Activism of institutional investors, corporate governance alerts and financial performance.

Lantz, Jean-Sebastien ; Montandrau, Sophie ; Sahut, Jean-Michel 等

I. INTRODUCTION

Since the mid-1990s, American fund managers have dropped their passive "Wall Street Rule" management style at general shareholders' meetings and replaced it with an 'active' style. The aim of these proxy battles is to ensure that the principles of corporate governance are applied in order to protect shareholder wealth by strengthening controls over the firm's managers. In the United States, the large pension funds (Calpers (2), TIAA-cref, (3) etc.), as well as associations such as the USA and CII, (4) were the first to adopt such practices. This approach even leads some to develop a reputation as actively 'hostile' pension funds due to their regular communication of black lists of companies which they advise against investing in. Empirical research, including the most recent studies (Song et al., 2002; Gillan and Starks, 2003) has attempted to demonstrate the impact of the various forms of activism, but the results to date, chiefly measured by the firm's financial performance, are open to debate.

In France, two elements can explain different corporate governance practices in big companies. Firstly, the American pension fund system does not exist. Secondly, we can observe a network of intricate links between the State and large industrial firms (most CEO's are former civil servants or government officials, and companies make use of, and abuse, pyramidal structures in the large French groups in order to retain control and prevent foreign takeovers). However, in recent years, individual shareholders' associations and the Association Francaise de la Gestion Financiere (the AFG) (5) have tried to protect the interests of shareholders with the introduction of various initiatives. Since 1999, the AFG has sporadically published alerts regarding major French listed companies for non-compliance with the rules set out in the association's code on best practices in corporate governance, drawn up in 1998. The aim of these alerts is to mobilise fund managers and encourage them to vote against resolutions proposed by the management board of the company in question at future general shareholders' meeting.

The aim of this article is twofold; first, it sets out to assess the short-term effects of the activism of French institutional investors, and second, it attempts to demonstrate that the effects of an alert can differ according to the target company's past performance. Our empirical study uses the methodology of event studies, and attempts to measure the impact of AFG alerts regarding major French companies during the period 1999 to 2002.

The article is divided into three parts. The first part presents the arguments developed in the relevant literature in order to formulate the hypotheses to be tested empirically. The second part investigates the choice of the AFG alerts and the event studies methodology. In

the third part, we present and discuss the short-term results of the activism of French institutional investors.

II. THE ACTIVISM OF INSTITUTIONAL INVESTORS AND THE INFLUENCE OF CONTEXT VARIABLES

A. Theories behind Activism

Activism can be defined as involvement by institutional investors to influence the management of companies in their investment portfolio, following a conflict between the shareholders and the company's management. Thus, the theoretical framework of activism is directly linked to the theory of corporate governance, agency theory and the theory of the entrenchment of managers. As Romano (2001) argued, "In brief, there is an apparent paradox: notwithstanding the development of shareholder activism and commentators' generally positive assessments of it, the empirical research indicate that such activism has little or no effect on targeted firms' performance." Recent studies, such as the one conducted by Thomas and Cotter (2007), have identified small and insignificant market reaction, suggesting that the activism of institutional investors apparently only leads to minor changes in a firm's corporate governance structures and does not measurably affect stock prices or earnings (Barber, 2006; Del Guercio et al., 2006; and Gillian and Starks, 2007).

In France the theoretical framework of activism is limited. In particular, Girard (2003) considers that it chiefly involves lobbying activities with regard to a legal process. In our view, however, over the last few years, there has been a progressive shift in terms of institutional investor behaviour towards proxy contests, which also involve media coverage of the conflict.

We began therefore by addressing the question: "is the activism of French institutional investors neutral from the point of view of financial markets?" If not, what are the factors which explain a positive or negative market reaction at the time of an activism process?"

As several literature surveys have shown, the results of activism have been disappointing. In this context, traditional empirical literature on activism has put forward several reasons for predicting a sometimes positive and a sometimes negative reaction, as illustrated below.

Existing literature on the subject highlights several factors that help to explain a positive reaction. First of all, research on management changes indicates a positive impact which depends on the economic context or the planned restructuring involving the target company (Furtado and Rozeff, 1987; Mikkelson and Partch, 1997). Pound (1992) sets out a "new policy model" based on proxy battles, in which dissatisfied shareholders no longer hesitate to contest decisions or revoke the mandate of entrenched company managers. Del Guercio and Hawkins (1999), and Prevost and Rao (2000) show that any proposal made upstream of the general shareholders meeting leads to a massive vote by dissident shareholders or helps to avoid an open conflict which is the subject of media coverage and can prove extremely costly for the company.

Recent studies have also identified the conditions necessary for a positive reaction. According to Solh (2000), if a fund manager adopts a long-term management approach in addition to his monitoring role, he will influence the long-term investment decisions and encourage the company's management to choose the optimal projects from the point of view of shareholder interest. Woidtke (2002), on the other hand, argues that bonuses and promotions provide private fund managers with strong incentives to adopt an activist approach, fostering a convergence of interests with other categories of shareholders.

In such a context, any announcement of institutional investor activism should generate a favourable market reaction.

H1: The activism of institutional investors implies a positive reaction to the stock market price of companies targeted for non-compliance with the corporate governance criteria.

We shall now turn to the factors that explain a negative reaction. To begin with, we should emphasise the different objectives of fund managers and other categories of shareholders, the former preferring to attain their personal objectives at the expense of those of the latter. Barber (2006) argues "that institutional activism should be limited to shareholder activism where there is strong theoretical and empirical evidence indicating that the proposed reforms will increase shareholder value ... Portfolio managers should pursue the moral values or political interests of their investors rather than themselves." The result can depend on the variety of the fund. Unlike mutual funds and pension funds, Brav et al. (2008) find that US hedge funds are able to influence corporate boards and managements due to key differences arising from their different organizational form and the incentives that they face (highly incentivized managers and concentrated positions in small numbers of companies).

