Economic Notes by Banca Monte dei Paschi di Siena SpA, vol. 34, no , pp

Transcription

1 CE: ANP CSE: AR ECNO 155 Economic Notes by Banca Monte dei Paschi di Siena SpA, vol. 34, no , pp Determinants of Corporate Governance in the Italian Financial Market EMILIO BARUCCI* JURY FALINI We evaluate the determinants of corporate governance of companies listed at the Italian Stock Exchange. We consider ownership structure, balance sheet data, company performance and some qualitative features as determinants. We evaluate the convergence of Italian companies governance towards a system with effective governance mechanisms. Our analysis shows definitely that companies with a large shareholder and/or minority blockholders controlling a large stake and companies belonging to a pyramidal group or controlled by a shareholders coalition are characterized by a poor governance/shareholders protection standard. Institutional investors play a positive role. (J.E.L.: G30, G34, G38). 1. Introduction This paper aims to evaluate the corporate governance of companies listed at the Italian Stock Exchange. For each company, we consider two main sources of information: (i) the 2002 and the 2003 by-laws, (ii) the 2002 and the 2003 statements on compliance with the code of best practice introduced in 1999 by the Italian Stock Exchange (the so-called Preda report). Our main goal is to investigate determinants of corporate governance and shareholder protection standards. To this end, we have collected data on company qualitative features, performance, balance sheet information and ownership structure. Up to 1990s, Italian financial markets were characterized by concentrated ownership, weak investor protection and state ownership. Banks and institutional investors did not play an active role on monitoring companies and on their corporate governance; companies were mainly * Dipartimento di Matematica, Politecnico di Milano, Via Bonardi 9, 20133, Milano, Italy. ebarucci@ec.unipi.it 2 Dipartimento di Statistica e Matematica applicata all Economia, Universita` degli Studi di Pisa, Via C. Ridolfi, , Pisa, Italy. We thank for comments Marcello Bianchi, Fabrizio Mattesini, Marcello Messori and participants at seminars held at Universita` della Svizzera Italiana, Perugia, Pisa, Roma Tor Vergata, Rostock, SOAS London, Verona. We thank a referee for useful comments and insights to improve the quality of the presentation of the paper. The usual disclaimers apply. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford, OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.

2 372 Economic Notes : Review of Banking, Finance and Monetary Economics under the control of the state or families. The system was quite opaque and the stock market was underdeveloped. The privatization process started in 1992 was joined by a deep process of modernization of the financial market regulation. The main goal was to increase investor protection favouring dispersed ownership and to make corporate governance effective. This process has transformed the Italian financial market. Since 1992, four main regulatory pieces have affected financial markets and corporate governance: the Testo Unico Bancario (TUB) in 1993, the privatization law in 1994, the Testo Unico della Finanza (TUF) in 1998 and the Preda code of best practice for listed companies in Some of these norms are normative-mandatory, some others have a dispositive feature, and some others have no impositive power at all. In particular, the code of best practice is an example of self-organization: the Italian Stock Exchange and listed companies agree on a set of rules on corporate governance to maximize the shareholder s value; companies are not forced to accept them, but they have to release a report on compliance with the code. In our analysis, we intend to address three main topics: analysis of corporate governance determinants in the Italian financial market, effectiveness of the new regulation on corporate governance and convergencepath dependence of the Italian governance system. The literature on corporate governance mainly deals with the US experience (Denis et al., 1997; Agrawal and Knoeber, 1998; Denis and Sarin, 1999; Deli and Gillan, 2000; Adams and Ferreira, 2003); little is known on other countries, and few exceptions are provided by cross-country analyses considering an aggregate index of corporate governance (Klapper and Love, 2002; Durnev and Kim, 2003; Doidge et al., 2004) and papers addressing manager 3 turnover (Kaplan, 1994a, b; Gibson, 1999; Dahya et al., 2000; Franks et al., 2001; Volpin, 2002; Renneboog and Trojanowski, 2003). While in a public company, governance devices have mainly to address the manager shareholders conflict, the main problem in a system with concentrated ownership is provided by the exploitation of minority shareholders with private benefits for the controlling shareholder. Hence, corporate governance problems are different and also governance mechanisms and their determinants may be different. In our analysis, we intend to address this topic. The second topic of this paper is the effectiveness of the new pieces of regulation introduced in the Italian financial market. It is not easy to evaluate the effectiveness of mandatory rules, e.g. the fact that derivative suits have not been promoted by minority shareholders does not mean that the regulation does not work; it may act as a deterrent. It is easier to test the effectiveness of dispositive rules or of a code of best practice: being based on a voluntary basis, we can evaluate the degree of compliance of companies. The degree of compliance signals effectiveness of the regulation. In what follows, we mainly consider dispositive rules contained in the TUF and in the code of best practice on internal governance and

