Stock repurchases and fundamental analysis : an empirical study of the brazilian market in the period from 1994 to 2006

This study examines whether there is a relation between stock repurchase announcements and abnormal returns of companies classified as winners and losers. The classification as winners or losers followed the method suggested by Piotroski (2000), as adapted to the Brazilian market by Lopes & Galdi (2007). The data were obtained from the Economática database, referring to companies listed on the São Paulo Stock Exchange (Bovespa) in the period from 1994 to 2006. The hypotheses were tested through regression with panel data. The results show no abnormal returns for both winners and losers that announced share repurchases in the entire study period, without specifying whether this was before or after issuance in 1999 of Instruction 299 by the Brazilian Securities Commission (CVM), which enhanced protection of minority shareholders in public tender offers. After segregating the study period into two subperiods, before and after the new rules, we found a negative relation with abnormal returns for losers that announced repurchases in the earlier period (1994 to 1999), while there was no statistically significant relation for winners. For the 2000-2006 period, however, there was a positive relation with abnormal returns for both winners and losers.


INTRODUCTION
he aim of this study is to determine whether there is a relation between announcement of share repurchases by companies classified as winners or losers and abnormal returns in the year of the offer, among companies listed on the São Paulo Stock Exchange (Bovespa), based on data for the period from 1994 to 2006.The firms' classification as winners or losers was based on their results two years before the repurchase announcement.The empirical results are partly contrary to the findings of other authors in the Brazilian and international literature.
Shareholders obviously seek to invest in companies that provide abnormal positive returns, meaning returns that outperform the market as a whole.Piotroski (2000) stressed that choosing firms likely to attain positive abnormal returns in the future (winners) by means of financial indicators is a task that requires further study.
The present study is an outgrowth of the findings presented by Piotroski (2000), Gabrielli & Saito (2004) and Lopes & Galdi (2007).Lopes & Galdi (2007) adapted and tested the proposition of Piotroski (2000) in the Brazilian market and observed that it is possible to obtain abnormal positive returns by analyzing nine accounting indicators to classify firms as winners or losers, according to the expected future return.Gabrielli & Saito (2004), also studying the Brazilian market, found that companies that announced share repurchases in the period from 1994 to 1999 produced negative abnormal returns.They explained this because of the issuance of CVM Instruction 299 in 1999, which among other measures enhanced the protection of minority shareholders in public tender offers.They also found that the variation in the concentration of winners and losers could be an explanation for the variation in the sign of abnormal returns.
In light of the above, this article investigates the question: Is there a relation between share repurchase announcements and abnormal returns of firms classified as winners and losers in the Brazilian market?
Here I use panel regression with fixed effects to study the relation between abnormal returns, as the dependent variable, and the following explanatory variables: classification as winner or loser; firms that announce share repurchase; and the temporal impact found by Gabrielli & Saito (2004).I also employ some of the main control variables indicated in the literature, such as price-to-book (Fama & French, 1996), liquidity, size and indebtedness (Piotroski, 2000 andLopes &Galdi, 2007).
This study helps give a better empirical view of the relationship of share repurchase programs and fundamental analysis with abnormal returns, by providing new information about the Brazilian market.

Portfolios chosen by fundamental analysis
The relevance of accounting for the capital market has been examined by Ball & Brown (1968), Baruch & Thiagarajan (1993), Fama & French (1996), Abarbanell & Bushee (1997), Abarbanell & Bushee (1998), Ali & Hwang (2000), Bird, Gerlach & Hall (2001), Piotroski (2000;2005), Mohanram (2005), and Lopes & Galdi (2007), among other researchers who have studied the relation of accounting indicators and stock returns, to give investors a better chance to obtain good returns.Piotroski (2000) used financial information on American firms in the period from 1976 to 1996 to calculate an "F_Score" to classify firms as winners or losers and suggested that one of the strategies investors could use to find a balance between risk and return would be to observe financial indicators, distribution of dividends, accruals, stock issues and repurchase programs.Lopes & Galdi (2006) and Lopes & Galdi (2007) suggested some adaptations to the F_Score proposed by Piotroski (2000), to adapt it to the peculiarities of the Brazilian market, and also proposed the use of another score, called the R_Score, based on some quasi-continuous indicators.
Both Piotroski (2000) and Lopes & Galdi (2007) observed an opportunity to use financial information as a mechanism to reduce information asymmetry.However, Lopes & Galdi (2007) stressed that the economic reality represented by financial ratios appears to be incorporated more slowly in the price of Brazilian shares than in American ones.
Size was used as a control variable by D 'Mello & Shroff (2000), Dittmar (2000) and Ho, Liu & Ramanan (1997).In the Brazilian market, Lopes & Galdi (2007) concluded that if an investor bought shares of winners and sold them in one year or two years, he would obtain a positive abnormal return of 8.3% or 11.5%, respectively.
If this same investor invested only in winning small and medium sized firms, the abnormal return would rise to 34.5% if selling after one year and 98.2% if waiting two years.
This result is different than that of Piotroski (2000), according to which the efficiency of a portfolio of winners is independent of the firms' size.
According to Bezerra & Lopes (2004), in the Brazilian market preferred shares have the highest liquidity.On the matter of liquidity, Lopes & Galdi (2007) found evidence that the returns of winners and losers are different for firms with low liquidity in the Brazilian market.
The leverage indicator (debt/assets) was studied by Piotroski (2000), who observed that a strategy of separating winners from losers works independently of the level of firms' indebtedness.In the Brazilian market, Lopes & Galdi (2007) found evidence that a strategy of separating winners from losers is more efficient for firms with higher leverage.