In addition, any threat by fund managers to withdraw their funds sends out a negative signal to the market linked to the appearance of additional costs, as well as a risk of profit-taking by fund managers. Charreaux (2003) points out that there is a timeframe that is necessary for a company's managers to develop the necessary competencies, without which investors will not be able to make their investment profitable. At the same time, if there is too much pressure on company managers, there is a risk that they might decide to delist the company, thereby depriving shareholders of the monitoring powers that accompany a stock market listing, which is also a source of market financing.

The third model of reflection, concerns the different stages of the institutional investors' activism. Smith (1996) demonstrates that the company's book value increases in the case of target companies that have concluded an agreement with Calpers, and falls for companies that have not signed such an agreement. It would therefore seem that when a conflict is given media coverage, the market sanctions the reluctance of company managers to accept organisational change. Thus, while the hope of seeing the problems resolved is initially strong, a succession of failures and the increasing proximity of the date of the general shareholders' meeting send out a signal to the market of increasing difficulties in resolving the conflict.

H2: The activism of institutional investors implies a negative reaction of the stock market price of companies targeted for non-compliance with corporate governance criteria.

B. Empirical Studies on the Results of Targeting in Terms of Financial Performance

In this section, we set out the results of the companies targeted by pension funds in the United States in terms of performance.

Karpoff et al. (1996) studied their performance by using measures such as "book-to-market" ratio, operating profit, and sales, and tried to identify differences in reactions using univaried tests. They showed that shareholder resolutions have very little influence on the operating result, the stock value or the rotation of the company's managers. These results remain unchanged, even when the resolution is adopted by a majority vote of the shareholders at the general meeting. Wahal (1996) showed that for a series of public funds, activism has a positive effect related to performance, when it is assessed in terms of return on equity and return on assets.

Using the lists circulated by the CII, Opler and Sokobin (1997) studied the long-term influence of the publication of a list of underperforming companies for the period 1991-1994. The criteria used to measure performance are the return on equity and the return on assets. They show that the effects of activism are highly positive and significant over the long term. Two other, more recent American studies have built on the study by Opler and Sokobin. The research conducted by Caton et al. (2001) to ascertain the short-term effects of the activism of the CII over the period 1991-1995, confirms the short-term impact of the publication of a list of underperforming companies, especially when the level of performance is assessed according to Tobin's Q.

The second study, by Song, Szewczyk and Safiedine (2002), extends the period of study until 1996. The authors concluded that the activism of the CII is not an effective method of increasing a company's stock market value. In France, Herve (2003) replicated the last two studies by observing the short-term and long-term effects of the activism of the CII over the period 1991-1998. He found a positive, significant impact on the performance of the target companies in the long term. On the other hand, over the short-term, his event studies did not reveal any positive, systematic effect on the stock market performance of the target companies. By way of conclusion, Romano (2001) found that shareholder proposals resulting from activist movements since the mid-1980s did not have any positive effect on the company's performance.

C. The Intensity of the Signal Issued

The recommendations made by the Vienot reports (1995) and the Law of 15 May 2001 on the New Economic Regulations identified several variables likely to be of interest to active institutional investors. In particular, the higher the number of poor management practices revealed by the activism of institutional investors, the greater the intensity of the signal received by the financial markets.

H3: The more reasons invoked for a corporate governance alert, the more negative the impact on the financial markets.

D. The Signal Effect Sent out by Performance

Studies of the information asymmetry show that minority and external shareholders pay special attention to annual reports and their publication. Accordingly, if activists increasingly use past performance as a criterion for the selection of targets, we cannot ignore the question of the risks for shareholders resulting from the manipulation of the accounting results.

If the accounting performance is poor, it will be in the interests of the company managers to window dress the accounts in order to make these results less visible and less transparent for the minority and external shareholders. To reduce the significance of the financial indicators, company managers have several means at their disposal to manipulate the figures. They can firstly disclose 'optimistic' information, which does not reveal the fall in profits and therefore misleads shareholders. Ali et al. (2000) shows that in companies that were subject to legal proceedings between 1988 and 1990, managers of companies with financial problems communicated accounting information that had been manipulated and was therefore opaque in order to avoid shareholders selling their stake in the company or any contestation by dissatisfied shareholders. However, this kind of manipulation reduces the possibility for shareholders to realise a capital gain, since as Lambert et al. (2005) point out, the specificity of shareholders' investments lies in the mobility of their stock holding linked in particular to liquidity. In fact, the protection offered by this capital is only valid ex-ante. Ex-post, it can only lead to capital loss. It should be borne in mind that if a company is in financial difficulty, the shareholders often lose out when the company's net assets are distributed among creditors precisely because, as shareholders, they have the last claim against assets.

Consequently, it is all the more in the interests of institutional shareholders to adopt an activist approach when the company's accounting indicators are weak. This is precisely the approach pursued by Calpers, the USA and CII. Romano (2001) confirms the interest of targeting underperforming companies, considering that the better the performance of the target company (first third in a league table by performance), the more likelihood there is of the impact of a shareholder's proposal being compromised, since there will be less scope for such companies to significantly improve their performance.

Moreover, the choice made by shareholders to sell or hold onto their shares is also linked to the information disclosed by the financial markets as the latter sanction poorly performing company managers. Thus, any fall in the stock price will make minority and external shareholders aware of a major loss in terms of their remuneration, in other words, the capital gain realised on the stock. Likewise, any fall in financial markets will lead to liquidity problems for certain companies when a fund has a large stake. If the latter wants to sell its holding, it will have to accept the fall in the share price resulting from the significant impact that any decision to sell will have on the financial markets (Chan and Lakonishok, 1995). At the same time, institutional investors have the alternative of spreading a stock sale over a period of time, which will help them to avoid liquidity problems. It is therefore in the interests of institutional investors to adopt an activist approach whenever a company's stock market performance is poor. In addition, such targeting of a company by institutional shareholders due to the company's poor performance can represent an opportunity to avoid selling costs or more important residual losses when selling the stock.