3 E. Barucci and J. Falini: Italian Financial Market shareholders protection. Considering 2002 and 2003 governance of companies, we are able to evaluate the revision of the code of best practice put forward in July Italian financial markets during the last 10 years offer an interesting example of transition from a non-well-developed financial market to a modern financial market. By modern financial market, we do not necessarily refer to the UK US system but only to a financial market where governance mechanisms work properly and the stock-debt market is developed. We have the opportunity to check for formal and functional convergence. A recent debate motivated by the literature on finance, law and growth (La Porta et al., 1997, 1998; Demirguc-Kunt and Maksimovic, 1998; Levine and Zervos, 1998; Rajan and Zingales, 1998; Beck and Levine, 2001; Beck et al., 2002a, b) has posed the following question: Are corporate governance systems going to converge towards the AngloSaxon-type corporate governance model based on stock markets, diffused ownership and management focused on the shareholder value? Some authors stress formal convergence driven by new legal rules increasing investor protection, and some others point out that convergence will occur through more market-based arguments such as market integration, free bargaining and self-regulation (Coffee, 1999; Coffee, 2000; Hansmann and Kraakman, 2000; La Porta et al., 2000a). By contrast, there are those 4 who point out path dependency on corporate governance which impedes convergence (Bebchuk, 1999; Roe, 2000; Rossi, 2001; Roe, 2002). The first hypothesis is mainly driven by a legal hypothesis and by the supremacy of the diffused ownership model to pursue efficiency and growth. The second hypothesis mainly relies on a political-incumbent hypothesis: institutions and politics affect the corporate governance system and may hamper convergence of governance systems. Advocates of the (functional) convergence thesis stress the relevance of self-regulation experiences (e.g. code of best practice): self-regulation precedes normative rules. According to this perspective, evaluating effectiveness of the new regulatory system, we can assess convergence of the Italian corporate governance system, and looking for determinants of corporate governance, we can identify convergence driving forces and forces impeding it. Our analysis shows that in general the rate of compliance with the code of best practice is high, but it is low on some crucial points (role of independent directors, director selection and minority shareholders protection). While forces contrasting convergence towards an effective governance system (large shareholder s stake, companies belonging to a pyramidal group or controlled by a shareholders agreement) work against convergence (they negatively affect the governance of the company), forces that should enhance convergence (monitoring by large outside blockholders, good growth options) do not succeed to guarantee good governance devices. Institutional investors positively affect governance quality.

4 374 Economic Notes : Review of Banking, Finance and Monetary Economics Large companies and companies recently listed in the market seem to have a better governance. No significant effect is associated with the code revision in This paper is organized as follows. In section 2, we briefly describe recent developments on the Italian financial markets regulation. In section 3, we present the dataset used in our analysis and some descriptive statistics. In section 4, we describe our research strategy. In section 5, we provide the empirical analysis on corporate governance determinants. 2. Financial Markets and Corporate Governance in Italy in the Last Decade Up to 1990s, Italian financial markets were characterized by concentrated ownership, weak investor protection and state ownership. The financial system was neither market based nor bank based; companies were mainly under the control of families directly, through pyramidal groups and/or through coalition cross-holding. Financial intermediaries did not play an active role on corporate governance. The system was quite opaque and the stock market was underdeveloped (on the Italian financial market system and corporate governance, see (Ciocca (2000), Bianco and Casavola (1996) and Bianchi et al. (2001)). The start of the privatization process in 1992 was joined by a process of modernization of the regulation of financial markets. The main goal was to increase investor protection favouring dispersed ownership, to help financial markets development and to make corporate governance effective. There are four main regulatory pieces that have affected financial markets and corporate governance in the last 10 years and that are relevant in our analysis: the TUB in 1993, the privatization law in 1994, the TUF in 1998 and the code of best practice in The TUB allowed commercial banks to detain stakes in non-financial companies. For 50 years, commercial-retail banks could not hold stakes in non-financial companies. The privatization process is relevant in our research for two main reasons. First of all, it has reduced the presence of the state in the economy, and in particular banks are not anymore under the control of the state; second, it has contributed to raise the debate on corporate governance and financial markets in Italy, and political parties discussed for a long time on the best way to privatize companies (Fulghieri and Zingales, 1994; Jaeger, 1995; Marchetti, 1995). The law on privatizzazioni (law nr. 474, 30 luglio 1994) introduced some relevant norms for privatized companies: golden share, limits to shareholding, voto di lista mechanism to appoint directors of the board and members inside the board of auditors (with 1 per cent as threshold to present a list) and mailing vote. According to the Civil Code, directors and auditors are appointed on a majority rule basis; the voto di lista mechanism introduced

