchange, which are the stocks of relatively small rms. Such a sample choice might aect the end results. Later studies, such as those by Seyhun (1986), Roze and Zaman (1988), Eysell and Arshdai (1993), Bettis and Vickrey (1997), and Chang and Suk (1998), pursue the question that Givoly and Palmons study raises, that of whether outsiders, by mimicking insider trade transactions, can also earn abnormal prots. There exists, however, no conclusive evidence on this issue. For example, Chang and Suk, by examining the reaction of securities prices to the publication of the Insider Trading Spotlight column in the Wall Street Journal, discover that there were signicant stock price changes and increases in trading volume following insider trading and 17

the publication of the column. Such results directly point to the outside investors ability to gain abnormal prots and the inconsistency with the semi-strong market eciency that all publicly available information should be already reected in the securities prices. Furthermore, their ndings are consistent with the argument of Givoly and Palmon (1985) that insider trading triggers additional trading by other investors. On the other hand, the evidence presented in the studies of Seyhun (1986) and Roze and Zaman (1988) refute the notion that an outsider may benet from mimicking insider trades. Both studies examine the realizable returns to outsiders after the reporting to SEC of insiders trade transactions and the release of the Ocial Summary to the public. Seyhuns article reports that the studys examination of the abnormal return to outsiders net of the bid-ask spread plus the commission fee nds that there was no signicantly positive abnormal return to outsiders. Similarly, Roze and Zaman discover that when a 2% trading cost is imposed, the abnormal prots of outsiders decrease so substantially that the prots approach the point of statistical insignicance. Since outsiders cannot use publicly known information to earn abnormal prots, the two studies suggest that the semi-strong form of market eciency holds.

3.2

Current Debate: Insider Trading Harmful or Good?

In addition to the numerous investigations on the potential protability generated by insider trading practices, a large body of literature has also been devoted to the debate of whether such practices are generally harmful, and thus, require harsh regulation, or whether such practices may be benecial. As Meulbroek (1992) notes in

18

her article An Empirical Analysis of Illegal Insider Trading, though there has been a great deal of debate about the pervasiveness and harmfulness of insider trading, nancial economists remain divided over the need to regulate insider trading. Opponents of insider trading contend that insider trading leads to decreases in market liquidity, produces abusive managerial practices, and is unfair to public investors. Proponents of insider trading, like Manne (1966) and Carlton and Fischel (1983), however, promote insider tradings extensive benets. Manne (1966) documents the ability of insider trading practices to improve the accuracy of stock prices by incorporating a large fraction of insider information into the share prices before the information is made public, thus fostering an ecient market. Results of Meulbroeks investigation conrm Mannes contention. Unlike prior eorts in examining insider trading, Meulbroek utilizes not the self-reported transactions that are published by the Securities and Exchange Commission, but rather non-public SEC documents of civil or authoritative cases of insider trading violations. The ndings from her study might have new implications for public policy on the regulation of insider trading. The author documents the uncanny ability of the U.S. stock market to detect the possibility of illegal insider trading and nds that about half of the pre-announcement price run-up observed before takeovers occurs on insider-trading days. The results from Meulbroeks study point to the rampant price run-ups before takeovers and the pervasiveness of insider trading, which some might see as demanding stricter regulations. Nonetheless, Meulbroek also shows that insidertrading practices can in fact help speed up the price discovery process, which might be benecial in promoting more informative prices and should be taken into consideration when evaluating future legislation concerning insider-trading practices. And 19

consistent with Meulbroeks ndings, the investigation on Anheuser-Buschs 1982 tender oer for Campbell Taggart by Cornell and Sirri (1992) nds that insider trading in Campbell Taggart led to a substantial price run-up before the tender oer announcement. The authors also note that the insider trading practices did not reduce liquidity, but, instead, created more informative prices. In a similar vein, when examining the eectiveness of the insider trading regulatory laws in the Amsterdam Stock Exchange, Kabir and Vermaelen (1996) nd that instead of promoting liquidity and informational eciency, the insider trading restrictions resulted in what the authors referred to as a regulatory overkill. They contend that the argument for eliminating insider trading, that it will increase liquidity, ignores the actual liquidity-enhancing role of insider trading. These restrictions constitute a regulatory overkill in the sense that not only did the liquidity actually decline after the enactment of restrictions, but so did the stock markets speed of adjustment to positive earnings announcement news. This result is consistent with Meulbroeks and Cornell and Sirris ndings in the U.S. markets.

3.3

Empirical Studies on the Stock Exchange of Hong Kong

The majority of past research eorts on insider trading have primarily focused on the U.S. capital markets. Little research of the same nature, however, has been conducted on the Asian stock markets, especially on the Stock Exchange of Hong Kong. Moreover, prior studies on the Hong Kong stock market have mainly focused on testing its market eciency rather than on insider-trading practices. One such study is found in an earlier research eort of Kwong and Wong (1984). The authors test the weak form of market eciency by conducting tests using serial correlation