Firtha et al. (2007) demonstrates that accounting indicators are used by dissatisfied shareholders to highlight the company's poor management performance to passive minority shareholders, so that the latter can add their dissenting voice at the general meeting of shareholders. DeAngelo and DeAngelo (1989) have shown that dissident coalitions prefer to use accounting indicators rather than market measurements, since the measures of activists have a positive impact on stock market returns. According to Gordon and Pound (1993), shareholders do not have specific expertise for calculating abnormal returns and take these results into consideration when voting at general meetings. Accordingly, they base their opinion on three elements: the target company's economic performance, the identity of the dissident shareholders and the type of resolution on which they are asked to vote. They conclude that accounting performance is a good market indicator of the management team's ability to improve shareholder wealth.

At this stage of the analysis, the object of our hypothesis is not to show the long-term results of activism, but rather to reveal a possible signal effect of the target company's past performance in the event of a corporate governance alert, hence the following hypotheses:

H4: An AFG corporate governance alert on a company that has a poor past financial performance has a negative impact on its market returns.

Because of the diversity of performance indicators used in the empirical literature, we decided to select several of them, depending on the various cleavages underlined by Dherment-Ferere and Renneboog (2000):

(1) performance indicators according to the maximisation of shareholder wealth:

* ex-ante measures: (6) we firstly used Marris' ratio (H4.e), as well as a measure based on portfolio theory: (7) Treynor's index (H4.f). The latter has the advantage of taking the risk incurred in measuring performance into account.

* ex-post measures: (8) two ex-post accounting measures were chosen to assess the return on equity. The return on equity assessed by book value (H4.a) and the variation in the return on equity assessed by book value (H4.b).

(2) performance indicators according to the maximisation of the company's overall value:

The first ex-ante measure used is Tobin's Q (H4.d), together with an ex-post measure represented by the return on capital employed, assessed on the basis of book values (H4.c).

III. METHODOLOGY AND DESCRIPTIVE STATISTICS

A. The Data and Study Methodology

Since 1999, the AFG has published alerts regarding major French listed companies for non-compliance with the rules set out in the association's code on best practices in corporate governance (9). The AFG initiated activism, thus encouraging its 1500 members to vote in the annual meetings. The only political motivation (no incentive) for the AFG in the selection of companies which are the object of alerts is to incite the fund managers to vote at the annual meeting. Our study covered the first 107 alerts relating to listed firms included in the SBF 120 market index, issued between 15 April 1999 and 28 March 2002. The AFG lists 9 reasons for calling into question a company's corporate governance policy. These mainly involve any anti-takeover measures introduced (38% of cases), plurality of mandates (23% of cases) and the absence of specialised committees (17% of cases). Other reasons remain fairly scarce, i.e. appointment of an excessive number of directors, absence of information on a candidate director, absence of specialised committees (audit, remuneration, etc.), the age of a director, or the amount of money a director receives. Moreover, 73% of the companies involved have received only one alert. In fact, it is not possible to survey the incidence of repeated alerts on a sufficiently large subsample. In order to calculate past performances, we collected book and market values. Book values rely on the accounting data published in both the last annual report available and the previous one, in order to calculate the variations.

We used the event study methodology developed by Brown and Warner (1985), with daily abnormal return (AAR). Concerning the possible perturbations of other events around the date of the announcement, the method used is generally designed to minimize these effects (i) by focusing only on abnormal returns, thus precluding the effect of systematic factors on the stock exchange and (ii) by aggregating samples of announcements at different dates, which randomizes and averages out other possible noisy effects to zero.

Our event period goes from date (-30) to date (+30), on eight observation windows (see table 1). By convention, date 0 is that of the issuance of the alert on the AFG web site. This date is relevant because it is the only source of information for the market. As the general meeting is scheduled 40 days after the issue of the alert, we have limited our survey to this period.

Abnormal return is measured by the difference between the stock's actual return ([R.sub.it]) and a theoretical return (corresponding to the return on the stock if the event announcement had not been made) on a given date (t). The average abnormal return on date [t.sup.10] (AAR) and the average cumulative abnormal returns between dates t1 and [t.sub.2] (ACAR([t.sub.1], [t.sub.2] )) is calculated as follows:

[[AAR].sub.t] = 1/N [[i=N].summation over (i=1)][RA.sub.it] (1)

where [RA.sub.it] = the average abnormal return of stock i on date t and N = the number of companies in the sample.

In the framework of event studies, the hypothesis Ho is the following: the AAR, whether accumulated or not, are equal to zero and the model used to define abnormal returns is clearly specified. We first need to test whether the average abnormal return for the total sample is different from zero. Cai and Xu (2006) calculate average abnormal returns as follows:

We then have to test whether the average cumulative abnormal returns (ACAR) between two dates in the event period are different from zero. To this end, a second statistic is calculated:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

where [T.sub.0] = the initial date of the estimation period, [T.sub.e] = the final date of the estimation period, and T = the length of the estimation period.

We used the three most commonly found models in the literature: the market index, the market model and the CAPM. (11)

Market model: [R.sub.it] = [[alpha].sub.i] + [[beta].sub.t] [R.sub.mt] + [[epsilon].sub.it]

Market index: (12) [R.sub.it] = [R.sub.mt] + [[epsilon].sub.it]

CAPM: [R.sub.it] = [R.sub.f] + [[beta].sub.t] ([R.sub.mt] - [R.sub.ft]) + [[epsilon].sub.it]

where [[alpha].sub.i] = estimation of the parameter whose value is such that the expected value of [[epsilon].sub.it] is nil (or expected value of [R.sub.it] when [R.sub.mt] is nil), [[beta].sub.t] = estimation of the volatility coefficient or beta coefficient, which is specific to each stock i and indicates the relationship which exists between fluctuations in stock i and fluctuations in the general market index (a measure of the systematic risk of stock i), [[epsilon].sub.it] = random residual variable specific to share i; its standard difference is a measure of the stock's specific risk. These coefficients were estimated over a period of 270 days before the event period. [R.sub.f] = risk-free interest rate, [R.sub.it] = return on stock i in t, [R.sub.mt] = market return measured by a general index in t.