5 E. Barucci and J. Falini: Italian Financial Market 375 some proportionality in the representation of the board of directors: shareholders with a stake above a certain threshold may present a list, and then directors are chosen among those indicated in the lists on a pure proportional basis or according to rules establishing that the majority of directors is in any case drawn by the most voted list. Instead, for the selection of the board of auditors, one over three auditors or two over five auditors are chosen from the lists that have not received the relative majority of votes. The TUF establishes a new regulation on takeovers, allows crossholding below 5 per cent, introduces a strict regulation on the flow of information to the market and reinforces shareholders rights. On this point, the main innovations are the following: shareholders quorum to convene a shareholders meeting (from 20 per cent to 10 per cent), quorum to appeal to the board of auditors (from 5 per cent to 2 per cent), quorum to appeal to the court against CEOs and auditors (from 10 per cent to 5 per cent), 5 per cent as a quorum to promote a law suit (azione di responsabilita`) against directors without shareholders meeting approval (as before the TUF), 2/3 quorum for extraordinary shareholders meetings and not simple majority as before, and minority shareholders can be represented inside the board of auditors (one over three or two over five members); control on the management is assigned to the board of auditors, and control on accounting is assigned to the societa` di revisione; proxy voting through banks is allowed. Moreover, the TUF provides large space to autonomia statutaria (free bargaining among shareholders): companies can provide stronger investor protection (lower quorums, proxy and mail voting, size of the board of auditors) through their by-laws. According to the index on investor protection computed by La Porta et al. (1997, 1998), after the TUF, Italy goes from one to five over six. Efficiency of corporate governance is handled by the TUF through external governance mechanisms (takeovers and surveillance by the CONSOB) and ex post legal 5 actions by shareholders and the board of auditors. It is difficult to evaluate the effect of the TUF on financial markets and corporate governance. PricewaterhouseCoopers (2000) investigates the corporate governance of more than 100 listed companies and large privately held companies. They show that only a small fraction of companies has a board of directors with independent directors and specialized committees inside the board. Excluding privatized companies, only a small fraction of companies has reinforced investor protection by adopting the voto di lista mechanism to appoint directors and lowering quorums/ thresholds to exert shareholders rights. Moreover, since the TUF, ex post legal rights have not been exerted by minority shareholders (Spaventa, 2002). Only once an extraordinary general shareholders meeting has been called by minority shareholders, and only in one extraordinary general shareholders meeting, the 2/3 quorum was not reached

6 376 Economic Notes : Review of Banking, Finance and Monetary Economics to approve a merger. Mail voting is included in a small fraction of company by-laws and has never been exerted in the last 2 years. Proxy voting is not used, in no general shareholders meeting we have assisted to a solicitation of votes. No azione di responsabilita` has been promoted by minority shareholders. These facts are striking. They say two interesting things. First, since the TUF shareholders rights have not been exerted frequently, this observation does not mean that they are not relevant; rules protecting shareholders rights may well act as a deterrent. Second, the TUF has not affected the internal governance of companies, and therefore at the beginning of the third millennium, the main reference continues to be the Civil Code (1942). The Civil Code is not designed to make the company work in 6 an efficient way; in fact, the Civil Code does not provide many insights on the management organization of the company. According to the code, there is a supremacy of ownership; almost no word is dedicated to the powers of managers, i.e., they are deduced from those assigned to the shareholders meeting (art ) and from the fact that they can be charged for bad management through a law suit (azione di responsabilita`) after a vote by the shareholders meeting (art. 2393). The management of the company should be directed to achieve the interesse sociale (social interest) which is not necessarily the shareholder value (Jaeger, 2000; Ferrarini, 2002). In this perspective, the Preda best practice code marks a step towards an effective governance. The code concentrates on corporate governance defined as the system of rules according to which the company is managed and controlled. The main goal of the recommendations of the code is the maximization of the shareholder value under the assumption that a good governance system will have good effects also on other stakeholders. We concentrate our attention on the following topics addressed by the Preda code: 1 Role of the board of directors: The Preda report recommends that companies define exclusive powers of the board of directors. The report includes among exclusive powers of the board the approval of strategic plans, of significant transactions and of transactions with related parties. Executive directors (including in case the chairman) shall have delegated powers defined by topic and/or by monetary limits. Transparency on delegated powers is expected. 2 Composition of the board of directors: The report identifies two classes of directors: executive and independent directors. Non-executive directors shall balance executives views in taking board decisions consistent with shareholders interests. An adequate number of non-executive directors shall be independent. Independency refers to independency from owners, the company and executive directors. Independent