20

coecients. They conclude that, in comparison with other studies for the U.S. and Great Britain, their own serial correlation tests do not support the weak form of market eciency while their runs tests do. Since the latter contain individual cases of signicant discrepancies from non-random patterns, the overall conclusion must be that the Hong Kong stock market does not exhibit signs or symptoms consistent with the weak ecient form. The unication in 1986 of the four stock exchanges in Hong Kong into the Stock Exchange of Hong Kong (SEHK) generated a large centralized market where information to all the participants is made instantaneously available through screen trading. An improvement in the overall eciency of the market was thus expected. Cahn and Donleavy (1989) test for this conjecture by using a before and after unication methodology. They use daily securities prices for all 260 of the shares traded between September 1985 and August 1986, and apply three standard eciency tests: the serial correlation test of the daily price changes, the runs test and a time series regression to test the existence of a time trend. The authors conclude that it is difcult to claim that the Hong Kong stock market is ecient even in the weak form, and thus, contrary to expectation, eciency did not improve after the unication of the stock exchanges. Despite the importance of insider trading and its topicality, to my best knowledge, there has been only one published empirical analysis on the insider trading practices in the Hong Kong stock market. The study by Cheung and Wu (2000) specically aims to investigate such insider-trading practices in the Stock Exchange of Hong Kong. As the largest stock market in Asia outside of Japan, SEHK is considered to be one of the largest emerging stock markets in the world. The authors explore the question of whether the widespread belief of the ineciency of emerging 21

markets is applicable to SEHK, and also examine the eect of the unique ownership structure of Hong Kong companies on insider-trading practices. In particular, the paper addresses the following three questions: (1) Do corporate insiders in Hong Kong make abnormal prots through trading their own stocks? (2) Can quasi-insiders, who may have access to the information of insider trading, make abnormal prots? and (3) Can ordinary investors who mimic the publiclyavailable information of insider trades earn abnormal returns? The authors discover that abnormal prots associated with insider trading are prevalent only in smallsized rms, especially with trades of large volume. By following the information of insiders in small rms, quasi-insiders are also able to benet from abnormal returns. In addition, the authors conclude that the medium-sized and large rms in the SEHK exhibit the strong form of market eciency as insiders make no abnormal prots. Cheung, Wong, and Wu utilize the event-study approach where the transaction date, report date, and publication date as reported in the Director/Chief Executive Notications Weekly Summary (DENWS) are adopted as the three respective event dates. The authors examine and compare the abnormal returns before, after, and on the event dates during the window period between September 1991 and June 1993. If there were dishonesty in the market because insiders possess an asymmetrical information advantage, one would observe higher (lower) stock prices after insiders buy (sell) stocks. In an eort to reduce the potential noise from trades not associated with information initiative, the authors opt to use only intensive trading events where consecutive buys or sells occur within a ten-day period, with no opposing transactions in the same period. To examine the rst question of insider protability, the daily abnormal return 22

for each individual stock in their sample is calculated as an excess daily return. For each day in the event period, the average excess return is calculated by averaging all individual excess returns for that day, and the signicance of the average excess return is then calculated by test statistics. The cumulative daily average excess returns are then computed by summing up all daily excess returns, and their significance tested. In exploring the second and third question on whether quasi-insiders and outsiders can also earn abnormal prots, the authors examine the abnormal returns made on the day the SEHK receives an insiders report and on the day the report is released to the public. If the market exhibits the semi-strong form of market eciency, one should not observe any abnormal returns after the publication of information on insider trading. In studying the results attained from statistical regressions, the authors found that a signicant positive abnormal return of 0.33% is shown on the day following the insider purchases, and similarly, a signicant negative abnormal return of 0.28% is found on the second day following the insider sales. These results imply that the price run-ups are associated with insider sales and that the insider may benet from abnormal returns. However, when cumulative daily excess returns are separated by rm size according to each rms market capitalization (S1= smallest, S2 = medium-sized, S3 = largest), only S1 indicates signicant returns within the postevent period. Turning to testing whether quasi-insiders and outsiders can mimic the information of insider trading to earn abnormal prots, Cheung, Wong, and Wu apply a similar excess return approach to an event window which includes the report date and the publication date of the insider trades. The authors nd that both S1 and S1V2 have a signicantly positive cumulative abnormal return after insider pur23

chases, and other subgroups (according to trade volume, with V1 being the smallest trade volume among S1 rms) exhibit insignicant returns. These results suggest that only quasi-insiders who mimic insider purchases of small rms can earn abnormal prots. In addition, in examining the cumulative abnormal return in the post-publication period, only S1V3 sales (sales of a small rm with the greatest trading volume) have a signicantly negative abnormal return while those of all other subgroups are insignicant. This implies that only the information of insider sales associated with small rms with a high level of trading volume can help outsiders in making abnormal prots. The overall results of Cheung, Wong, and Wus study suggest that only insider purchases associated with small rms can lead to abnormal prots and only the insider sales information of small rms with a relatively large trading volume is valuable to quasi-insiders and outsiders. The inability of medium and large rms to earn abnormal prots from insider information may signal both semi-strong and strong forms of market eciency in the Hong Kong stock market for such rms. However, the studys results might call for a stricter regulation on insider trading, especially among small rms in Hong Kong. None of these prior studies on insider trading reaches conclusive ndings on the questions of the protability of insider trading, whether outsiders can mimic the protable trading strategies of insiders, or whether insider trading practices are harmful or benecial. As the greater part of the previous studies reviewed here are based on the U.S. stock markets, it is the primary aim of this paper to provide a more in-depth empirical analysis on the prevalence of insider trading in the Hong Kong stock market.