Repurchase of shares and fundamental analysis
The repurchase by firms of their own shares is defined by Ross, Westerfield & Jaffe (2002), Grullon & Michaelly (2002, p. 1675), Gabrielli & Saito (2004) and Piotroski (2000) as a strategy that can have several aims: to reduce agency cost, to take the place of paying dividends, to manage capital structure and to signal to the market that a firm's shares are undervalued.
The fact that some firms signal events that generate positive abnormal returns and others signal events that generate negative abnormal returns is based on agency conflict, according to Jensen & Meckling (1976) and Akerlof (1970).With respect to signaling theory, Spence (1973) argued that by adopting certain practices, firms can send neutral, positive or negative signals to interested parties.
Searching for an explanation, by empirical testing, for the fact that firms repurchase their own shares, Dittmar (2000) observed that firms do this to distribute cash, manage their leverage and as part of employee and management stock option plans.
Bens & Wong ( 2007) corroborated the result found by Dittmar (2000), referring to management of earnings per share to increase the value of stock options.Jensen & Meckling (1976, p. 308) argued that agents with access to information about the firm can use it in their own personal interests, in detriment to those of the shareholders.
In analyzing stock repurchase based on signaling theory, Dielman, Tinothy & Wright (1980), Dann (1981), Vermaelen (1981) In contrast, the results of the study by Stephens & Wisbash (1998) indicated that share repurchase announcements are inversely related to returns.In Brazil, Moreira (2000) and Gordon (2002) concluded that repurchasing shares signals positive abnormal returns.Gabrielli & Saito (2004), on the other hand, concluded that the abnormal returns were negative in the period from 1994 to 2002, for companies that announced stock repurchases.2004) about its importance as a mechanism to protect the interests of minority shareholders.
Article 3 of CVM Instruction 345/2000 amended Art. 12, VIII, of CVM Instruction 299/1999, establishing that if holders of more than one-third of the free float shares express their intention of participating in the repurchase auction, the company must: "(i) withdraw the offer; (ii) proceed with the acquisition in proportion to the number of shares owned by shareholders accepting the offer, up to the limit of one-third of the shares in circulation; or (iii) initiate a new public tender offer procedure." Gabrielli & Saito (2004) concluded that before the application of CVM Instruction 299/1999, share repurchases led to accumulated abnormal returns of negative 10%, while afterward this figure was positive 4%.In other words, there is evidence that this instruction benefited minority shareholders (Gabrielli & Saito, 2004).
The result of the study by Gabrielli & Saito (2004) indicates that the new rules provided by CVM Instruction 299/1999 were a factor to explain the variation in abnormal returns of companies that announced share repurchases in the period analyzed.
However, besides this change in the regulatory framework, there are other factors that can explain the abnormal returns after share repurchase offers.
As explained by Fama & French (1996), Piotroski (2000) and Lopes & Galdi (2007), among these factors are price-to-book, liquidity, size, leverage and the classification as winner or loser.The strategy of building a portfolio containing the shares of companies satisfying certain financial indicators and that repurchase shares is aimed at increasing the probability of obtaining positive abnormal returns.In this respect, Ho, Liu & Ramanan (1997), besides stressing the importance of accounting numbers to investors, observed that share repurchase is positively related to earnings per share and growth of sales between the period before and after the announcement.
D 'Mello & Shroff (2000, p. 2422) separated firms by their stock price as undervalued or overvalued, by applying the model of Ohlson (1995) and controlling for size.D 'Mello & Shroff (2000) concluded that small firms that announce share repurchases attain better returns than large firms doing the same.
Lie (2005) studied companies that announced share repurchases and analyzed their earnings disclosures, concluding that the market reacts more favorably to disclosure of earnings and operational performance in the case of companies that carry out the repurchase in the same year as announced than to firms that do not do so within the same year.that by using accounting information it is possible to separate winning from losing companies according to the expectation of abnormal stock returns.In turn, Gabrielli & Saito (2004) found that Brazilian firms that repurchased shares in the period before CVM Instruction 299/1999 obtained a negative average abnormal return, while firms that did so after the change in rules attained a positive average abnormal return.

