Corporate Governance, Regulatory Compliance and Insider Trading

Transcription

1 Corporate Governance, Regulatory Compliance and Insider Trading Abstract This paper investigates the role of corporate governance on regulatory compliance and the performance of Italian corporate insider trading. Exploiting a unique enforcement and reporting framework, we present three main findings. First, corporate governance does not influence the propensity of firms to comply with insider trading regulation. Firms that have concentrated ownership and control, together with those run by families and cooperatives are most likely to comply with regulation. Second, multiple trading, but not transaction size, affects the performance of corporate insider trading. Third, the market responds less to reporting by compliant and family firms. JEL Classification: G14, G38 Keywords: Corporate Governance, Corporate Insider Trading, Regulation, Italy; Family Firms - 1 -

2 INTRODUCTION In the wake of recent corporate scandals in the US, Europe, and South East Asia, there has been a concerted effort by regulatory authorities to improve the overall corporate governance culture of firms. However, external and internal monitoring of firm management can only be effective so long as enforcement measures are meaningfully implemented and managers themselves are willing to cooperate with such disciplinary regimes. The reporting of corporate insider trading is an area where management and regulator objectives may be in opposition and, therefore, is a good subject to test the effect of governance on regulatory compliance. In addition, the sensitivity of market participants to the reporting of insider trading by firms who show different degrees of compliance can provide much insight into the nuances of market efficiency. In recent years, considerable interest has focussed on the relationship between corporate governance, firm behaviour and market performance [see for example, Hermalin and Weisbach (1991), Yermack (1996), Core, Holthausen, and Larcker (1999), Gompers, Ishii, and Metrick (2003), and Anderson, Mansi and Reeb (2004)]. Research has also investigated how new regulation impacts upon the behaviour of managers [Dahya, McConnell and Travlos (2002)]. Surprisingly, very little work has examined whether corporate governance affects the propensity of firms to comply with regulatory regimes and if this affects the market reaction to corporate events. The arrival of the Sarbanes-Oxley Act has motivated many researchers to examine the role of regulation on corporate behaviour and performance [Cunningham (2003), Romano (2004), Chhaochharia and Grinstein (2007), De Franco, Guan and Lu (2005), and Zhang (2005)]. The US is fairly unusual in that - 2 -

3 recent enforcement of punishments relating to non-compliance of regulation have been diligently implemented. However, in most other countries, regulatory authorities have much less success in ensuring firms comply with regulation and non-compliance within countries is common. Italy is a particularly interesting case to study the effects of regulatory compliance. Whereas there is quite extensive regulation in place, which governs the trading behaviour of corporate executives, non-compliant firms and executives are very rarely punished with meaningful penalties. Italian firms can fully adhere to the regulation and bear the concomitant transparency and administrative implications. They may also partially comply, or completely ignore the regulation and experience penalties of varying magnitude. Our paper makes three main contributions to the literature. First, we investigate whether corporate governance influences the degree to which firms comply with regulation. A fundamental rationale for improving corporate governance is the implicit belief that better governance leads to improved and more accountable decision making. This is also the assumption underlying most corporate governance-firm performance studies [see, for example, Core et al. (2006)]. Regulatory compliance is taken as given, yet if corporate governance does not improve the chances of firms adhering to regulation, its link with performance, accountability and transparency is weak and possibly inappropriate. Our second contribution is to examine the sensitivity of the market to corporate news (in this case, corporate insider trading activity) in the presence of different degrees of regulatory compliance. Semi-strong form tests of market efficiency assume that every company equally complies with regulation. An interesting question is whether the market reacts differently to corporate events in - 3 -

4 firms that a) are fully compliant and promptly announce every corporate event (insider trade); b) are partially compliant and only selectively announce corporate events (insider trades); or are c) non-compliant and make no prompt announcements whatsoever. Traditional market efficiency theory predicts that fully compliant firms would elicit the greatest market reaction to corporate announcements given that information quality in these firms is highest [see Easley, Hvidkjaer and O Hara (2002)]. However, a counter-argument can be given that firms with few announcements would have a stronger response to new information given that the information set is poor prior to the announcement [Kim and Verrecchia (1997)]. Finally, we examine whether corporate governance within firms has an impact on the performance of corporate insider trading. The conventional view that corporate governance improves accountability and transparency naturally leads to the conclusion that corporate insiders in firms with good governance are less likely to earn excess profits from their personal trading activities. However, given that the link between corporate governance, regulatory compliance, and corporate accountability has yet to be proven, a study of corporate insider trading performance conditional on these factors provides valuable insights into the debate. Our results are as follows. We find that the corporate governance and ownership characteristics of firms have a definite impact on the extent of regulatory compliance. Family firms with high levels of ownership and control and cooperative banks are more likely to comply with the insider trading regulations. Firms that have a greater need for non-executive directors are more likely to be noncompliant with the regulation. Compliance does not appear to be affected by firmlevel characteristics such as size, leverage, or book to market equity