24

Research Design

As outlined in the literature review, prior studies of insider trading have mostly utilized the insider-trade transactions self-reported to securities commissions as data. It is questionable whether such data really reect the occurrences of insider trading. It is likely that corporate insiders would refrain from reporting their questionable transactions to the securities commissions, and by relying only upon the data provided by securities commissions studies of insider trading are likely to produce skewed results. In this study, therefore, instead of using the insider trades reported in the Director/Chief Executive Notications Weekly Summary (DENWS) that is published by the Stock Exchange of Hong Kong, I adopt a broader method to test for information leakage and/or abuses of the possession of private information before public news announcements by listed companies. Through an event-study approach, patterns of price changes for the periods preceeding public announcements could yield interesting evidence about market eciency. If insider trading rules are strictly obeyed, there should be no abnormal price changes before the public release of relevant news, but rather, a clean jump on stock price only on the announcement date. Thus, if an abnormal price behavior occurs prior to a news announcement, such a phenomenon might be attributed to either anticipation by market participants or insiders abusing their knowledge of private information. Anticipation plays a valid role for certain types of corporate announcements, like earnings, mergers and acquisitions, however, as pointed out by Bodie, Kane, and Marcus (1999), evidences of leakages appear even when the publics access to the information is not gradual, thus implying that insiders trading on private information play a more signicant role in accounting for

25

abnormal price changes prior to news announcements.

4.1

Data

This paper investigates insider-trading behavior surrounding good-news events (large earning increases, large dividend increases, merger announcements, and the like) and bad-news events (large earning decreases, large dividend decreases, bankruptcy announcements, and the like). I use daily trading data collected from Datastream terminals, and corporate news announcements obtained from Dow Jones Interactive. Given the limitations of the data collecting process, I have chosen to limit the period under study from January 1994 through December 2000. Due to the large number of dormant stocks listed in the Stock Exchange of Hong Kong, the data of securities are screened using the following process. I select only stocks which are among the 126 constituents of the newly established Hang Seng Hong Kong Composite Index, and only rms that had actively traded stocks in the period of study. I have chosen to eliminate all stocks that had no event-worthy news in Dow Jones Interactive in that time period. Event-worthy news comprises, but is not restricted to, all restructuring announcements. These announcements include news about changes in capital structure, mergers, takeovers, acquisitions, spin-os, sell-os, joint ventures, and board change announcements. Atypical earnings announcements are also included. The same process is used to select stocks for the U.S. from the Dow Jones Industrial Index constituents and for the China-aliated stocks from the newly formed Hang Seng China-Aliated Corporations Index (the Red-Chips) and the Hang Seng China Enterprises Index (the H-shares)constituents. Only 46 Hong Kong and 19 China-aliated stocks survived this screen, while all of

26

the 30 DJI constituents passed.7 Table 1 lists a total of 542 events involving both Hong Kong and China aliated rms, and the 489 events that involved U.S. rms. The number of announcements are listed by type and study periods. The day of the event/news announcement is taken to be the date on which the news was rst reported by Dow Jones Interactive. Since it is dicult to pinpoint the exact time the news was made public, for the purpose of this study I dene the event window to include the day before, up until and including one day after, the news announcement. I also dene an event period to be from 30 days before to 30 days after the news announcement. In some cases, if data is missing at the beginning or end of the event period, the event period will be shorter. I collected the daily closing prices, and the daily trading volumes for the study period from January 1994 through December 2000. The daily closing price and turnover volume of the Hong Kong market proxy, the Hang Seng Index (HSI), are also collected over the same period. The HSI is a major indicator of the Stock Exchange of Hong Kong and consists of 33 stocks that represent over 75% of total market value on the SEHK over the sample period (Factbook 2000). For the U.S. sample, the S&P500 serves as the market proxy. To examine abnormal returns and trading volumes, I divide the samples into two main groups, namely, good news announcements and bad news announcements. The direction of the price change after a news announcement determines whether a news event is classied to be either good or bad. The samples are then subdivided into two groups, Prior-1997 and Post-1997, for the investigation for evidence of suspicious
The U.S. sample will also include pre-merger Exxon, Mobil, Citicorp, and Travelers Group as their mergers took place during the study period. Thus the U.S. sample will be composed of a total of 34 stocks.
7

27

Table 1: Number of Announcements by Type for the Period 1994-2000

A. Good News Announcements

Earnings Before 1997 Hong Kong China-aliated After 1997 Hong Kong China-aliated All Periods United States Hong Kong China-aliated 110 79 17 57 27 10 94 84 27 261 190 54 37 9 12 4 40 11 89 24 42 8 15 6 44 16 101 30 Non-Corporate Corporate Total

B. Bad News Announcements

Earnings Before 1997 Hong Kong China-aliated After 1997 Hong Kong China-aliated All Periods United States Hong Kong China-aliated 80 95 40 39 27 7 83 72 31 202 194 78 50 18 15 5 42 19 107 42 45 22 12 2 30 12 87 36 Non-Corporate Corporate Total

28

insider-trading activities in the periods both before and after the handover of Hong Kongs sovereignty. The samples are then further subdivided into the dierent types of announcements: earnings, corporate, and non-corporate.8 Table newsstat reports summary statistics for the frequencies of the dierent types of news announcements for the rms in my sample from 1994 to 2000.