This
Based on these findings, one would expect there to be a relation between abnormal returns and the announcement of share repurchases by winners and losers.
The null hypothesis tested here is thus: H0: There is no relation between the announcement of stock repurchase by winners or losers and abnormal returns.
I tested this null hypothesis considering the periods before and after the publication of CVM Instruction 299/1999, to verify whether this variable had any influence on the abnormal returns beforehand and afterward.The method for classifying C a p i t a l S t r u c t u r e ∆LIQUID (AtivoCir .i c t2 / P.Circ .it2 ) (AtivoCir .i c t3 / P.Circ .it3 ) ∆LIQUID > 0 (1) ∆LIQUID < 0 (0) ∆LEVER ((P.Circ itt2 PEL i P t2 )/ Ativo it2 ) ((P.Circ it3  PEL i P t3 )/ Ativo it3 ) ∆LEVER < 0 (1) ∆LEVER > 0 (0) EQ_OFFER If the company issued equity in the last year before forming the portfolio, then it receives a value of zero (o), and one (1) if it did not issue equity.EQ_OFFER = 0 (1) EQ_OFFER > 0 (0) O p e r a t i o n a l E f f i c i e n c y ∆TURN< 0 (0) Source: Adapted from Piotroski (2000) and Lopes & Galdi (2007). Where: CHART 1: VARIABLES FROM Capital structure indicators -I assigned a score of one to firms with variation in liquidity greater than zero, and zero to firms with variation in liquidity less than zero.I also assigned a value of one to firms with variation in leverage less than zero, and zero to firms where this variation was greater than zero.Finally, I attributed a score of one to firms that issued shares (equity offer) in the year before formulation of the portfolio and zero to firms that did not issue shares.Operational efficiency indicators -I attributed a score of one to firms presenting a change in margin indicator greater than zero, and zero to firms with a variation in margin less than zero.Likewise, I assigned a score of zero to firms with variation in asset turnover greater than zero and zero to firms where this variation was less than zero.Based on this scheme, the nearer to nine the total score is, the more winning the firm is.
After formulating the scheme for calculating the F_Scores, I examined the Economática database for firms that announced a share repurchase in the study period.
There were 1,268 such announcements, of which 281 announcements (at least one announcement per company in each year) corresponded to companies with financial information and the other variables available.These formed the panel for testing.
The use of a portfolio chosen two years before the repurchase announcement was based on the results found by Lopes & Galdi (2007), according to which the strategy of separating firms into winners and losers with respect to abnormal returns particularly occurs two years after choosing the portfolio.From the F_Score of the companies that announced share repurchases, I classified those in the top 20% as winners and those in the lowest 20% as losers, following the classification parameters proposed by Piotroski (2000) and Lopes & Galdi (2007).Also following Piotroski (2000) and Lopes & Galdi (2007), I excluded firms with negative equity.I also CHART 3: VARIABLES OF EQUATION 2 Source: Mellagi Filho e Ishikawa (2000) Again, the highest and lowest 2.5% were excluded from the abnormal returns, as well as from the control variables.Equation 3 is the panel regression equation, without considering the period before and after issuance of CVM Instruction 299/1999, and Equation 4 contains a dummy to represent the temporal effect observed by Gabrielli & Saito (2004).The control variables used here were included to reduce the stochastic error, already foreseen in Equation 2. I adopted this procedure based on a review of the literature and chose variables already used in relation to abnormal return by Brown, Lo & Lys (1999), Fama & French (1996), Piotroski (2000) and Lopes & Galdi (2007).
(RA it )  abnormal return of asset i and t  year of the repurchase announcement  Winner it  2 = company that announced a share repurchase and that was classified two years beforehand as a winner.It is a dummy variable that assumes the value of one for winners that announced a repurchase and zero for companies that are not winners.
 Loserit2  company that announced a share repurchase and that was classified two years beforehand as a loser.It is a dummy variable that assumes the value of one for losers that announced a repurchase and zero for other companies.
Debts / assets it  (current + long-term liabilities) / total assets size it Ln of assets R F  risk-free rate of return, represented in this work by the rate paid on passbook savings accounts;  it  measure of an asset's return versus the overall market return; and E(R M )  expected market return.
Another test conducted to make the results more robust is the use of leverage as a control variable.Both Piotroski (2000) and Lopes & Galdi (2007) employed the debtto-assets ratio as the leverage indicator.
In this study I also used the debt-to-equity ratio and found no significant difference in the Brazilian market between the outcomes, as shown in Appendix C.
Therefore, I only consider debt/assets as the leverage variable in the results.Wooldridge (2006, p. 445) states that using panel regression with "fixed effects is the same as allowing a different intercept for each observation, and we can estimated these intercepts by including dummy variables...".
Based on this, I carried out a pooled regression analysis to check whether the results found by panel regression with fixed effects were consistent with those found by pooled regression (Appendix B), although the robust tool is not used in pooled regressions.
The results are constructed from panel regression analysis with fixed effects and the robustness tool, which according to Greene (1997, p. 635) can correct possible problems due to the existence of heteroskedasticity.Appendix A presents the descriptive statistics of the variables used, in general and separately for firms classified as winners and losers.