B. Statistical Description of the Data

The Komolgorov Smirnov test indicates that the distributions of the abnormal returns differ from a normal law (right skewness and flattening coefficients higher than 1, which is fairly frequent in the framework of daily returns). However, in their 1985 study, Brown and Warner conclude that "the non-normality of daily returns does not have any impact on the effectiveness of the event study methodology. In fact, the average abnormal return calculated on a series of stocks converges rapidly towards normality when the number of stocks increases, this result being obtained using a sample size of 50 stocks."

The test for the autocorrelation of the remainders shows that the remainders are independent from one period to another. This stems chiefly from the fact that the announcement dates often differ from one company to another, which eliminates the crude dependency adjustment problems referred to in the methodology section.

Problems with the stability of beta are moderate because the estimation period only covers 240 days. (13) Finally, the homogeneity of variance tests of Fisher show that the variance does not increase between the estimation period and the event period. (14)

IV. RESULTS OF THE EMPIRICAL STUDY

A. Impact of an Alert for Non-Compliance with the Principles of "Corporate Governance"

1. Impact on the Stock Return of an AFG Alert

Table 1 shows that the average abnormal return that was positive before the announcement, becomes negative from date +1 (-0.54%), a significant result at 10%. In windows (+1,+20) and (+1,+30) the negative impact accentuates up to -3.21% (significant at 2%) and -3.32% (significant at 5%). Brown and Warner (1985) specify that in studies where the event days do not intersect, the power of the market index is similar to another more elaborate model. We therefore focus first of all on these results.

In line with the hypothesis of a negative signal being sent out on the financial markets to shareholders and other stakeholders, the impact of a corporate governance alert is negative (-0.54% over a session) and statistically significant at a threshold of 10% from the date of receipt of the alert mail by fund managers. This trend persists and leads to negative ACAR (-3.32%) over the 30 sessions following the announcement. These results corroborate the conclusion of the second tests carried out by Prevost and Rao (2000): the longer the conflict lasts, the more the market perceives a negative signal regarding the company's performance. This corresponds to our case since the AFG alerts appear on average 40 days before the general meeting of shareholders, which is relatively close to the date of the opening of the conflict if applicable. Moreover, as alerts are formulated from resolutions on which shareholders are to be asked to vote at the general meeting, the situation goes far beyond the simple stage of private negotiations between shareholders and the company's management.

The principle whereby the closer we move to an official conflict at the general meeting, the greater the likelihood of a negative market reaction, is in line with the conclusions of Smith (1996), for whom, if no agreement is signed between the fund and the target company, the company's value will fall and vice versa.

We therefore validate hypothesis H2: the activism of institutional investors implies a negative reaction of the stock market price of companies targeted for noncompliance with corporate governance criteria.

2. Impact According to the Number of Non-Compliant Resolutions Having Motivated the Alert

The tests carried out to asses the impact of an alert according to the number of noncompliance grounds show in Table 2 that this is almost nil or insignificant when only one reason is invoked. If the AFG bases an alert on several grounds, the impact is highly negative (-8.06%) over the next 30 days and significant at 2% for the market index. The results obtained using the market and CAPM models confirm this trend but are only significant at 5 and 10%.

Thus, the more institutional investors note 'infringements' of the corporate governance recommendations, the more the market sanctions the target companies by reactions involving a fall in stock market prices. This in turn leads to an anticipation that the probability of the situation being turned around will be lower or that higher costs will be incurred. These conclusions enable us to validate hypothesis H3: the more reasons invoked for a corporate governance alert, the more negative the impact on the financial markets.

B. Influence of Past Performance on the Impact of the Publication of a Corporate Governance Alert Regarding A Target Company

The total sample was divided into two groups according to the median of each performance variable. Table 3 shows the results of the event studies carried out for each performance variable using the market index.

1. Influence of Past Performance Measured by the Return on Equity by Book Value or Their Variation over the Last Financial Year

If the target company belongs to the sample below the median in terms of return on equity by book value, the negative reaction of the financial markets increases following the publication of the alert. These results are significant and equally valid from the first date of the announcement and for the following thirty days. When the performance is good, the market reacts very weakly since there is a reduction in the ACAR which become negative for several days, but very rapidly become positive again. Therefore, we validate hypothesis H4a.

Our conclusions corroborate the results of the American studies (Wahal, 1996; Bizjak and Marquette, 1997; and Opler and Sokobin, 1997) which found that the target companies had an accounting performance significantly below that of the companies in the control sample. However, these studies used the return on assets or the market-to-book ratio. The only study which used the return on equity by book value was that of Strickland et al. (1996). The authors did not find any difference in reaction when using this performance measure.

The impact of a corporate governance alert is significantly more unfavourable to companies whose return on equity by book value has fallen over the last financial year. We go from a positive, significant ACAR of +2.26% in the window (-30,-10) to a low significant negative ACAR of 5% for the last three windows (-2.54%, -3.77%, -4.82% respectively). For these targets, we therefore note a severe, constant price fall, as if the market which expects good news (performance in line with that of the previous year) sanctions the bad news associated with the alert concerning corporate governance criteria more severely. For companies whose return on equity is considerably higher, the negative abnormal returns are not significant, but a graph would show that the ACAR are almost always below zero over our analysis period. This could lead us to conclude that when the performance is good, investors react negatively, although extremely progressively, when an alert is announced. A good performance cushions the negative signal (due to the publication of the alert) on the financial markets.

Thus, while our study methodology does not enable us to verify that the target companies under-perform the market, we nonetheless show that the greater the deterioration of past performance, the more the target company's share price will fall after an alert. The two negative announcement effects combine to signal 'doubly' bad management to the market. This result, combined with that of the previous test, calls into question the conclusions of Romano (2001) who emphasised the higher probability of a positive impact in underperforming (since it would be easier to improve their performance).