4.2

Abnormal Returns

The daily returns of an individual security i are calculated as Ri,t = ln(Pi,t+1 ) ln(Pi,t )

(1)

where Pi,t represents the closing price for security i on day t. The same method is applied in computing the returns of market indices. In order to test each rm-series in my samples for abnormal returns for every day in the event period, I use the market model as outlined by MacKinlay (1997). The return for any security i at time t is

Ri,t = i + i Rm,t + i,t

(2)

where Ri,t and Rm,t are the returns on security i and on the market at time t respectively. To eliminate bias for the impact of the event, the securities data during the 120 days prior to the event window are utilized to estimate the parameters and for
Corporate news include major investments, joint-ventures, mergers & acquisitions, management changes, lay-os, spin-os, and sell-os. Non-corporate news include winning or losing contracts/bids, legal disputes, and nancial warnings.
8

29

each announcement of security i. Let be used to index returns throughout the event period, where = 0 represents the announcement date for an event, = T1 to = T2 is dened to be the event window, and = T0 to = T1 constitutes the estimation window. And let L1 = T1 T0 and L2 = T2 T1 represent respectively the period length of the estimation window and the event window. The estimated abnormal return is ARi, = i, = Ri, ( i + i Rm, ), (3)

where i and i are the ordinary least squares estimates of i and i and are calcu lated as follows,
T1

i =

T0

(Ri, i )(Rm, m )
T1 T0

(4) (Rm, m )2

i = i i m
T1

(5)

i = 2

1 L1 2

(Ri, i i Rm, )2 .
T0

(6)

For period , the average abnormal return for N events is 1 AAR = N

N

ARi, .
i=1

(7)

The cumulative average abnormal returns are then computed by aggregating over the event window = T1 to = T2 ,

30

CAAR(1 , 2 ) =
=1

AAR ,

(8)

Assuming no clustering of events, statistical inferences about the CAARs can be examined by testing the null hypothesis that the abnormal returns dier signicantly from zero. Their signicance can be tested by using the following t-statistic, CAAR(1 , 2 ) var(CAAR(1 , 2 )) 2
1

tCAAR =

N (0, 1).

(9)

4.3

Abnormal Volume

In addition to the stock return data, trading volume surrounding news announcements can also provide evidence of suspicious insider-trading activities. In the market microstructure literature, high trading volumes are associated with the release and reception of information. To investigate the trading volume eects, I test for evidence of unusual trading volume prior to the event window. If one observes a relatively high fraction of total trading volume change within the event period prior to a news announcement, one might associate such a phenomenon with the exercise of their knowledge of private information by insiders. In the absence of early leakage of insider information, the corporate news announcements are expected to aect the volume of trade, and one would expect to see a big spike around the event date. In order to investigate this hypothesis, I rst oset the non-normality of trading volume around an event by dening it to be ln 1 + V olume.9 The daily natural log
Trading volume around an event is generally found to be positively skewed. See, for example, Harris (1986).
9

31

volume for each security i at time t is thus, ln(1 + V olumei,t ) = ln(1 + V olumem,t ) + ln(1 + V olumei,t1 ) + ln(1 + V olumei,t2 ) +i M ondayi,t + i T uesdayi,t + i W ednesdayi,t +i T hursdayi,t , (10)

where V olumem,t is the market turnover volume at time t, V olumei,tn is the lagged volumes, and M ondayi,t ,..., are the weekday dummy variables which equals one for rm i if the trading took place on that day and zero otherwise. The abnormal volume AVi, is then calculated to be

AVi, = ln(1 + V olumei,t ) ln(1 + V olumem,t ) ln(1 + V olumei,t1 ) ln(1 + V olumei,t2 ) i M ondayi,t i T uesdayi,t i W ednesdayi,t i T hursdayi,t . (11)

Since we can assume that the trading volume, once it has been adjusted for rst order serial correlations and the day-of-the-week eects, is approximately normally distributed, we can apply the same signicant tests for the abnormal returns to the abnormal volumes.

32

5
5.1

Empirical Results
Evidence of Pre-Event Stock Price Run-Ups or DownwardDrifts?

The cumulative average abnormal returns (CAAR1 ,2 ) for good and bad news announcements for days -30 to +30 are shown in Figures 1 and 2. The trends of (CAAR1 ,2 ) for good news announcements depict a more informative picture of suspicious price changes prior to the events. A price run-up is evident as early as day -15 for China-aliated stocks, increasing sharply around day -10 and reaching almost 10% at the announcement date. In contrast, both Hong Kong and U.S. stocks exhibit no such suspicious run-ups prior to good news announcements, signalling market eciency for both groups of stocks. Tables 2 and 3 present the t-statistics for all the CAARs of dierent announcement types over the study period from 1994 to 2000. With the exception of non-corporate events, the t-statistics for the entire event window for all categories in the good news group are signicant at least at the 10% level, while similar forceful results are not as apparent among bad news announcements with no denitive downward-drifts prior to the event date. Within the pre-event window for the good news samples, only China-aliated stocks indicate signicantly positive CAARs, around 14% for earnings announcements and 7% for corporate announcements. The U.S. stocks do show positive CAARs but these are statistically insignicant. If there does exist questionable insider-trading on non-public information and its leakage prior to news announcements, one should expect to observe positive CAARs in the pre-event window for good news and negative CAARs for bad news.