7 E. Barucci and J. Falini: Italian Financial Market 377 directors shall not detain a stake enabling them to exercise a considerable influence on the control of the company. 3 Role of the chairman of the board of directors: The chairman shall call meetings of the board, shall co-ordinate the activities of the board and in particular shall provide information documentation to members reasonably in advance of the date of the board meeting. 4 Information to the board on delegated powers: The executive committee and managing directors shall periodically report to the board of directors and board of auditors on the activities performed in the exercise of their delegated powers. 5 Appointment of directors: Proposals for director appointments, accompanied by detailed information, shall be deposited 10 days before the shareholders meeting. The same recommendation is made for appointments to the board of auditors. In case an appointment committee is established, the majority of the committee shall be made of nonexecutive directors. 6 Remuneration of directors: The board of directors shall form a committee on remuneration. The committee, the majority of whose shall be non-executive directors, shall submit proposals to the board on the remuneration of managing directors and of directors who are appointed to particular positions. Proposals are taken avoiding conflicts of interests. A part of the remuneration shall be linked to the company s profitability achievement of specific goals. 7 Internal control: The board of directors shall define an internal control system to monitor the efficiency of the company s operations. The board of directors shall establish an internal control committee, charged with the task of making proposals on the control system. 8 Transactions with related parties and conflict of interests: Transactions with related parties shall comply with the criteria of fairness. Directors who may have conflicts of interest on some specific decisions shall abandon the board. 9 Relations with institutional investors: The board shall designate an investor relator. The code stresses the monitoring role of the board; the board maintains a central role in defining strategic plans and delegated powers, but the management of the company is mainly delegated to executive directors. Recommendations included in the report are similar to those adopted in countries with well-developed markets and diffused ownership. A system with concentrated ownership may need a regulation different from that adopted in countries with dispersed ownership. Rossi (2001) and Spaventa (2002) point out serious doubts on the effectiveness of the code. They argue that the code does not address the main negative peculiarities of the Italian financial system: interlocking directories (there is no limit to

8 378 Economic Notes : Review of Banking, Finance and Monetary Economics multiple directorships), pyramidal groups, blocking minorities, extensive cross-holding and non-transparency on decisions taken by coalitions. In fact, codes of best practice are designed to allow shareholders to exert their voice more easily, but the effect is weak when exit is the unique option for small shareholders. A code of best practice is intended to make a public company work well and not to protect minority shareholders. On effectiveness of codes of best practice, we refer the reader to the works of Dahya et al. (2000), DeJong et al. (2001) and Drobetz et al. (2003). Papers mainly deal with the effects of codes of best practice on firm valuation and CEO turnover. Dahya et al. (2000) address the effect on board effectiveness of the UK s best practice code (Cadbury Committee report). The report recommends that boards of UK corporations include at least three outside directors and that the positions of chairperson and CEO be held by different individuals. Authors show that CEO turnover and its sensitivity to performance has increased after the code has been introduced. Sensitivity increase appears to be due to the increase of the fraction of outside directors. DeJong et al. (2001) analyse the effect of the self-regulation experience in the Netherlands (the best practice code proposed by the Peters committee in 1997); no effect on firm-value-asset price is detected. They conclude that self-regulation which relies on monitoring without enforcement by either exchanges or governments, or where there is limited or no outside monitoring, is unlikely to be successful. The results also cast considerable doubt on the success of other (similar) selfregulation initiatives undertaken by European Unione countries. Drobetz et al. (2003) concentrate on the German code of best practice experience; they show that the market-value-to-book-value ratio is positively related to the rating of governance proxies. As far as expected returns are concerned, a negative relationship between them and corporate governance is observed: companies with a good corporate governance can reduce their cost of capital. The rationale for this result is that a good corporate governance reduces agency costs and therefore a lower cost of capital is observed. Summing up, the TUF does not affect companies governance; it reinforces shareholders rights, but new rules do not seem to be effective; the code of best practice has introduced internal corporate governance rules that are designed to pursue the shareholder value in a public company system. On Italian corporate governance, see Volpin (2002), Barontini and Caprio (2002), Bianchi and Enriques (2001) and Belcredi et al. (2002). 3. The Dataset: Variables and Descriptive Statistics To get a picture on corporate governance of listed companies in , we consider two pieces of information: (i) the 2002 and the