5 Broadly consistent with prior corporate insider trading studies, executives earn significantly positive cumulative abnormal returns from their buy trades and avoid significantly negative CARs after their sell trades. In addition, multiple insider buy and sell trades are a stronger signal to the market than single isolated cases of insider trading. Buying activity in full and partially compliant firms is associated with positive abnormal returns in contrast to non-compliant firms, which elicit no market reaction to insider trades. Moreover, selling activity is only significantly negative in fully compliant firms. These results are strongly supportive of market efficiency theory that posits increased price sensitivity for better quality information announcements [Easley et al. (2002)]. Of major interest is whether corporate governance and ownership characteristics, together with the degree of regulatory compliance by firms to the Italian corporate insider trading restrictions, has any influence on the performance of corporate insider trades. Firms that promptly disclose insider buying activity, particularly by chief executives, to the market are more likely to experience positive price responses. In addition, companies with the greatest need for external monitoring, i.e. those with higher numbers of non-executive directors, have strong positive price reactions to insider buying activity. Buying activity by insiders in family firms elicits the opposite effect with a significantly lower share price reaction. All other corporate governance characteristics, such as a split role for the chief executive and chairman, or compliance with the insider trading regulation codes, have no impact on the performance of corporate insider buying activity. In the next section, the Italian regulatory reporting environment for corporate insider trading is described in detail. In Section II, we develop the hypotheses that - 5 -

6 are tested in the paper. Section III presents the data and methodology and Section IV discusses the empirical results. Section V concludes the paper. 1. REGULATORY FRAMEWORK During our sample period, Italian regulation on corporate insider trading was quite distinct from that of other countries, including the US, in the expectation it placed on company executives to disclose their trading activity. Unlike most other countries, Italy had a differential reporting system for insider trading ranging from periodic to prompt disclosure depending on the scale of trading activity. The very first Italian insider trading disclosure regulation was introduced on 1 January 2003, and was part of a broader regulation ( Regolamento ) affecting companies listed on the Italian Stock Exchange. The regulation required firms to report and update a list of persons who could have access to private information and periodically disclose trades made by such persons. The code also identified blackout periods when insiders were forbidden to trade (typically before approval of the financial statements or quarterly reports). In addition, all insiders were to declare the exercise of stock options or pre-emptive rights. According to the Italian Exchange rules (article 2.6.4), listed firms had to respect two kinds of disclosure requirements concerning insiders trading in their own company s securities. Depending on the total trade size, trades needed no disclosure (total transactions below 50,000), had to be disclosed quarterly (total transactions between 50,000 and 250,000), or promptly within three days of the transaction (total transactions greater than 250,000). The exchange also allowed - 6 -

7 firms to declare their own lower threshold amounts and more stringent deadlines if they so wished. These disclosure requirements are significantly different than any other regulation that is in place in other developed countries. By far the most common type of disclosure rule requires insiders to promptly notify their company of any trading activity they have undertaken (normally within five days). The onus is then on the firm to notify the relevant regulatory authority within a short time period. By having different regulatory regimes in place for the same insiders and the same market, much insight can be gained into the way that the market reacts to different levels of information quality. Clearly trades that are not disclosed promptly (but through the quarterly announcement) are private information whereas trades that are promptly disclosed are public information. Firms that did not comply with the above rules could be sanctioned by the Italian Exchange in one of three ways: a private warning; a public warning; or a public warning plus a fine ranging from 5,000 to 100,000 for each instance of non-disclosure that broke the regulatory requirements (article 2.6.9). The number of private warnings given to Italian firms over the sample period is obviously unknown. It is interesting to note that there were no public warnings or fines imposed on non-compliant firms until September It was recognised by the Italian authorities that non-compliance with the insider trading regulations could also be the fault of insiders not communicating relevant trades to the firm. Non-complying insiders could only be sanctioned by the listed company in the ways set by the firm s internal dealing code. On this issue, the Italian Stock Exchange advised firms to adopt such possible sanctions as a written warning, a pecuniary fine equal either to a percentage of the undisclosed transaction, - 7 -