33

Figure 1: Good News Announcements

34

Figure 2: Bad News Announcements

35

Table 2: Cumulative Average Abnormal Returns for Good News Announcements

Event Window Period Pre-event Post-event 1.62% (1.37) 0.35% (0.23) 14.21% (3.75) -2.20% (-1.74) -4.25% (-1.63) -9.80% (-1.47) 0.55% (0.37) -0.10% (-0.06) 6.54% (2.25) 0.40% (0.51) -0.51% (-0.50) 5.93% (2.63) 3.84% (3.18) 5.69% (4.01) 10.03% (2.63) 1.62% (1.27) 2.16% (0.82) 5.85% (0.85) 4.62% (3.05) 4.56% (2.68) 12.43% (4.25) 3.64% (4.56) 4.69% (4.56) 10.46% (4.59)

Announcement Type
Earnings United States Hong Kong China-aliated Non-corporate United States Hong Kong China-aliated Corporate United States Hong Kong China-aliated All Types United States Hong Kong China-aliated

All 5.95% (3.48) 5.95% (2.89) 24.17% (4.45) -0.60% (-0.33) -2.38% (-0.63) -3.50% (-0.36) 5.12% (2.38) 0.65% (4.92) 20.57% (4.92) 4.22% (3.73) 4.96% (3.38) 17.25% (5.32)

Pre-event window period refers to a 30-day period prior to the news announcement. Post-event period refers to a 30-day period after the announcement. Numbers in parentheses are the test statistics. Signicant at the 10% level. Signicant at the 5% level.

36

Table 3: Cumulative Average Abnormal Returns for Bad News Announcements

Event Window Period Pre-event Post-event -2.57% (-2.11) 1.15% (0.80) 2.89% (1.06) 2.26% (1.04) -2.39% (-0.77) -0.37% (-0.05) 0.14% (0.11) 2.75% (1.25) 4.06% (1.25) -0.05% (-0.65) 1.26% (1.20) 3.07% (1.51) -1.82% (-1.46) -5.42% (-3.76) -10.12% (-3.67) -1.38% (0.62) -6.82% (-2.21) 7.82% (1.12) -2.03% (-1.62) -7.16% (-4.20) -3.01% (-0.91) -1.29% (-1.55) -6.26% (-6.02) -5.68% (-2.80)

Announcement Type
Earnings United States Hong Kong China-aliated Non-corporate United States Hong Kong China-aliated Corporate United States Hong Kong China-aliated All Types United States Hong Kong China-aliated

All -4.45% (-2.53) -4.14% (-2.01) -7.78% (-1.98) 3.32% (1.06) -9.60% (-2.17) 6.16% (0.60) -1.02% (-0.57) -5.18% (-2.12) 2.32% (0.49) -1.54% (-1.31) -5.29% (-3.55) -2.52% (-0.87)

Pre-event window period refers to a 30-day period prior to the news announcement. Post-event period refers to a 30-day period after the announcement. Numbers in parentheses are the test statistics. Signicant at the 10% level. Signicant at the 5% level.

37

The signicantly positive CAARs for China-aliated stocks listed in Hong Kong directly signal suspicious insider-trading activities prior to the release of news events. Another method for depicting the abnormal price changes prior to a news announcement is to calculate the fraction of the CAAR on dierent days prior to the good news announcements over the CAAR at the end of the event window. Depicted in Figures 3, 4, and 5, this measure displays the proportional magnitude of the total abnormal price changes that occur before the event. Overall, the China-aliated rms exhibit the most striking results, consistent with the observation of the signicantly positive pre-event CAARs. For good earnings events, more than 50% of the total CAAR occurs ve days before the actual announcement, and around 16% for corporate news announcements. It is a surprising observation that over 10% of the total CAAR of the U.S. stocks arises as early as 15 days prior to the event date for both good earnings and corporate news announcements. Hong Kong stocks also exhibit a similar trend for good earnings events. This may imply the possibility for both U.S. and Hong Kong rms to have certain information leakages before actual announcements; however, their CAARs in the pre-event windows are shown to be insignicant. The dierence in means tests between the CAARs of U.S., Hong Kong, and China-aliated rms can provide a clearer portrait of the relative dissimilarities in each markets degree of eciency. Table 4 displays the t-statistics obtained from dierence in means tests between the three groups for both good news and bad news events.10 For good news events, the CAARs of China-aliated rms seem to express a much greater relative degree of abnormality in returns prior to news
10

Dierences in means t =

x1 2 x
(n1 1)s2 +(n2 1)s2 1 2 n1 +n2 2 1 [n 1 1 +n 2

, where s2 = s2 i CAAR N .

38

Figure 3: Good Corporate News Announcements

39

Figure 4: Good Earnings Announcements

40

Figure 5: Good News Announcements

41

events, consistent with previous observations. This constitutes stronger evidence that, compared to Hong Kong and U.S. rms, the China-aliated counterparts are more embroiled in suspicious insider-trading practices. Table 4: Dierences in Means for Cumulative Abnormal Returnsa
Dierence Hong Kong China-aliated United States China-aliated United States Hong Kong
a These

Good News -3.78 -4.51 -0.41

Bad News -0.93 0.37 1.98

are cumulative average abnormal returns in the pre-event window period from day -30 to day -1.

5.2

Abnormal Volume

Past literature has documented the development of high trading volumes surrounding the arrival of new information.11 Such phenomenon is particularly evident for the Red-Chips and H-shares. The buildup of China-aliated stocks CAARs in the pre-event window for good news announcements is paralleled by a dramatic increase in trading volume, lending further support to the insider information leakage hypothesis among the Red-Chips and H-shares. Table 5 reports the t-statistics of the cumulative average abnormal volumes, CAAVs, in the pre-event window. If insiders trade on non-released information, then we would expect to see higher abnormal trading turnover volume before the announcement date. Among the three groups, only the China-aliated Red-Chips and H-shares display signicant abnor11

See, for example, Bolster, J., and M. (1992) and Kyle (1985).