9 E. Barucci and J. Falini: Italian Financial Market by-laws, (ii) the 2002 and the 2003 statements on compliance with the code of best practice introduced in 1999 by the Italian Stock Exchange (the so-called Preda report). On the size of the board of directors and on the size of the board of auditors, we rely upon official data by CONSOB. On members of the board of auditors appointed by minority shareholders, we directly refer to shareholders meeting minutes. We exclude companies belonging to the mercato ristretto (the market for small companies) and companies based as exchange market in other countries (e.g. STM electronics, Banco de Bilbao, Banco de Santander) because they comply with different governance codes. We are left with a sample of 277 companies in 2002 and 275 companies in Companies are not forced to meet the recommendations of the code of best practice, but they have to evaluate their compliance with the code each year, explaining in case non-compliance; note that this feature is a peculiarity of the Italian code with respect to those of other European countries DeJong et al. (2001). The board of directors presents each year a report on the corporate governance at the shareholders meeting pointing out how the company has met the recommendations of the code. These reports are available on-line ( The code has been promoted by the Preda Committee in 1999; companies started to comply with the code in 2000; reports are available on-line for 2001, 2002 and We have collected the 2002 and 2003 reports. We consider 2002 reports because some inertia was observed on compliance with the code in Some companies do not comply with the code, and for them, we do not have any information on corporate governance. We also consider 2003 governance because in July 2002 a revised version of the code has been provided. Hence, we can evaluate the effect of this revision. The main revisions concern a stricter definition of independent directors and deals with related counterparts: the board should introduce criteria to identify this type of deals and in case should rely on expert advices. We start by observing that 2002 (2003) corporate governance reports as well by-laws are prepared and conceived by the end of 2001 (2002); hence, we consider ownership structure and balance sheet data by the end of 2001 (2002) as exogenous variables Endogenous Variables We consider governance features regarding three main governance areas: board of directors composition and selection of directors, board of directors activity and board of auditors. As far as the composition of board of directors is concerned, we consider the number of directors sitting on the board (boardsize) and the fraction of independent directors (fracind) (Table 1). In a public company, 7

10 380 Economic Notes : Review of Banking, Finance and Monetary Economics Table 1: Description of Variables Variable blockhold boardsize ceochair chairind comcon(i) connom comre(i) down dual excl fracind infodis infopre inform inst ipo limdel lista listasin lrgshr lrgshr2 lsales mvbv numsin par privat pyram synd state Definition Stake detained by shareholders with more than 2% not related to the largest shareholder Number of directors Dummy variable assuming value equal to 1 if the chairman of the board is also a CEO Dummy variable assuming value equal to 1 if the chairman of the board is independent Dummy variable assuming value equal to 1 if there is an internal control committee (fully independent) Dummy variable assuming value equal to 1 if there is an appointment committee Dummy variable assuming value equal to 1 if there is a remuneration committee (fully independent) Dummy variable assuming value equal to 1 if the largest shareholder is also an executive Dummy variable assuming value equal to 1 if the company is also listed in UK or in US Dummy variable assuming value equal to 1 if exclusive powers of the board are defined Fraction of independent directors Dummy variable assuming value equal to 1 if there exists a procedure to disclose information to the market Dummy variable assuming value equal to 1 if the chairman provides information to directors before meetings Dummy variable assuming value equal to 1 if information on candidate directors is released before appointment Stake detained by banks, insurance companies, foundations, investment funds Dummy variable assuming value equal to 1 if the company entered the market in the period Dummy variable assuming value equal to 1 if delegated powers from the board of directors to CEO are defined Dummy variable assuming value equal to 1 if the voto di lista to appoint directors is adopted Threshold to present a list to appoint auditors Stake detained the largest shareholder/coalition lrgshr*lrgshr log of sales Market value over the book value Dummy variable assuming value equal to 1 if the board of auditors size is larger than 3 Dummy variable assuming value equal to 1 if exclusive powers of the board of directors on transactions with related companies are defined Dummy variable assuming value equal to 1 if the company has been privatized Dummy variable assuming value equal to 1 if the company belongs to a group Dummy variable assuming value equal to 1 if the company is controlled by a shareholders coalition Dummy variable assuming value equal to 1 if the company is controlled by the state a large board of directors is a bad governance feature because it weakly monitors the management of the company and its governance. In a company with concentrated ownership, the relation between the size of board

11 E. Barucci and J. Falini: Italian Financial Market 381 of directors and governance is ambiguous, i.e. a small board is more likely to be controlled by the controlling shareholder. However, there is a wide empirical evidence showing that companies with a large board are characterized by a low market value even in a concentrated ownership system (Barucci and Ceccacci, 2005). Independent directors are supposed to balance executives interests and to pursue the shareholder s value. Therefore, a small board of directors and a large fraction of independent directors indicate a good governance. Appointment of directors is a crucial issue in the company life. Unless the by-laws dispose in a different way, directors are appointed by the shareholders meeting and all directors come from the most voted list, i.e. shareholders representing the relative majority at the meeting appoint all the directors of the board. As pointed out in section 2, a system that makes easier for minority shareholders to appoint directors is provided by the voto di lista mechanism. To catch this fact, we consider the dummy variable lista which assumes value equal to one when the voto di lista mechanism is adopted and zero otherwise. The Preda code recommends transparency on candidate directors selection, i.e. candidate directors should be declared 10 days before the shareholders meeting; when the company satisfies this recommendation, the dummy variable inform assumes value equal to one and zero otherwise. These two variables stand for transparency on the appointment of directors and for greater protection of minority shareholders. Companies adopting the voto di lista mechanism also satisfy information disclosure requirements. The role of a chairman of the board of directors as an independent figure is stressed by many authors/commentators. To catch this feature, we consider the dummy variable assuming value equal to one if the chairman of the board is also the CEO of the company (ceochair) and a dummy variable assuming value equal to one if the chairman of the board is independent (chairind). While coincidence of the chairman of the board of directors and CEO is a proxy of poor governance, independency of the chairman is a good indicator. As far as the activity of board of directors is concerned, we consider two different features. First of all, we look at the definition of exclusive powers of the board by considering a dummy variable assuming value equal to one if exclusive powers of the board are defined (excl). The Preda code and the TUF have reinforced the role of the board of directors on the management of the company. One of the major governance problems in the Italian financial market is that management decisions are taken by controlling shareholders outside the board of directors, e.g. shareholders coalition, the controlling family. In this environment, definition of exclusive powers of the board of directors is a good governance feature because it reinforces its accountability to shareholders. The Preda code has introduced committees inside the board. The rationale is that introducing