8 a fixed amount, a refund of the expenses, or damages directly or indirectly borne by the firm. 1 In most cases, the actual penalties imposed on non-compliant firms or insiders did not reflect the gains that were made from the insider s trading activity. An insightful example is that of Emilio Gnutti, director of Unipol, the Italian insurance company. Between 1 September 2003 and 30 January 2004, Mr. Gnutti bought 12.7 million Unipol preferred shares, equal to approximately 3.9% of the total outstanding preferred equity. Italian regulation required that he promptly disclose the trades, which he did not do. When the trades were eventually disclosed, it transpired that Mr. Gnutti was up to 42 days late with the disclosure and was fined 150,000. This only partially offset the gross capital gain made from his trading, which was equal to 4.42 million. 2 The Italian regulation of scaled disclosure leads to some interesting issues that can be explicitly examined. Trading is required to be disclosed promptly, at the quarterly announcement, or not at all. The assumption underlying this rule is that small trades lack information and total transaction size is the only metric by which to gauge information quality. This view is championed by the theoretical models of Kyle (1985) and Admati and Pfleiderer (1988) who show that insiders will seek to take advantage of their informational advantage by trading as much as they can before prices move. Empirically, it has been shown that transaction size alone is unable to predict information quality with other proxies, such as proportionate 1 The regulation changed on 1 April 2006, which required prompt reporting of trades (within 5 days of the transaction) greater than 5,000. In addition, the group of individuals who can be classified as insiders was increased to anyone who held a stake greater than or equal to 10% of a company s outstanding share capital. 2 The issue was widely covered in the Italian press at the time. For more detailed information, see Il Sole 24 Ore, 3 and 5 February, 2004, and 6 May 2004; and Unipol s press release of April 29 th, Formattato: Inglese (Regno Unito) - 8 -

9 shareholdings or multiple trading signals [Hillier and Marshall (2002)], providing better predictive performance. Given the lax enforcement regime, many insider trades will be noncompliant with the disclosure requirements of the Italian regulation. This will lead to the complex situation whereby market participants not only have to assess the information or signal in disclosed insider trading activity but also place this in the context of wider reporting compliance within the firm. Market observers will be uncertain whether companies are reporting all their trades or only partially disclosing trading activity, and whether the selective disclosure pertains to informative or uninformative trades. 2. HYPOTHESES 2.1 Determinants of regulatory compliance The most recent and profound case of regulation in the US is the Sarbanes-Oxley act, which places a considerable burden on the disclosure activities and accountability of management. Enforcement of the act has been diligent and many senior executives have experienced the full brunt of the punishment associated with non-compliance. It is easy to understand why most firms are complying with this regulation [Zhang(2005)]. The regulatory environment in most countries lacks the same power of enforcement as it does in the US. Many firms flout disclosure regulations and have little fear of receiving punitive penalties. The Italian corporate environment, which is similar to most countries in the developed and developing world, is also quite different from the US. In Italy, most firms have high levels of ownership and control concentrated in small blocks of family units [La Porta et al. (1999), Mengoli et al - 9 -

10 (2006)]. This can have markedly different impacts on the objectives of these firms because of the altruistic pressures placed on managers to consider wider and nonaffiliated stakeholder groups [See Anderson and Reeb (2003), Burkart et al. (2003), and Villalonga and Amit (2006)]. Surprisingly, while many studies have investigated the impact of new regulation [Garfinkel (1997), Eleswarapu, et al. (2004), Grount and Zalewska (2006)], very little research has gone into the determinants of regulatory compliance [Healy and Palepu (2001)]. Some systematic factors have come to light which provide a degree of insight into the area. In the US, firm size, external financing needs, ownership structure, and profitability have all been shown to increase the likelihood that firms will comply with voluntary disclosure regulation [Lang and Lundholm (1993, 2000), Healy and Palepu (2001), Bushee and Leuz (2005)]. The general wisdom is that smaller firms with more concentrated ownership and lower external financing needs will more likely be non-compliant. In corporate environments with very strong, extensive family ties and business networks, this is unlikely to be the case. Where improved disclosure brings positive liquidity and value benefits, the convergent objectives of management and shareholders would make it more likely that family-run firms would be compliant with regulation. In addition, managers in larger firms, where the principal-agent problem is more apparent, are in the case of lax regulatory regimes more likely to pursue their own pecuniary objectives than consider those of shareholders. Contradictorily then, larger and more widely held firms, with higher numbers of non-executive directors, may be associated with less compliance with regulation in these environments. This is especially true where the independence of nonexecutive directors is in doubt [Santella et al. (2007)]

11 Notwithstanding the effects of families and wider business networks on the propensity for firms to comply with regulation, many have a compliance infrastructure explicitly developed by the company. With respect to the Italian insider trading regulation, firms may adopt several policies. Specifically, the company may have a) voluntarily adopted the regulation prior to its mandatory enforcement in January 2003; b) introduced one or more blackout periods when insiders are forbidden from trading; c) required the prompt disclosure of stock option exercises; d) had more stringent disclosure requirements; or e) published the internal compliance policy on the company website for external verification. In an environment where managers are held fully accountable by their actions, all of these factors would be expected to increase the propensity of firms to comply with disclosure regulation. H1: Family firm status and high levels of ownership and control will increase the probability of regulatory compliance. H2: Larger firms with more non-executive directors will decrease the probability of regulatory compliance. H3: Firms with a stronger voluntary disclosure regulation will more likely be compliant. 2.2 Information quality and determinants of insider trading performance An interesting feature of the first Italian insider trading regulation is that it assumes that trade size alone reflects information content. Given that corporate insiders have