Moreover, we confirm the results of the study conducted by Lambert et al. (2005), for whom the risks incurred by shareholders in the event that the company performs badly are high because of the low possibility of being able to withdraw from the company or realise a capital gain if the information regarding poor accounting performances was, in addition, manipulated to look better. At the same time, the proportion of capital owned by investors increases the risk of illiquidity and forces them to spread the sale of their stock (Chan and Lakonishok, 1995), which could be intensified in the event of twofold bad management signalled to the market, as we have just seen. These results lead us to confirm hypothesis H4b.

2. Influence of Past Performance Measured by Return on Capital Employed, Tobin's Q and Marris' Ratio

The value of a group's shares and debts is based on the value of the capital employed. The capital employed is financed exclusively by equity capital and net bank and financial debt. (15) In Table 4, the tests carried out on a company with a low return on capital employed show a stronger and more significant impact of an alert (-3.24% for (+1,+20) and -5.01% for (+1,+30)). Nevertheless these market reactions are not as significant as those based exclusively on measuring the return on equity, which is the ultimate measure of performance for shareholders. The share price of target companies with strong return on capital employed also falls at the time of the alert, but the fall is not as pronounced, and above all, it reverses and becomes positive over the last ten days of the observation period, which means that the abnormal returns are never significant.

In accordance with Table 4, we can corroborate hypothesis H4c: there is a negative relationship between past performance and the negative impact of an AFG corporate governance alert, this being measured by the return on capital employed. Tobin's ratio is in principle a measure of all the anticipated returns over an infinite time horizon. A ratio above one means that the return that can be generated by all the company's assets and anticipated by the market is higher than the average weighted cost of the capital, hence the conclusion that the company is outperforming.

Unlike the event studies conducted using ex-post measures of performance, the tests carried out by dividing the sample according to the median of Tobin's Q (value of 1.2) reveal a far more striking negative reaction (-4.35%) for the windows (+1,+20), which is significant at 5% for outperforming companies. This stronger decline in comparison with that of the sample of underperforming companies (which had a more progressive decline) shows the dissatisfaction of investors who anticipated a good performance and who learn that the proposed resolutions to shareholders could undermine these prospects over time. On the other hand, in the case of underperforming companies, we observe a significant negative market reaction for windows (+1) and (+1, +30), then stabilisation. In fact, the impact of the alert does not compromise the somewhat poor projected performance. On the contrary, investors may see in this form of activism a possible modification of the company's governance which could in time modify its performance. We therefore invalidate hypothesis H4d.

As with the tests carried out using Tobin's Q, the measure ofperformance based on Marris' ratio (which uses only the capital directly invested by the shareholders and is therefore in line with the maximisation of the company's equity capital), produces results that run counter to the assumption made. In fact, the negative impact of the alert is more important for outperforming companies. When ex-ante measures of performance are used, the market revises the valuation of outperforming target companies downwards, as if the forecasts had gone up in smoke with the poor governance revealed by the AFG alerts. In consequence, we do not corroborate hypothesis H4e.

3. Influence of Past Performance Measured by Treynor's Index

Treynor's index is a relative measure of performance calculated ex-ante (risk premium for the future) that aims to assess the level of maximisation of the equity capital value of the target company. In Table 5, the results of event studies produced for companies with a negative Treynor's index (underperforming) show that the target companies record, from date +1, a negative reaction of-0.67%, which is not, however, significant (t=1.35). On the other hand, for windows (+1,+20) and (+1,+30), the results are extremely significant since the prices generate abnormal returns of-6.71%, significant at 2% and -8.58%, also significant at 2%. (16)

In the sample of companies with a positive Treynor's index, the abnormal returns are almost nil and the Student's t coefficients are very low. This result corroborates the hypothesis whereby the market is more conciliatory towards companies that offer the hope of future positive flows.

By using Treynor's index it is possible to measure the 'reward' or risk premium for investors by unit of risk, the latter being defined by the beta of the stock (systematic risk). The use of such a performance index in our study enables us to compare the results of event studies based on ex-ante measures (expectancy of future flows) and those calculated ex-post (a posteriori). If the three previous measures did not enable us to validate our hypotheses, the results of these event studies show a very strong difference in the reactions of the financial markets for our two samples. In fact, while outperforming companies do not record significant abnormal returns, the targets which have a lower possibility of increasing shareholder wealth in the future (companies with a negative Treynor Index) have an ACAR that is extremely negative and significant at 1% over the twenty and thirty days which follow the alert. This reaction, which remains the most marked of all the performance measures used, shows that the market remains very attached to the combination of risk and return. Thus, any company which does not offer a risk premium to investors or corporate governance practices in line with the principle of maximising shareholder wealth will immediately be doubly sanctioned. We therefore corroborate hypothesis H4f.

V. CONCLUSION

Independently of all context variables, the overall impact of a corporate governance alert has a negative impact on shareholder wealth (-0.54% * from date +1 and -3.21% *** in the window (+1,+20)). Our results tend to support the notion that persistently negative effects indicate that French funds managers and institutional investors are not inclined to increase their stakes in a company suspected of being a haven for entrenched managers. Investors, whether individual or institutional are basically portfolio diversifiers, and the motive of control plays a lesser role than claimed a few years ago (e.g. Jensen 1989 and 1997). Each time they feel a company is trying to protect its managers, they vote with their feet and divest. This behaviour is in line with the conclusions of Prevost and Rao (2000) whereby the repetition or extension of the conflict related to the failure of the negotiation phase merely signals to the market the growing inability of activists to find a way to engage the dialogue and organisational changes necessary to maximise shareholder wealth. The findings concerning the influence of the number of reasons underlying the alert show, as was to be expected, a negative and significantly more unfavourable impact for these companies.

We also attempted to compare the alerts issued on corporate governance criteria linked to the possible impact of the target company's past performance. Since the mid-1990s, certain institutional activists have preferred to select their targets from among the least performing companies on account of their performance potential. Our sample indicates that: (1) the market is more measured in its negative reaction when the targets have a high return on equity, or when they have improved their return on equity between (N-1) and (N); (2) on the other hand, when performance is measured by the return on capital employed, Tobin's Q and Marris' ratio, the results of event studies are inconclusive. After twenty days of a fall in returns, the abnormal returns of the outperforming target companies revert in line with the average, which is not the case of underperforming companies whose returns continue to decline; and (3) the use of Treynor's index as the last performance measure shows a highly unfavourable reaction for companies which do not offer a positive future risk premium. The positive abnormal reaction for companies which have a future risk premium shows in the final analysis the supremacy of past performance in restoring confidence between the company and investors, even in a context of corporate governance shortcomings.