42

mal volume movement prior to good news events. Figure 6 shows the CAAVs for the good news events for China-aliated stocks. It is apparent that there exists upward sloped trends of abnormal volumes starting as early as around day -15. The conclusion I draw is that the Red-Chips and H-shares seem to have an unusually great amount of trading activity prior to the actual announcement event, further strengthening the evidence of ineciency among the China-aliated stocks. Table 5: Cumulative Average Abnormal Volume t-statistics
Announcement Types Earnings United States Hong Kong China-aliated Non-corporate United States Hong Kong China-aliated Corporate United States Hong Kong China-aliated All United States Hong Kong China-aliated
Signicant at the 10%level. Signicant at the 5% level.

Good News -7.03 -0.33 3.88 0.51 -1.11 1.43 -3.98 0.16 1.73 -6.55 -0.49 4.13

Bad News -2.92 -0.34 -2.19 -2.47 2.44 3.93 -3.46 1.00 -3.23 -5.13 1.25 -2.43

43

Figure 6: China-aliated Good News Announcements

44

5.3

Impact of the 1997 Handover

I now turn to the question of whether there has been any dierences in Hong Kongs market eciency after its sovereignty handover in 1997. Tables 6 and 7 show the cumulative average abnormal returns for the pre-event, post-event and the entire event window, and their t-statistics for both before and after the Handover. To detect for suspicious trading behavior, I focus on the CAARs in the pre-event window, prior to the arrival of news announcements. As can be observed from the data, only Chinaaliated stocks show signicant CAARs for good news announcements in both before and after the Handover. I test the dierence in the means of CAARs before and after 1997 for both Hong Kong and China-aliated samples in the pre-event window. Hong Kongs dierences are insignicant, while China-aliated stocks report a positively signicant dierence for good news events, with a t-statistic of 1.37, and a negatively signicant t-statistic of -2.66 for bad news events. This indicates that there exists fewer abnormal returns in the post-Handover period for good news events, but that abnormal returns have a relative increase after 1997 for bad news events. I also examined the dierence in means tests for CAAVs. Hong Kongs dierence in means is only signicant at a 10% level for the bad news event, implying that greater abnormal volumes appear post-Handover for negative announcements. As for the Red-Chips and H-shares, they show signicant dierences at the 5% level, with t-statistics of -2.52 and -3.00 for good news and bad news respectively. Thus for China-aliated stocks, for both good and bad news events, the magnitude of abnormal volume and volatility increased after the Handover. The results, however, do not reect a strong conclusion about the impact of the Handover on the market

45

Table 6: Cumulative Average Abnormal Returns for Good News Events in Dierent Event Window Periods
Pre-event Before After 0.20% (0.14) 19.08% (3.80) -0.23% (-0.10) 8.39% (1.73) 0.28% (0.14) 3.59% (1.13) 0.17% (0.16) 8.68% (3.66) 0.84% (0.19) 9.89% (1.77) -9.27% (-1.80) -37.10% (-2.47) -0.52% (-0.19) 10.85% (1.99) -1.28% (-0.71) 2.50% (0.61) Post-event Before After 5.42% (3.90) 12.80% (2.48) 2.27% (3.90) 4.30% (0.89) 1.89% (0.95) 2.23% (0.69) 3.42% (3.12) 5.46% (2.27) 5.99% (2.31) 7.56% (1.36) 2.01% (2.31) 8.17% (0.53) 7.49% (2.65) 27.27% (5.00) 6.13% (3.39) 16.70% (4.01) All Before 5.13% (2.58) 30.42% (4.18) 1.04% (2.58) 13.79% (1.99) 4.03% (1.41) 6.18% (1.35) 4.04% (2.58) 14.12% (4.15) After 6.75% (1.81) 18.62% (2.34) -6.65% (1.81) -29.44% (-1.35) 9.11% (2.27) 41.50% (5.32) 6.00% (2.33) 21.10% (3.57)

Announcement Type
Earnings Hong Kong China-aliated Non-corporate Hong Kong China-aliated Corporate Hong Kong China-aliated All Types Hong Kong China-aliated

46

Before refers to the sample period from January 1994 through June 1997, and After refers to a sample period from July 1997 through December 2000. t-statistic signicant at the 5% level. t-statistic signicant at the 10% level.

Table 7: Cumulative Average Abnormal Returns for Bad News Events in Dierent Event Window Periods

Announcement Type
Earnings Hong Kong China-aliated Non-corporate Hong Kong China-aliated

Pre-event Before After 0.49% (0.33) -4.29% (-1.51) -1.72% (-0.68) 0.95% (0.08) 4.90% (2.54) -0.54% (-0.14) 1.71% (1.59) -2.75% (-1.22) 1.75% (0.73) 11.67% (2.34) -2.92% (-0.56) -0.89% (-0.10) 1.21% (0.47) 6.96% (1.47) 0.09% (0.53) 8.05% (2.50)

Post-event Before After -1.99% (-1.31) -9.80% (-3.43) -4.53% (-1.31) -2.87% (-0.25) -4.23% (-2.18) 1.18% (0.49) -3.11% (-2.86) -5.55% (-2.43) -8.51% (-3.58) -10.51% (-2.09) -8.65% (-1.68) 12.09% (1.40) -9.25% (-3.60) -6.05% (-1.26) -8.82% (-5.30) -5.80% (-1.80)

All Before -1.97% (-0.92) -13.75% (-3.37) -6.42% (-1.77) -2.01% (-0.12) 0.04% (0.13) 1.17% (0.21) -1.18% (-1.15) -8.13% (-2.51) After -6.09% (-1.79) -0.48% (-0.67) -12.15% (-1.64) 9.42% (0.73) -9.14% (-2.48) 3.05% (0.45) -8.14% (-3.41) 2.29% (0.50)

47
Corporate Hong Kong China-aliated All Types Hong Kong China-aliated
t-statistic signicant at the 10% level.

Before refers to the sample period from January 1994 through June 1997, and After refers to a sample period from July 1997 through December 2000. t-statistic signicant at the 5% level.

eciency of Hong Kong.