12 382 Economic Notes : Review of Banking, Finance and Monetary Economics internal committees on appointment, internal control and remuneration, the board of directors work more efficiently. In what follows, we consider a dummy variable assuming value equal to one if an internal control committee has been introduced (comcon), a dummy variable assuming value equal to one if a fully independent internal control committee has been introduced (comconi) and a dummy variable assuming value equal to one if a remuneration committee has been introduced (comre). The board of auditors is made up of three up to five members. The company chooses the size of the board; since 1994, privatized companies are forced by the law on privatization to adopt a board with five members. The size of the board of auditors is crucial because in a board with five members, two of them are appointed by minority shareholders (if they present a list), independently of their votes. Two auditors have a lot of power; they can convene the shareholders meeting and can inspect the company. To consider companies with a board of auditors made up of more than three members, we consider a dummy variable assuming value equal to one in that case and zero otherwise (numsin). The threshold needed to present a list to appoint members of the board of auditors (listasin) is an important governance feature because a low threshold makes easier for minority shareholders to appoint auditors. To catch the presence of an auditor appointed by minority shareholders, we consider the dummy variable sindmin. We have considered other variables representing the governance of a company, but we have not been able to select a significant model for them and therefore we only present some descriptive statistics. We have considered the following variables: a dummy variable assuming value equal to one if the chairman of the board provides information to directors before board meeting (infopre), a dummy variable assuming value equal to one if delegated powers are defined by topic and/or by a monetary upperbound (limdel), a dummy variable assuming value equal to one if deals with related counterparts are approved ex ante by the board (par), a dummy variable assuming value equal to one if the company defines a procedure to disclose information to the market (infodis) and a dummy variable assuming value equal to one if an appointment committee has been introduced (comnom) Exogenous Variables We have collected data on qualitative features of companies, on their performance balance sheet and on ownership structure. As far as ownership structure is concerned, information is gathered by CONSOB official releases. Each stake above 2 per cent is identified by the percentage of shares of the company, the direct shareholder and the ultimate shareholder

13 E. Barucci and J. Falini: Italian Financial Market 383 (dichiarante). The direct shareholder in many cases is another company; the ultimate shareholder is the relevant one. In case of a family, we have grouped all the stakes owned by people of the family and we assign it to the family as a whole unless it is well known that members of the family do not agree on the management of the company. In case of a person who detains a certain stake directly and a stake through a company, we have grouped them and we assign the stake to him. As an additional information, we consider details on shareholders agreements (patti di sindacato) that were in force by the end of 2001 and 2002; details on coalitions involving shareholders of listed companies are made public since the TUF on the official web site of CONSOB. Note that there are many types of coalition; a coalition may deal with voting at the general shareholders meeting, talks about the management of the company and decisions to be taken by the board of directors, and option or preemptive duty on selling shares. Shareholders stakes are grouped in two categories: the controlling shareholder coalition or the largest shareholder (lrgshr) and shareholders with a stake above 2 per cent different from lrgshr (blockhol). It is difficult to classify a shareholder as belonging to lrgshr or to blockhol. We proceed as follows: in companies classified by CONSOB as non-controlled, we consider lrgshr as the stake detained by the largest shareholder; in companies controlled by absolute majority or de facto with no coalition agreement, we consider lrgshr as the stake detained by the largest shareholder; in companies characterized by a coalition among shareholders (independently of its nature), lrgshr is given by the stake of all shareholders participating to the coalition. We assume that any form of patto di sindacato is relevant to define a coalition controlling the company. In the literature, it is claimed that institutional investors holding a large stake of the company and having a long period view should be particularly keen on the governance of the company. To address this point, we consider the stake detained by all institutional investors (inst). Depending on the controlling shareholder, we consider the following dummy variables: synd assuming value equal to one if the company is controlled by a shareholders coalition and state assuming value equal to one if the controlling shareholder is the state or a local authority. Note that when the company is controlled by a shareholders coalition, there is a wedge between voting rights and cash flow rights, i.e. for control it is relevant the fraction of shares inside the coalition. 8 Entrenchment of the management is evaluated considering a dummy variable assuming value equal to one if the largest shareholder is either a CEO or the chairman of the company (down). Qualitative features of a company are important in our analysis. In what follows, we consider the following dummy variables: ipo assuming value equal to one if the company is listed in the market during the last