This article contributes an initial series of responses regarding the short-term results of the activism of French institutional investors, as well as the influence of the target company's past performance. On the basis of the abnormal reactions recorded in the short-term in our sample, we can question certain pension fund practices. In fact, in the USA, since the mid-1990s, the CII (17) has abandoned corporate governance criteria for pure performance criteria by targeting the worst-performing companies for their growth potential. In the light of our results, it would seem that this method of selecting targets by institutional activists is unlikely to be the most 'creative' of wealth in the short-term.

REFERENCES

Admati, A.R., P. Pfleiderer, and J. Zechner, 1994, "Large Shareholder Activism, Risk Sharing, and Financial Market Equilibrium", Journal of Political Economy 102: 1097-1130.

Admati, A.R., and P. Pfleiderer, 2005, "The Wall Street Walk as a Form of Shareholder Activism", Working paper, Stanford University.

Ali, A., C. Durtschi, B.I. Lev, and M.A. Trombley, 2000, "Informed Trading by Institutions and Quality of Accounting Information", Available at SSRN: http://ssrn.com/abstract=247506

Barber, B.M., 2006, "Monitoring the Monitor: Evaluating CalPERS' Shareholder Activism", Journal of Investing, Winter: 66-80.

Bizjak, J.M., and C.J. Marquette, 1998, "Are Shareholder Proposals Allbark and No Bite? Evidence from Shareholder Resolutions to Rescind Poison Pills", Journal of Quantitative and Financial Analysis 33: 499-521.

Brav, A., W. Jiang, F. Partnoy, and R. Thomas, 2008, "Hedge Fund Activism, Corporate Governance, and Firm Performance", The Journal of Finance 63 (4): 1729-1775.

Brown, S.J., and J.B. Warner, 1980, "Measuring Security Price Performance", Journal of Financial Economics 8: 205-258.

Brown, S.J., and J.B. Warner, 1985, "Using Daily Stock Returns, The Case of Event Studies", Journal of Financial Economics 14: 3-31.

Cai, F., and L. Xu, 2006, "Will Investors Profit By Following The Stock Recommendations Published In Barron's?", Journal of Business & Economics Research 4 (3): 69-80.

Caton, G.L., J. Donaldson, and J. Goh., 2001, "The Effectiveness of Institutional Activism", Financial Analysts Journal 57 (3): 21-26.

Chan, L.K.C., and J. Lakonishok, 1995, "The Behavior of Stock Prices Around Institutional Trades", Journal of Finance 50 (4): 1147-74.

Charreaux, G., 2003, "Le gouvernement d'entreprise," in J. Allouche, Encyclopedie des ressources humaines, Paris, Vuibert: 628-640.

DeAngelo, H., and L. DeAngelo, 1989, "Proxy Contests and the Governance of Publicly Held Corporations", Journal of Law and Economics 23: 29-59.

Del Guercio, D., and J. Hawkins, 1999, "The Motivation and Impact of Pension Fund Activism", Journal of Financial Economics 52: 293-340.

Del Guercio, D., L. Wallis, and T. Woidtke, 2006, "Do Board Members Pay Attention when Institutional Investors "Just Vote No"? CEO and Director Turnover Associated with Shareholder Activism", Working paper, University of Oregon and University of Tennessee.

Denis, D.J., and D.K. Denis, 1995, "Majority Owner-Manager and Organizational Efficiency", Journal of corporate finance 1: 91-118.

Dherment-Ferere, I., and L. Renneboog, 2000, "Share Price Reactions to Ceo Resignations and Large Shareholder Monitoring in Listed French Companies", CentER Working Paper no 2000-70, available at SSRN: http://ssrn.com/abstract=244678.

Dodd, P., and J. Warner, 1983, "On Corporate Governance: A Study of Proxy Contest", Journal of Financial Economics 11: 401-438.

Fama, E.F., M. Fisher, M. Jensen, and R. Roll, 1969, "The Adjustment of Stock Prices to new Information", International Economic Review 10: 1-21.

Firtha, M., P.M.Y. Fungb, and O. Ruic, 2007, "Ownership, Two-Tier Board Structure, and the Informativeness of Earnings--Evidence from China", Journal of Accounting and Public Policy 26 (4): 463-496.

Furtado, E.P.H., and M.S. Rozeff, 1987, "The Wealth Effects of Company Initiated Management Changes", Journal of Financial Economics 18: 147-160.

Gillan, S.L., and L.T. Starks, 2003, "Corporate Governance, Corporate Ownership, and the Role of Institutional Investors: A Global Perspective", Journal of Applied Finance 13 (2): 4-32.

Gillan, S.L., and L.T. Straks, 2007, "The Evolution of Shareholder Activism in the United States", Journal of Applied Corporate Finance 19: 55-73.

Girard, C., 2003, "La Visibility de la Cause Actionnariale", Banque et Marches 63: 22-33.

Gordon, L., and J. Pound, 1993, "Information, Ownership Structure, and Shareholder Voting: Evidence from Shareholder-Sponsored Corporate Governance Proposals", Journal ofFinance 48 (2): 697-718.

Herve, F., 2003, "L'impact de l'activisme des Fonds de Pension Americains: l'exemple du Conseil des Investisseurs Institutionnels", Banque & Marches, no 62.

Jensen, M.C., 1989, "Eclipse of the Public Corporation", Harvard Business Review, Sept-Oct., revised 1997, Available at SSRN: http://ssrn.com/abstract=146149 or DOI: 10.2139/ssrn.146149

Lambert, C., and S. Sponem, 2005, "Corporate Governance and Profit Manipulation: A French Field Study", Critical Perspectives on Accounting, 16 (6).