5.4

Sensitivity to Good vs. Bad News

The observation of fewer signicant t-statistics for bad news announcements may be explained by the hypothesis that good news disseminates at a faster rate than bad news. I hypothesize that it is much more likely for insiders to trade on and disseminate good news than bad news since advantages can be more easily reaped from it. If an insider is in possession of bad news unknown to the public, it is probable that s/he is more inclined to withhold that information, resulting in less leakage prior to the actual announcement date. To test the dierence of the two information types, I conduct a dierence of means test between good and bad news to see if prices are more sensitive to one than the other. The dierence is positively statistically signicant for all three groups at a 10% level, with a t-statistic of 3.489 for U.S. stocks, and t-statistics of 4.900 and 1.743 for Hong Kong and China-aliated stocks respectively. The results imply that price changes are signicantly more responsive to good news than bad news. This observation is consistent with the ndings presented by Fung and Poon (2000) that the China-aliated stocks show more sensitivity to good news than bad news. I then conduct a dierence in means test is also conducted on cumulative average abnormal volume between good and bad news. For all three stock groups, the dierences are found to be signicant. Both Hong Kong and China-aliated stocks are positively signicant at the 10% level with t-statistics of 1.806 and 7.338 respectively, representing the greater eect of good news on volume volatility surrounding

48

news announcements. On the other hand, the U.S. has a negatively signicant tstatistic of -2.178 at the 10% level, which points to the opposite of volume being more sensitive to bad news than good news. I conclude, therefore, that, relatively, China-aliated stocks are the most sensitive to good news events.

5.5

Clustering

My analysis in examining aggregate abnormal returns has assumed that there is no cross-sectional correlation or clustering within the securities event windows. This made it possible to compute the variances of the sample CAARs without taking covariances into consideration since they will be zero under the assumption. To accommodate the issue of clustering is beyond the scope of this study. Although this paper has not accounted for clustering, it is extremely unlikely that the results documented here will be rendered invalid since my sample is subdivided into different specic announcement type groups, which should have lessened the eect of clustering, and furthermore, the reported outcomes remain statistically signicant among the subdivided groups. There are several methods to adjust for the clustering eect. One way is to form portfolios of securities that are dated by the day of news announcements. This method allows for cross-sectional dependence among the abnormal returns, and the procedure outlined in the study design section on computing abnormal returns and their inferences can be used. Another way to solve the clustering issue is to avoid aggregation and to examine each individual abnormal return, as Bhattacharya, Daouk, Jorgenson, and Kehr (2000) did in a recent event study. Other methods have been developed by Collins and Dent (1984) and Schipper and Thompson (1983).

49

Conclusion

Using a data set of corporate news announcements in Hong Kong and the U.S. from January 1994 to December 2000, this study investigates whether there is evidence of suspicious illegal insider-trading activities by examining the behavior of abnormal prices changes and uctuating trade volumes. It also tries to consider how the dynamics of securities listed on the Stock Exchange of Hong Kong have been affected by the transfer of sovereignty, while addressing the dierence in sensitivity of securities to good and bad news events. Since Hong Kong regulations on insider trading, in comparison with the United States, are more lenient and penalties upon conviction are less stringent, it is surprising to nd that both Hong Kong and U.S. samples are shown to be relatively honest. But consistent with the concerns raised by the Hong Kong media, the Red-Chips and H-shares of China-aliated rms document abnormally signicant pre-announcement price changes and volume uctuations. As corporate insiders are the only persons with access to non-public information, the observed results in this study are attributed to illegal insider-trading activities and/or trading by outsiders to whom inside information has been leaked. In another words, there are strong signals that China-aliated stocks have been engaged in dishonest trading activities on the basis of non-public information. With their corporate parents strong ties to mainland China, enforcing the regulations on the Red-Chips and H-shares rms may prove to be a dicult hurdle for the Securities and Futures Commission of Hong Kong to cross. The ndings of the study did not, however, support the local Hong Kong medias worries about the negative impact on the integrity of Hong Kongs nancial market.

50

There exists no strong indication of changes in eciencies of the Hong Kong stocks and those with ties to the Peoples Republic of China before and after the Handover. This paper also nds that good news events seem to have a stronger eect on security prices for all three sample groups. This might be due to the fact that good news events are more probably disseminated and/or traded on by insiders than bad news events since nancial advantages can be more eortlessly attained. The method employed in this paper is only one of many ways to examine a markets eciency. I hope, however, that this study has made a small step towards a wealth of future research on the market eciency of other world nancial markets, research which could provide an invaluable picture of the relative eciency/honesty of the various world markets.