12 finite levels of wealth and varying capacities to trade, absolute transaction size is not an efficient measure of information quality. Finance theory lacks a definitive argument on the most important trade characteristics that reflects superior information. An alternative proxy for information can be that of multiple trading signals [(Easley and O Hara, 1987, Bagnoli and Khanna, 1992] where more than one insider trades within a particular period or one insider makes more than one trade. Multiple trading signals are consistent with the models of John and Mishra, 1990, John and Lang, 1991, and Bagnoli and Khanna, 1992, who argue that corporate insiders use their trades as signals to market participants to increase the efficiency of prices. H4: Multiple trading signals rather than total transaction size will affect the performance of corporate insider trades. Existing research has examined the impact of job role within the firm [Seyhun (1986), Lin and Howe (1990)], ownership structure [Fidrmuc et al. (2006)], firm value [Lakonishok and Lee (2001)], analyst following [Penman (1984)], and multiple trading signals [Hillier and Marshall (2002), Fidrmuc et al. (2006)] on the performance of corporate insider trades. However, to date, the role of compliance has not been examined. There are many reasons to expect that these factors will be significant determinants of insider trading performance. Fidrmuc et al. (2006) has argued that external monitoring by blockholders would be expected to reduce the performance of insider trading. It is natural to extend this argument to other forms of corporate governance or to the level of firms regulatory compliance

13 H5: The level of firm s compliance will affect the performance of corporate insider trades. 2.3 Determinants of market reaction to insiders trades Within the context of the present study, firms can be compliant, partially compliant or non-compliant. In addition, within each firm, insiders can be compliant, partially compliant, or non-compliant with the regulation. It is important to emphasise the difference between insider and firm compliance. It is possible that a firm may have very good compliance structures but an insider, whether accidentally or proactively, does not comply fully with the disclosure regulation. It would be expected in this situation that the market would react differently to insider trades in each of the different scenarios. However, the empirical impossibility of disentangling the confounding influences of insider versus firm compliance means that we only focus on aggregate firm s compliance in our analysis. Where disclosure improves the information set within markets, compliance with regulation that mandates the prompt reporting of material corporate events will lead to an improvement in the quality of information. In markets with good information quality, share prices respond quickly and accurately to the arrival of news [Patell and Wolfson (1984), Mitchell and Mulherin (1994), Chan (2003), Verga (2006)]. In contrast, when the quality of information is uncertain or poor, share prices will be noisy and not fully reflect the information content of news. From this perspective, the market will respond more to firms that are fully compliant with disclosure regulation. However, an opposite view, proposed by Kim and Verecchia (1997), is that the arrival of new information, which substantially improves the quality of information in a market, will lead to significant price

14 changes. This is because market expectations change discretely in line with the arrival of new but infrequent events. H6: The market reaction to insider trading will be different between compliant and non- compliant firms. Insiders belonging to family firms should be less inclined to exploit information advantages by trading relatively small amount of shares. In fact, the potential profit from the exploitation of private information in small insider trading activity would be more than offset by the loss of reputation. At the same time, non-family insiders will be likely monitored by family members, reducing the chance to observe information-based transactions. These arguments help to explain why family firms are generally more compliant to insider trading regulation and allow us to assume that trades made by family firms insiders carry lower information content. H7: The market reaction to insider trading will be lower for family firms. 3. DATA AND METHODOLOGY The insider trading data in our analysis consists of 5,853 trades reported on 748 filings sent to the Italian Exchange during 2003, the first year of internal dealing regulation. These were made by 411 insiders from 146 listed firms 3. Each filing contains the name of the company, the name of the director or insider who traded, the date of the transaction, the date of the public filing, i.e. the announcement date, the number of shares traded and the transaction price. 3 Our 146 firms represent the whole universe of Italian listed firms who reported at least one internal dealing filing in 2003 and the 5,853 trades represent the total number of declared insiders trades made in The total population of Italian firms during the period was

15 From the initial sample, we exclude 106 trades on derivatives or hybrid securities, such as convertible bonds, options and pre-emptive rights; 151 trades on firms whose fiscal year did not match 2003; and 8 trades with missing information either on the transaction price or on the type of trade. The final sample therefore consists of 5,588 observations of which 3,062 trades are insider buys and 2,526 are insider sales. Trades made on the same trading date are cumulated and netted to have a single day net buy or sell observation. After cumulation and netting the final sample is 2,728 transactions split between 1,713 net buy and 1,015 net sales. Aware that relevant insiders trades may not have been reported to the Italian Exchange, we also collect evidence of non-compliance to the law. Italian listed companies financial statements must report directors holdings as well as their change from the previous year (article 79, Consob rule 11971/99). We are thus able to identify insiders who changed their holdings but did not disclose trades in the required fashion. From a manual checking of every Italian firm s financial accounts in 2003, we recorded 1,471 changes in designated insider holdings in a total of 217 firms financial statements. The transactions carried out by the same insider during 2003 are cumulated and then divided by four, to get the minimum quarterly trade that may have occurred (based on the hypothesis of equally split trades). The overall amount is then multiplied for the lowest stock price in the firm s fiscal year. If the minimum estimated traded amount results above euros, it had to be disclosed and reported. We are left with 195 cumulative insider holdings changes which certainly required disclosure according to insider trading regulation. For each insider we measure the degree of compliance to the insider trading regulation in two different ways:

16 a) a compliance ratio, which is the ratio of the number of disclosed shares traded (by all insiders belonging to a single firm) over the total number of shares traded by insiders; and b) a categorical variable equal to no compliant (no trades disclosed), equal partial compliant (only some trades are disclosed) and equal to full compliant (all trades are disclosed). Similarly, we measure a firm s compliance to insider trading regulation in two different ways: a) a compliance ratio, which is the ratio of the number of disclosed shares traded (by all insiders belonging to a single firm) over the total number of shares traded by insiders; and b) a categorical variable equal to 1 if the firm does not comply at all (no directors disclosed any of their traded shares), equal to 2 for partial compliance (only some directors partially or fully disclosed their traded shares) and equal to 3 for full compliance (all directors partially or fully disclosed their traded shares). Corporate governance, ownership, financial accounting and internal compliance structures are manually collected from each firm s set of financial accounts and share price information comes from Datastream. Full descriptions of each variable are presented in the Appendix. Company Characteristics Table 1 reports summary statistics on our sample of firms. Panels A and B present the firm characteristics of industrial and financial firms respectively and Panel C relates to governance and ownership information. Panel D provides details

17 on the various additional provisions that firms adopted with respect the insider trading regulation. < INSERT TABLE 1 HERE > In Panels A and B, the sample is split into industrial and financial firms because of the inherent difference in financial characteristics of these two groupings. The size of financial firms far exceeds that of their industrial counterparts, primarily due to the presence of very large Italian banks in the sample. Interpreting leverage and market to book ratios can be difficult for financial firms because of their unique structure. However, it can be seen that the industrial firms in the sample had higher growth opportunities with mean market to book value of 2.14 compared to 1.78 for financial firms. Operating performance during the sample period was also better with return on assets approximately double for industrial firms. In Panels C and D the full sample descriptive statistics are presented since firms from all industries are more homogeneous across ownership, governance, and compliance characteristics. Taking ownership and governance first, Panel C, the proportion of the firm s cash flow (ownership) and voting rights (control) owned by the ultimate shareholder is, as would be expected for Italian firms, very high at 41.53% and 45.16% respectfully. This compares with significantly lower values in other countries [La Porta et al. (1999), Faccio and Lang (2002)]. We also categorised firms into ownership classifications dependent on the identity of the controlling shareholder. Family firms (defined as those firms where the ultimate shareholder is a family and controls more than 10% of the votes) have by far the most concentrated ownership of over 50%, followed by government controlled firms, such as Alitalia (an airline), and institution-owned firms. Cooperative banks, which follow a one-person one-vote system, naturally have very low levels of

18 ownership concentration. Similar to many international firms, non-executive directors outnumber other directors on corporate boards. Approximately half of the population of non-executive directors are designated as independent by the firms themselves. Compliance information is presented in Panel D. While the new disclosure code was introduced in 2003, a large majority of firms (126) introduced the reporting procedures early, in 2002, with a similar number publishing the firm s insider trading code on the company website. Only 24 companies introduced a trading black-out period similar in spirit to the closed period mandated in many European countries. This, in effect, permitted Italian insiders to trade at any point during the corporate financial year and exploit potential informational advantages at certain periods (such as before an earnings announcement). Less than half of the sample required prompt disclosure of trades relating to the exercise of executive options, possibly reflecting the lack of perceived information content in these trades. The Italian regulation allowed firms to set their own disclosure thresholds for prompt and periodic reporting of insider trades. Most kept the recommended thresholds of 50,000 and 250,000 respectively. However, a substantial number tightened the disclosure requirements. Of the 146 firms who reported insider trading, 35 imposed prompt disclosure thresholds between 15,000 and 35,000. With respect to periodic disclosures, 44 firms set lower thresholds of between 50,000 and 200,000 with a mean disclosure requirement of 122,000 for this group, less than half the level of the original requirement. Insider Trading Characteristics