Karpoff, J., 2001, "The Impact of Shareholder Activism on Target Companies: A Survey of Empirical Findings", Working paper, available at SSRN: http://ssrn.com/abstract=885365.

Karpoff, J., P. Malatesta, and R. Walking, 1996, "Corporate Governance and Shareholder Initiatives: Empirical Evidence", Journal of Financial Economics 42: 365-389.

Mikkelson, W.H., and M.M. Partch, 1997, "The Decline of Takeovers and Disciplinary Managerial Turnover", Journal of Financial Economics 44 (2): 205-228.

Mulherin, J.H., and A.B. Poulsen, 1998, "Proxy Contests and Corporate Change: Implications for Shareholder Wealth", Journal of Financial Economics 47: 279-313.

Opler, T.C., and J. Sokobin, 1997, "Does Coordinated Institutional Activism Work? An Analysis of the Activities of the Council of Institutional Investors ", working paper, Ohio State University.

Pound, J., 1992, "Belong Takeovers: Politics Comes to Corporate Control", Harvard Business Review 93: 83-92.

Prevost, A.K., and R.P. Rao, 2000, "Of What Value are Shareholder Proposals Sponsored by Public Pension Funds?", Journal of Business 73 (2): 177-204.

Rogers, P., and J.R. Securato, 2007, "Comparative Study of CAPM, Fama and French Model and Reward Beta Approach in the Brazilian Market", Working paper, available at SSRN: http://ssrn.com/abstract=1027134.

Romano, R., 2001, "Less is More: Making Institutional Investor Activism a Valuable Mechanism of Corporate Governance", Yale Journal of Regulation 18: 174-251.

Smith, M.P., 1996, "Shareholder Activism by Institutional Investors: Evidence from CalPERS", Journal of Finance 51 (1): 227-252.

Solh, M., 2000, "Fonds de pension et politique d'investissement a long terme des entreprises", Doctorate Thesis in Management Science, University of Paris X Nanterre.

Song, W-L., S.H. Szewczyk, and A. Safieddinne, 2002, "Does Coordinated Institutional Investors Activism Reverse the Fortune of Underperforming Firms?", Journal of Financial and Quantitative Analysis 48: 245-268.

Strickland, D., K.W. Wiles, and M. Zenner, 1996, "A Requiem for the USA, Is Small Shareholder Monitoring Effective?", Journal of Financial Economics 40: 319-338.

Thomas, R.S., and J.F. Cotter, 2007, "Shareholder Proposals in the New Millennium: Shareholder Support, Board Responds and Market Reaction", Journal of Corporate Finance 13 (2): 368-391.

Vienot M., 1995, Le conseil d'administration des societes cotees, CNPF, AFEP.

Wahal, S., 1996, "Public Pension Fund Activism and Firm Performance", Journal of Financial and quantitative analysis 31: 1-23.

Weisbach, M.S., 1998, "Outside Directors and CEO Turnover", Journal of Financial Economics 20: 431-460.

Woidtke, T., 2002, "Agents Watching Agents? Evidence from Pension Fund Ownership and Firm Value", Journal of Financial Economics 63 (1): 99-131.

Jean-Sebastien Lantz (a), Sophie Montandrau (b) Jean-Michel Sahut (c)

(a) Associate Professor, CEROG-CERGAM, IAE Aix en Provence, France [email protected]

(b) Associate Researcher, CEROG-CERGAM, IAE Aix en Provence, France [email protected]

(c) Professor, Amiens School of Management & University of Poitiers, France [email protected]

ENDNOTES

(1.) The authors wish to thank the Association Francaise de la Gestion Financiere, AFG (French Financial Management Association) for giving us access to the data.

(2.) California Public Employees Retirement System.

(3.) Teachers Insurance Annuity Association College Retirement Equities Fund.

(4.) United Shareholder Association and the Council of Institutional Investors.

(5.) AFG which represents 1495 members: portfolio management companies, UCITS, open-end investment companies and closed-end investment companies.

(6.) When it is based on expectancies of future flows generated by the stock.

(7.) Three risk-adjusted measures of performance based on portfolio theory are generally used in literature: the indexes of Treynor, Sharpe and Jensen. Although we have used Treynor's index in our study, the correlation between these three indexes is generally very strong for portfolios.

(8.) It is therefore an a posteriori measure.

(9.) On the AFG website (www.afg.asso.fr) we can also find full details about the nine reasons for alerts, and statistics on the votes of institutional funds managers. Those statistics show a significantly growing trend in votes over the ten last years.

(10.) Average of the average abnormal returns of each company on date t.

(11.) The use of these models has three main drawbacks: The beta used in the CAPM is an expected value calculated on the basis of historical values which means the following assumption must be made: future periods are similar to past periods, which is not always true in practice (Rogers et al., 2007). The second criticism concerns the fact that the CAPM and its derivatives are "two-period" models. They therefore integrate the assumption that the betas are stable over time, namely from one period to the next (tests carried out on the basis of the market index will produce results free of beta stability problems), which is something that has not been verified in empirical studies. The last problem results from the a and p which are estimated using linear regressions, leading for example to problems concerning the autocorrelation of the remainders. Thus, for each stock i, the remainders ei, or (ARit) must have a normal distribution irrespective of t, and be independent from one session to another (no autocorrelation of remainders).

(12.) This model corresponds to the market model assuming that a is equal to zero and [beta] is equal to 1 for all companies.

(13.) To take the stability problem of the beta into account, we chose to test the robustness of our results using three different models; we observed the same tendency with all three models.

(14.) These tests were repeated on the sub-samples and conclude that the variances are homogeneous.

(15.) Value of the capital employed = value of the equity capital + value of net debt.

(16.) We performed regressions but these are not presented in the paper. The variables were: the number of bad marks, Treynor index (H4F) and the variation on the return on equity assessed by book value (h4b). Because of the correlations, we had to exclude the other variables. The t coefficients are all above excepted for the variation on the return on equity assessed by book value debt ratio. The F values are also significant but the R2 are low (between 2.5% and 14%).