51

References
Banz, R. W. (1981): The Relationship between Return and Market Value of Common Stocks, Journal of Financial Economics, 9, 318. Bettis, C., and D. Vickrey (1997): Mimickers of Corporate Insiders who Make Large Volume Trades, Financial Analysis Journal, 53, 5766. Bhattacharya, U., H. Daouk, B. Jorgenson, and C.-H. Kehr (2000): When an Event is not an Event: the Curious Case of an Emerging Market, Journal of Financial Economics, 55, 69101. Bodie, Z., A. Kane, and A. J. Marcus (1999): Investments. Irwin/McGrawHill. Bolster, P. J., K. A. J., and P. J. M. (1992): Merger Announcements, Asymmetrical Information, and Trading Volume: An Empirical Investigation, Journal of Business Finance and Accounting, 19, 901910. Cahn, S. Y., D. J. P., and G. D. Donleavy (1989): Economic Eciency and Stock Market Unication, Securities Bulletin, Stock Exchange of Hong Kong, 37, 3034. Carlton, D., and D. Fischel (1983): The Regulation of Insider Trading, Stanford Law Review, 35, 857895. Chang, S., and D. Y. Suk (1998): Stock Prices and the Secondary Dissemination of Information: The Wall Street Journals Insider Trading Spotlight Column, Financial Review, 33, 115128.

52

Cheung, Yan-Leung, W. M., and L. Wu (2000): Insider Trading in the Hong Kong Stock Market, Asia-Pacic Financial Markets, 7, 275288. Collins, D., and W. T. Dent (1984): A Comparison of Alternative Testing Methodologies Used in Capital Market Research, Journal of Accounting Research, 22, 4884. Cornell, B., and E. R. Sirri (1992): The Reaction of Investors and Stock Prices to Insider Trading, Journal of Finance, 47, 10311059. Elliot, John, M. D., and G. Richardson (1984): The Association between Insider Trading and Information Announcements, Rand Journal of Economics, 15, 521536. Eysell, T. H., and N. Arshdai (1993): Insiders, Outsiders, or Trend Chasers?: An Investigation of Pre-Takeover Transactions in the Shares of Target Firms, Journal of Financial Review, 16, 4959. Fama, E. (1965): The Behavior of Stock Market Prices, Journal of Business, 38, 34105. Finnerty, J. E. (1976): Insiders and Market Eciency, Journal of Finance, 31, 11411148. Fung, H.-G., and W. Poon (2000): Red Chips or H-Shares: Which ChinaBacked Securities Process Information the Fastest?, Journal of Multinational Financial Management, 10, 315343. Givoly, D., and D. Palmon (1985): Insider Trading and the Exploitation of Insider Information: Some Empirical Evidence, Journal of Business, 58, 6987. 53

Harris, L. (1986): Cross-Security Tests of the Mixture of Distributions Hypothesis, Journal of Financial and Quantitative Analysis, p. 39. Ho, B. M. (1998): Public Companies and their Equity Securities: Priniciples of Regulation under Hong Kong Law. Klumer Law International, London. Jaffe, J. F. (1974): Special Information and Insider Trading, Journal of Business, 47, 410428. Kabir, R., and T. Vermaelen (1996): Insider Trading Restrictions and the Stock Market: Evidence from the Amsterdam Stock Exchange, European Economic Review, 40, 15911603. Kwong, K. S., and K. A. Wong (1984): The Behavior of Hong Kong Stock Prices, Applied Economics, 16, 905917. Kyle, A. (1985): Continuous Auctions and Insider Trading, Econometrica, 53, 13151335. Langevoort, D. C. (2001): Insider Trading: Regulation, Enforcement and Prevention. West Group. Lorie, J., and V. Neiderhoffer (1968): Predictive and Statistical Properties of Insider Trading, Journal of Law and Economics, 11, 3351. MacKinlay, A. C. (1997): Event Studies in Economics and Finance, Journal of Economic Literature, 35, 1339. Manne, H. G. (1966): Insider Trading and the Stock Market. The Free Press, New York. 54

McGuinness, P. (1999): A Guide to the Equity Markets of Hong Kong. Oxford University Press (China), Hong Kong. Meulbroek, L. (1992): An Empirical Analysis of Illegal Insider Trading, Journal of Finance, 47, 16611699. Oppenheimer, H. R., and T. E. Dielman (1985): Firm Dividend Policy and Insider Activity: Some Empirical Results, Journal of Business Finance and Accounting, pp. 525541. Penman, S. (1982): Insider Trading and the Dissemination of Firms Forecast Information, Journal of Business, 55, 479504. Pratt, S., and C. DeVere (1970): Relationships between Insider Trading and Rates of Return for NYSE Common Stocks, 1960-1966, in Modern Development in Investment Management, ed. by J. Lorie, and R. Brealey, pp. 259270. Praeger. Reinganum, M. R. (1981): Misspecication of Capital Asset Pricing: Empirical Anomalies Based on Earnings Yields and Market Values, Journal of Financial Economics, 9, 1946. Rozeff, M. S., and M. A. Zaman (1988): Market Eciency and Insider Trading: New Evidences, Journal of Business, 61, 2544. Schipper, K., and R. Thompson (1983): The Impact of Merger-Related Regulations on the Shareholders of Acquiring Firms, Journal of Accounting Research, 21, 184221. Seyhun, H. N. (1986): Insiders Prots, Costs of Trading, and Market Eciency, Journal of Financial Economics, 16, 189212. 55

(1992): Why Does Aggregate Insider Trading Predict Future Stock Returns?, Quarterly Journal of Economics, 107, 13031332.

56