19 Descriptive information on the corporate insider trades is presented in Table 2. As stated earlier, Italian insider trading regulation has differential regulatory reporting systems for different trade sizes. Three trade size categories exist that determine the promptness of disclosure. For the lowest trading activity (less than 50,000 unless stated otherwise by the firm), insiders are not required to disclose their trading activity but may do so voluntarily. Between 50,000 and 250,000, insiders must report their trades periodically every quarter and above 250,000 trades must be reported promptly. < INSERT TABLE 2 HERE > In total, there were 5,588 insider trades reported during the sample period. Most were small, with about 25% of the sample consisting of trades greater than 50,000. The smallest trade size category had a mean trade size of 11,600 compared to 104,940 for periodic filings and 2,417,770 for prompt filings. The smallest trade in the sample was 4,000 while the largest trade was 92,650,970, undertaken by a director of Camfin, the parent company of Pirelli-Telecom. Italy is unusual in its insider trading characteristics in that buy and sale trades are of approximate similar size and equally common. In other countries, such as US and UK, buy trades tend to be more numerous and considerably smaller than sales [Jaffe (1974), Friederich, Gregory et al. (2002)]. Insider Trading Compliance Behavior In Table 3, we present information on the extent to which corporate insiders in Italy adhere to disclosure regulation. All insiders are identified by their position within the firm and categorised according to their disclosure behaviour. If an insider promptly reports all their trades in accordance with the disclosure requirements, he

20 is viewed to be fully compliant. At the opposite end of the spectrum, an insider that ignores the disclosure requirements and reports none of their trading is labelled non-compliant. We define the remaining group, which selectively reports their trades, to be partially compliant. Panels A and B refer to family and non-family firms respectively. < INSERT TABLE 3 HERE > It can be seen from Table 3 that corporate insiders in Italy very rarely report their full trading activity (58 out of 195 insiders). Surprisingly, there were more insiders in the non-compliant category than either fully or partially compliant. Chief Executives seem to be particularly negligent in their reporting with fifty percent of the group completely non-compliant. Panels A and B illustrate quite striking differences between family and non-family firm compliance. In Italy, family firms are the dominant group of companies and this is replicated in the mix of our sample with 125 firms controlled by families compared to 70 with no family control. Overall, an examination of the two panels shows that 50.6% of the shares traded by insiders are disclosed in family firms relative to a lower 39.5% in non-family firms. Family firm compliance is therefore significantly higher than their non-family counterparts and this difference is significant at the 5% level 4. A robust analysis of individual job role compliance in family and non-family firms is difficult due to the small number of observations in many of the non-family job categories. However, an examination of the two panels shows that the mean (median) percentage of chairmen s trades which are regularly disclosed amount to 55.5% (71.8%) in family firms compared to a lower 38.9% 4 Tests not reported in the table. P-values respectively equal to (t-test) and (Wilcoxon test) for tests of different means and medians

21 (24%) in non family firms. The percentage of directors compliant with the regulation is roughly similar across groupings. 4. RESULTS In this section, we explicitly test the hypotheses developed in the paper. In Section 4.1, the determinants of regulatory compliance are investigated. The control variables are categorised into four main groupings and relate to structural and financial characteristics, ownership and board characteristics, internal governance and control, and supplementary internal compliance procedures. Section 4.2 examines the appropriateness of the Italian regulatory framework in establishing transaction size alone as the most relevant variable to proxy for information content. The performance of insider trades is related to firm s regulatory compliance in Section 4.3, while Section 4.4 reports the main determinants of market reaction to the announcement of insiders trades. 4.1 Determinants of Regulatory Compliance Table 4 presents coefficients from ordinary least squares and ordered probit regressions relating firm regulatory compliance to a number of control variables. There are various ways to measure regulatory compliance within a firm since there are different perspectives on what is actually meant by compliance. We utilise two metrics, a compliance ratio and an ordered compliance measure. Our compliance ratio is defined as the ratio of the number of traded shares disclosed by insiders to the total number of shares traded by insiders. The ordered compliance measure is equal to 1 if the firm is non-compliant (i.e. no directors complied with the disclosure

22 regulation), 2 if the firm is partially compliant, and 3 if the firm is fully compliant (i.e. all directors complied with the disclosure regulation). Both metrics are intuitive measures of regulatory compliance but capture different aspects of the concept. The compliance ratio treats corporate insiders as a homogenous body and regards the proportion of shares traded that have been disclosed as the most informative measure of compliance. The ordered compliance variable, on the other hand, treats insiders as distinct individuals and measures the proportion of those who are compliant, disregarding the number of shares that have actually been traded and disclosed. A number of factors can be judged to influence the propensity of regulatory compliance within a firm and these are captured in each of the panels of Table 4. Panel A considers the structural and financial characteristics of firms, Panel B examines ownership and board characteristics, Panel C reports the effect of the controlling shareholder group and the internal governance structures within the firm, and Panel D investigates whether more stringent compliance constraints within firms are likely to influence firm compliance. A detail description of each of the control variables is provided in the Appendix. <INSERT TABLE 4 HERE> Our first hypothesis relates to the main factors that influence firm compliance and in particular we strongly argued for the role of family and concentrated ownership as well as internal compliance structures in improving compliance levels. In contrast, we argued that firms with greater need for external monitoring within ineffective regulatory regimes, larger firms with more nonexecutive directors, are less likely to be compliant. In general, the results support the main thrust of this compound hypothesis