(17.) Council of Institutional Investors.

Table 1
Average abnormal returns produced by the publication of an alert

Window Market Index Market Model CAPM

 ACAR t ACAR t ACAR t

-30,-10 0.63% 0.45 0.96% 0.74 0.37% 0.28
-10,0 0.01% 0.01 0.88% 0.94 0.61% 0.64
 -0.30% -0.57 0.04% 0.07 -0.07% -0.14
0 0.11% 0.36 0.20% 0.69 0.16% 0.56
+1 -0.54% -1.76 * -0.25% -0.88 -0.29% -1.00
+1,+10 -1.09% -1.13 -0.24% -0.27 -0.56% -0.63
+1,+20 -3.21% -2.35 *** -2.03% -1.60 -2.53% -1.98 **
+1,+30 -3.32% -1.99 ** -2.13% -1.36 -2.74% -1.75 *

* significant at 10%

** significant at 5%

*** significant at 2%

Table 2
Impact of an alert when there is only one reason invoked (sample 1)
and when between 2 and 6 reasons are invoked (sample 2)

ACAR Market Index Market Model

Window Sample 1 Sample 2 Sample 1 Sample 2
 (N=58) (N=49)
 Only 1 Between 2
 reason and 6

-30-10 -0.22% 1.63% -0.08% 2.19%
-10,0 0.78% -0.92% 1.24% 0.46%
-1+1 -0.33% -0.27% 2.07% 0.16%
0 0.14% 0.07% 0.20% 0.19%
+1 -0.58% * -0.49% -0.35% -0.14%
+1+10 0.63% -3.12% * 0.83% -1.50%
+1+20 -0.50% -6.43% *** -0.59% -3.74% *
+1+30 0.68% -8.06% *** 0.19% -4.87% *

ACAR CAPM

Window Sample 1 Sample 2

-30-10 0.00% 0.79%
-10,0 1.30% -0.21%
-1+1 2.24% -0.08%
0 0.20% 0.11%
+1 -0.34% -0.22%
+1+10 0.94% -2.35%
+1+20 -0.35% -5.11% **
+1+30 0.70% -6.82% ***

* significant at 10%

** significant at 5%

*** significant a 2%

Table 3
Influence of past performance

 Impact according to the return on
 equity by book value for N

ACAR Index

Window Sample 1 Sample 2
 (N=54) (N=53)

 Return on equity Return on equity
 for N < 16.6% for N > 16.6%

-30-10 0.30% 0.96%
-10,0 -0.20% 0.21%
-1+1 0.00% -0.61%
0 0.21% 0.01%
+1 -0.76% * -0.31%
+1+10 -2.64% ** 0.49%
+1+20 -5.02% *** -1.37%
+1+30 -5.84% *** -0.75%

 Impact according to the variation in
 return on equity between N-1 and N

ACAR Index

Window Sample 1 Sample 2
 (N=55) (N=52)

 Variation in Variation in
 return on equity return on Equity
 <+1.4% >+1.4%

-30-10 2.26% -1.11%
-10,0 -0.06% 0.07%
-1+1 0.19% -0.83%
0 0.32% -0.11%
+1 -0.67% -0.39%
+1+10 -2.54% * 0.45%
+1+20 -3.77% ** -2.62%
+1+30 -4.82% ** -1.74%

* significant at 10%

** significant at 5%

*** significant at 2%

Table 4
Impact according to the return on capital employed, Tobin's Q and
Marris' ratio

 Return on capital Tobin's Q
 employed

ACAR Index Index

Window Sample 1 Sample 2 Sample 1 Sample 2
 (N=54) (N=53) (N=54) (N=53)

 Return on Return on Tobin's Q Tobin's Q
 capital capital <1.2 >1.2
 employed employed
 <8.7% >8.7%

-30-10 1.02% 0.23% 0.19% 1.06%
-10,0 0.22% -0.21% 0.77% -0.77%
-1+1 0.15% -0.77% 0.10% -0.71%
0 0.31% -0.09% 0.48% -0.27%
+1 -0.47% -0.60% -0.82% ** -0.25%
+1+10 -2.03% -0.13% -1.87% -0.30%
+1+20 -3.24% * -3.19% -2.09% -4.35% **
+1+30 -5.01% ** -1.60% -3.77% * -2.87%

 Marris' ratio

ACAR Index

Window Sample 1 Sample 2
 (N=55) (N=52)

 Marris Marris
 <1.5 >1.5

-30-10 -0.15% 1.45%
-10,0 0.62% -0.64%
-1+1 0.19% -0.83%
0 0.57% -0.38%
+1 -0.58% -0.49%
+1+10 -1.36% -0.80%
+1+20 -2.24% -4.25% **
+1+30 -3.05% -3.61%

* significant at 10%

** significant at 5%

*** significant at 2%

Table 5
Impact of an alert according to Treynor's index
(sample 1<0; sample 2>0)

ACAR Market Index Market Model

Window Sample 1 Sample 2 Sample 1 Sample 2
 (N=54) (N=53)

-30-10 2.09% -0.87% 3.54% -1.67%
-10,0 0.08% -0.07% 1.74% 0.01%
-1+1 -0.49% -0.11% 0.22% -0.15%
0 -0.09% 0.32% -0.04% 0.43%
+1 -0.67% -0.40% -0.12% -0.39%
+1+10 -2.57% * 0.42% -0.50% 0.02%
+1+20 -6.71% *** 0.35% -3.07% -0.97%
+1+30 -8.58% *** 2.03% -3.29% -0.95%

ACAR CAPM

Window Sample 1 Sample 2

-30-10 1.10% -0.38%
-10,0 0.52% 0.70%
-1+1 -0.17% 0.04%
0 -0.17% 0.50% *
+1 -0.26% -0.32%
+1+10 -1.77% 0.66%
+1+20 -5.33% *** 0.32%
+1+30 -6.67% *** 1.26%

* significant at 10%

** significant at 5%

*** significant at 2%