14 384 Economic Notes : Review of Banking, Finance and Monetary Economics 6 years, privat assuming value equal to one if the company has been privatized during the last 10 years, dual assuming value equal to one if the company is listed in UK or US (there are only seven companies listed in these markets) and pyram assuming value equal to one if the company belongs to a pyramidal group. Balance sheet data are collected from the Calepino dell azionista. We consider the market-value-to-book-value ratio as performance indicator. To evaluate the robustness of the results, we have also considered the average market value over the book value of the industrial sector of the company (mvbvs); results are quite similar and therefore have been omitted. We control for the size of the company through the logarithm of sales (logsales) Descriptive Analysis In Table 2, we present some descriptive statistics on compliance of companies with the code of best practice and on shareholders rights in 2002 and In general, the rate of compliance with the code of best practice is high, but it is low on some crucial features. It is quite rare to observe an independent chairman of the board of directors. While a remuneration and an internal control committee are established by a large fraction of companies, in a small fraction of companies, we observe a fully independent committee. Being ownership concentrated, an appointment committee usually is not introduced. Appointment of directors and of members of the board of auditors represent a critical governance area. The voto di lista mechanism is adopted by a small fraction of companies. The threshold to present a list to appoint members of the board of auditors is quite high (the median is 3 per cent) compared with the stake detained by institutional investors and outside blockholders (the median is 3 per cent and 5 per cent, respectively). As a consequence, only 1/4 of companies are characterized by a board of auditors with members appointed by minority shareholders. Overall, we observe that companies belonging to the MIB30 are characterized by a governance better than companies belonging to the entire market. On average, the fraction of independent directors, the likelihood of observing an independent chairman and the likelihood of observing a (fully independent) remuneration internal control committee is higher restricting our attention to MIB30 companies than for the entire sample. Minority shareholders are also better protected, i.e. the voto di lista is more frequent and the threshold to present a list to appoint members of the board of auditors is lower. However, companies belonging to the MIB30 are characterized by a higher voting rights cash flow rights wedge than companies of the entire sample.

15 385 Table 2: Descriptive Statistics on Corporate Governance and Exogenous Variables Governance variables Mean Median Mean MIB30 Mean 2002 Mean 2003 p-value Median 2002 Median 2003 p-value boardsize ceochair chairind comcon comconi comnom comre comrei excl fracind infodis infopre inform limdel lista listasin numsin par sindmin blockhold synd state dual inst ipo lrgshr mvbv privat pyram Notes: The first and the second column refer to the entire sample. The third column refers to MIB30 companies on both years. The fourth and fifth column provides the mean of the 2002 and of the 2003 sample, respectively, and the sixth column provides the p-value on their difference; the seventh and the eighth column refer to the median of the 2002 and of the 2003 sample, and the last column provides the p-value on their difference (for non-dummy variables).

16 386 Economic Notes : Review of Banking, Finance and Monetary Economics As far as governance standards in 2002 and 2003 are concerned, after a revision of the code in July 2002, we observe weak effects; for a positive evaluation, see Assonime (2004). The revision of the code has imposed a more restrictive notion of independent directors. Probably, this has caused a smaller fraction of independent directors in 2003 than in 2002 (the difference on median values is statistically significant). In general, no statistically significant effect is observed; we only observe that the chairman of the board of directors in 2003 is independent with a higher probability than in It is also more likely to observe a fully independent internal control committee. 4. Corporate Governance Determinants: Hypotheses and Models Little research both theoretically and empirically has been developed on corporate governance determinants. Durnev and Kim (2003) provide a model on determinants of governance. The quality of governance is supposed to be inversely related to the proportion of firm value diverted by the largest shareholder. They show empirically and theoretically that companies with better investment opportunities, higher concentration of ownership and greater needs of external financing practice better governance. A large outside blockholder and institutional investors, when they are not the controlling shareholders, should positively affect the governance quality. In fact, they detain a large stake and are often characterized by a long-term view and therefore they should be interested in pursuing the shareholder s value, in exerting their voice and in monitoring the governance of the company (monitoring hypothesis) (Kahn and Winton, 1998; Maug, 1998). Note that in many financial companies, the largest shareholder often is an institutional investor, and therefore to fully appreciate the monitoring role of institutional investors, we should look at the nonfinancial company sample. Empirical evidence on the role of institutional investors is weak. In the Italian financial market, there is some evidence that institutional investors prefer companies with a good governance and that their stake positively affects corporate value (Barucci and Falini, 2004; Barucci and Ceccacci, 2005); however, in general, there is weak evidence on their effective role to enhance the sensitivity of CEO turnover to performance, and the same thing can be said about outside blockholders (Denis et al., 1997; Franks et al., 2001; Renneboog and Trojanowski, 2003). As the stake of the largest controlling shareholder increases, we observe two competing effects on the management of the company and on its governance; on the one hand, managers interests and those of the controlling shareholders are more aligned and therefore agency costs