23 In Panel A, the coefficients from OLS and ordered probit regressions using firm and structural characteristics are presented. In acknowledgement that financial firms have very different characteristics from their non-financial counterparts, we include a financial firm dummy variable. Interestingly, this is the only variable that is significant in these regressions. Size, which is one of our core theoretical factors, has no influence on firm compliance although the sign is as predicted and consistent across all six models. The effect of ownership and board structure is presented in Panel B. The impact of concentrated ownership and control on firm compliance is very clear with both variables significant at least at the 5% level in all regressions. Consistent with the argument that more non-executive directors within weak enforcement regimes is actually a predictor of non-compliance, we report strong statistical evidence of a negative relationship between the number of non-executive directors in a firm and the likelihood of regulatory compliance. Hypothesis 1 predicts that family firms with the extensive peer and family networks are likely to have higher levels of compliance. The results in Panel C fully support this hypothesis. In particular, family firm status is shown to be highly significant in explaining the propensity for regulatory compliance. Notably, traditional recommendations for corporate governance such as separating the role of chairman and chief executive, performance related management incentives, the presence of audit and executive committees, and cumulative voting procedures have absolutely no impact on regulatory compliance. Panel D examines whether more stringent internal compliance structures lead to improved disclosure. Surprisingly, these do not appear to be important and may actually have a negative impact. A common criticism of corporate governance

24 regulation is that firms may follow the letter of the code but not necessarily its spirit. In Italy, this appears to be the case with not one internal compliance procedure improving compliance itself. It is also shown that lower disclosure thresholds statistically reduce the probability that insiders will adhere to the regulatory reporting requirements. In effect, the lower internal thresholds are generally being ignored by corporate insiders. 4.2 Is Transaction Size an Appropriate Proxy for Information Content? Hypothesis 2 presents the argument that transaction size is not an optimal identifier for private information when other, more appropriate, measures are available. The most commonly recognised alternative is that of multiple trading by insiders, possibly as a result of stealth trading [Barclay and Warner (1993)] or signalling activity by insiders [John and Mishra (1990), John and Lang (1991), and Bagnoli and Khanna (1992)]. To test this hypothesis, we compare the share price behaviour of insider firms for large (> 250,000) and small (< 250,000) transactions and around single and multiple trading events. Insider transactions that occur on the same day are netted out, to construct a composite trading signal. Longer netting intervals could be used but as with any research design, there are trade-offs between precision and number of observations or power. We define multiple trades those events when more than one insider trades or one insider makes more than one trade in one single day. We opted for multiple trades defined within one day because most multiple trades took place on the same day and the measurement of abnormal returns around the event was considerably more intuitive and simple. If the Italian disclosure regulation was appropriate, the

25 information content of large trades would be greater than all other information proxies. <INSERT TABLE 5 HERE> Table 5 presents summary information on cumulative abnormal returns (CAR) derived from market model regressions with a parameter estimation period of 160 days spanning 180 to 21 days before each insider transaction. Three CAR windows are presented covering 20 days pre-trade, the event date, and 20 days posttransaction. Of greatest interest is the post-trade period, when the market reaction to corporate insider trading can be ascertained. Taking the full sample first, the pattern of cumulative abnormal returns around insider trading is slightly different to studies that examine insider trading performance in other countries, such as the US. Whereas, many studies find a classical V shape pattern in CARs around insider buy transactions, Italian insider buying activity tends to occur after a period of positive share price performance, followed by further excess share price performance after the trade. The pattern in cumulative abnormal returns around sales transactions exhibits an inverse V shape and is similar to that seen in other countries [see, for example, Friederich, Gregory, Matatko and Tonks (2002), Biesta, Doeswijk ann Donker (2003)]. The difference in insider buying CARs may be due to a number of reasons. First, Table 5 reports abnormal returns from disclosed insider trades. Noncompliant (i.e undisclosed) trades may have significantly more information content, and return patterns for these may resemble that of the US and other countries. In addition, given the ownership structure of Italian firms and the dominance of family control, insiders may not fully exploit their informational advantage since trading is more likely to involve other members of the extended business or family network

26 The Italian regulation requires prompt disclosure of insider trading activity when trades are greater than 250,000. Periodic disclosure, every three months, is required for trades between 50,000 and 250,000 and disclosure is not required for trade sizes under 50,000. If the regulation is based on sound principles, large transactions will carry greater information content than small trades. An examination of Table 5 shows this not to be the case. For buy transactions, posttrade cumulative abnormal returns are actually significantly negative at -1.51% for large trades. This is in comparison to small buy transactions that experience a significant and positive CAR of 0.79% over 20 days after the insider trade. Sales transactions for both small and large trades have similar negative post-trade CARs in the region of -2%. A comparison of cumulative abnormal returns for buy and sell trades across transaction size shows no difference in post-trade CARs for large trades (mean difference = 0.34%), while for small trades the difference in post-trade CARS between buy and sell transactions is significant at 3.03%. An alternative measure of private information is multiple trading activity by insiders [Dufour and Engle (2000)]. In Table 5 it can clearly be seen that multiple trades experience greater absolute post-trade abnormal returns for both buy and sale transactions than single trades. Multiple buy trades experience post-trade CARs that are 73 basis points greater than single buy trades while Multiple sell CARs are more than 2.5% lower than single sell CARs. Moreover, multiple trades have significantly greater information content than large trades. Clearly, transaction size is not an appropriate benchmark for information content and other proxies should be considered by the Italian regulator. 4.3 Performance of insider trading and firm compliance