17 E. Barucci and J. Falini: Italian Financial Market 387 should be reduced with the management focused on the (largest) shareholder s value (incentive effect); on the other hand, a large minority shareholders conflict arises because managers controlling shareholder pursue private benefits and managers are entrenched, i.e. they look for perquisites (private benefits-entrenchment hypothesis). While the effect on companyvalue management is ambiguous, the effect on the governance of the company is likely to be negative, i.e. a company with the largest shareholder detaining a large stake or under strict control will be less inclined to introduce governance rules that limit private benefits of control and to protect minority shareholders. Both effects should be stronger when the largest shareholder is also an executive of the company. Note that theoretical and empirical results provided by Durnev and Kim (2003) show that the incentive effect should prevail, but Doidge et al. (2004) show that there 10 is a negative relation between ownership concentration and quality of governance. Moreover, Barucci and Ceccacci (2005) show that company value is negatively related to the stake of the largest shareholder. Companies with good growth opportunities should pay more attention to corporate governance. In fact, these companies have greater needs of external financing. To collect funds in the equity-debt market, the company is more likely to adopt a high-quality governance and a better minority shareholder protection to attract outside investors with a lower cost of capital. Evidence confirming this hypothesis has been obtained by Durnev and Kim (2003), Doidge et al. (2004) and Klapper and Love 11 (2002). As far as size is concerned, we expect large companies to have a better governance because they have greater agency/asymmetric information problems and therefore are more difficult to monitor; see Klapper and Love (2002) and Doidge et al. (2004) for empirical evidence confirming 12 this hypothesis. If the private benefits-entrenchment hypothesis prevails over the incentive effect, then a wedge between voting rights and cash flow rights of the largest shareholder (through a pyramidal group or a shareholders coalition) should negatively affect the governance quality. According to the functional convergence thesis, governance convergence can occur by listing in countries with high investor protection standards, i.e. companies listed in these countries are forced to adopt a good governance and high investor protection standards. A similar argument applies to companies that recently entered the market. As a consequence, we expect companies listed in UK/US and companies listed recently in the market to have a good governance. As far as control by state or privatized companies is concerned, we expect two different effects: protection of shareholders rights should be at a high level, but on the other hand, when the state controls a company, the management is not necessarily focused on the shareholder value, and therefore the internal governance of the company may be poor.

18 388 Economic Notes : Review of Banking, Finance and Monetary Economics It is difficult to build an aggregate index of the governance of a company; in our analysis, we analyse the determinants of specific corporate governance features. We consider three main governance areas: composition of board of directors and selection of directors, activity of the board of directors and board of auditors. In section 5, we provide an analysis of each governance feature. Our research strategy is to analyse each corporate governance feature varying exogenous variables in order to test different hypotheses and to investigate the robustness of the results. As far as the model is concerned, we proceed in three steps by building a reference model and an extended model estimated on the 2002 sample and then estimating the extended model on the 2003 sample. In the reference model, we control for the size of the company, measured through the logarithm of sales (logsales), and then we consider ownership structure by inserting the stake of the largest shareholder (lrgshr), the stake of outside blockholders (blockhold) and the stake of institutional investors (inst); growth opportunities and company value are measured through the market-value-to-book-value ratio (mvbv). Note that mvbv does not only proxy growth opportunities but also company value, and there is some evidence that a good governance positively affects company value (Klapper and Love, 2002; Black et al., 2003; Drobetz et al., 2003; Durnev and Kim, 2003; Doidge et al., 2004). To capture 13 entrenchment of the board, we consider the dummy variable (down) assuming value equal to one when an executive of the company (CEO or chairman) is also the largest shareholder. We control for industry effects by including industry dummy variables. Then we consider an extended model. The extended model is made up of the exogenous variables included in the reference model and of variables addressing qualitative features of the company. We control for companies recently listed in the stock exchange market through the dummy variable ipo, companies listed in the UK or in the US financial market through the dummy variable dual, companies privatized since 1992 (privat), companies controlled by the state (state), companies controlled by a shareholders agreement (synd) and companies belonging to a pyramidal group (pyram). 5. Determinants of Corporate Governance In this section, we provide the results on the determinants of corporate governance through three subsections: composition of board of directors, activity of board of directors and board of auditors. We estimate the reference with industry effects and the extended model on the 2002 sample (Tables 3, 4 and 5) and the extended model on the 2003 sample (Table 6). Then in subsection 5.4, we provide a synthesis of the results.