27 In this section, we investigate hypothesis 5 - how a firm s compliance to disclosure regulation affects the performance of insiders trades. Mean and median cumulative abnormal returns around insider trading activity are presented in Table 6. A standard event study methodology is used with an estimation period of 160 days spanning 180 days to 21 days before each insider transaction. Two-tailed parametric t-tests and non-parametric Wilcoxon signed ranks tests are carried out to determine statistical significance across buy and sell trades and full and partial or noncompliance. <INSERT TABLE 6 HERE> To improve statistical power and facilitate interpretation, the partial and noncompliance categories are combined. 5 Surprisingly, there seems to be very little evidence of any difference between the excess returns of compliant and partially or non-compliant firms. The analysis has different dimensions along pre- and posttransaction returns and across buy and sell trades. Taking pre-transaction excess returns first, the difference in excess price performance between fully compliant and non-compliant firms is statistically insignificant for both buy and sell trades (not reported in the table). Subsequent to the insider trade, sell transactions are followed by statistically significant mean excess returns of -1.01% (the median is -1.82%) over 20 days for compliant firms compared to insignificant mean excess performance of 0.38% (the median is %) for non-compliant firms. The difference in post-trade performance of sell transactions between compliant and non-compliant firms is significant at the 5% level and of the expected sign. That is, insiders anticipate stock performance more in compliant firms than in partial or non-compliant firms. One possible reason could 5 The partial and non-compliance firms were combined because the small number of transactions (7 sales and 38 buys) for non-compliant would lead to low power in statistical testing, even for nonparametric tests

28 be that insiders do not disclose their trades when they prelude severe drops in stock prices due to still undisclosed relevant information. Another way of investigating the information efficiency of firms with different levels of compliance is to examine the difference between the performance of buy and sell trades. Hypothesis 5 does not explicitly predict whether share price performance of compliant and non-compliant firms after insider trading will be more different or more similar across buy and sell trades. On one level, fully compliant firms will have greater information quality in the market and since prices are less noisy, insiders trades lead to stronger excess returns. However, an alternative view that follows Kim and Verrechia (1997) argues that fully compliant firms will have better information quality than non-compliant firms and, as a result, the information component in insider trades in non-compliant firms will be greater. The results in Table 6 suggest that the former is somehow true. In fully compliant firms, the mean post-trade share price performance of insider sell transactions is -1.01%, and is significantly lower compared with the non-compliant firms (+0.38%). 4.4 Determinants of market reaction to the announcement of insider trades In this section, we bring together corporate governance, internal regulatory compliance processes, and ownership forms to examine the main factors underlying the market reaction to corporate insider trades. In Table 7, three-day [-1,+1] cumulative abnormal returns around the announcement of insiders trade for buy and sell transactions are regressed against a number of different factors. We separate the analysis into buy and sell trades because of the different motivations and restrictions that may underlie each transaction. Insider buying activity is seen by the market to

29 be a positive signal and, as a result, companies are naturally less inclined to restrict such insider activity. Selling transactions, in contrast, can provide negative information to the market, bring attention to the company, and potentially have a negative effect on social and business networks if the counterparty is a close relative or acquaintance and prices respond negatively afterwards. Under these situations, it is possible that different factors may have varying effects on the market reaction to insider trades. <INSERT TABLE 7 HERE> Table 7 presents the results of our analysis. To facilitate comparison of buy and sell determinants, we reverse the sign of the sales coefficients since announcement sale returns are negative whereas the opposite sign is observed for insider buy trades. Expected signs, based on theoretical reasoning, are also presented in the table. Taking insider buy transactions first, trades that are promptly reported to the market experience a significantly stronger share price reaction than other trades. Since transaction size and multiple trading signals, both indicators of information content, have already been controlled for in the analysis, this is strong evidence that promptly disclosed trades lead to a stronger market reaction, and better information quality in the market in aggregate. Consistent with earlier research [Gregory, Matatko et al.(1994)]), the market reaction to buy trades is lower for larger firms, and higher for smaller firms, again indicative of greater information quality for this group of companies. Small firms will normally have very light analyst following, little exposure in the financial press, and illiquid stock. As a result, insider trading activity in these firms provides a much more important insight into the views and expectations of management. Family firm status and percentage of non-executive