ANALYSIS OF THE RESULTS
This section presents the results and analysis of the panel regression, to answer the research question: Is there a relation between share repurchase announcements and abnormal returns by firms classified either as winners or losers?
Table 2 provides indications, by means of the Hausman test, that there is no systematic difference in the coefficients when comparing the results of the panel regression with fixed effects against that with random effects.Therefore, I assumed panel regression with fixed effects, both for Equation 3 and Equation 4.  3).This means that investors do not respond in statistically significant form to stock repurchase announcements, for either winners or losers.Although the relation of the winners and losers is not significant, the sign of the winners' coefficient (p-value = 0.13) is positive, while the sign of the losers' coefficient (p-value = 0.74) is negative.
In the panel analysis (Equation 4), I separately analyzed the winners and losers that announced share repurchases in the two sub-periods (1994-1999 and 2000-2006), by multiplying the time dummy by the dummy variables D.Winner or D.Loser.
Table 3 shows there was of a statistically significant (at 1%) inverse relationship between abnormal returns and announcements by losers of share buybacks in the earlier period (1994)(1995)(1996)(1997)(1998)(1999).In other words, investors reacted negatively, by means of the stock price, when firms classified as losers announced share repurchases in that period.
The same analysis for winners in that period did not reveal any significant relation, although the sign of the coefficient of the firms classified as winners was positive.The analysis of the relation between abnormal returns and share repurchase announcements in the later period (2000)(2001)(2002)(2003)(2004)(2005)(2006) revealed a positive and significant relation for both winners and losers (Table 3).
The signs of beta1, beta2, beta3 and beta4, (Table 3) indicate a change in investors' reaction to share repurchase announcements by winners and losers before and after 2000.This corroborates the results found by Gabrielli & Saito (2004) as well as those of Lopes & Galdi (2007), that accounting information can help investors to obtain better returns in the Brazilian stock market, Some explanations can be suggested as to why there was a positive relation between abnormal returns and share buyback announcements for both winners and losers in the period from 2000 to 2006.
 The announcement of the intention to repurchase shares is a positive sign to the market, which dominates "poor" fundamentals in the case of losers;  The period was characterized by a substantial appreciation of the market in general and this can be a "poor" adjustment to risk; and  To calculate return, it may be better to use an arbitrage pricing theory (APT) market model instead of the CAPM;

CONCLUSION
In this study I analyzed the effect on abnormal returns of share repurchase announcements by companies listed on the Bovespa in the period from 1994 to 2006.In the first analysis, without separating this interval into sub-periods before and after issuance of CVM Instruction 299/1999, the results indicate there was no relationship between these announcements and abnormal returns for either companies classified as winners or losers.
However, the analysis of these two sub-periods separately for winners and losers showed that in the period from 1994 to 1999, the market reacted negatively to stock repurchase announcements by losers, while for winners there was no significant effect on abnormal returns.
In contrast, in the period from 2000 to 2006, announcements by both winners and losers had a positive effect on abnormal returns.Therefore, the null hypothesis that "there is no relation between the announcement of stock repurchase by winners or losers and abnormal returns" was rejected.
tested some possibilities, such as cash (available and expected) and earnings management.
paper is in the positive accounting research tradition, based on analysis of empirical data about companies listed on the São Paulo Stock Exchange (Bovespa) from 1994 to 2006.It is an extension of the observations of Lopes & Galdi (2007) and Gabrielli & Saito (2004) regarding the Brazilian market.Lopes & Galdi (2007) stated The legal support for repurchase of shares in Brazil is provided byLaw 6385 of   1976 (Brasil, 1976), which covers the capital market and created the Brazilian Securities Commission (CVM).According to the basic law on corporations, Law 6404/76 (Art.30, § 2), the acquisition by a listed company of its own shares "shall obey, under penalty of nullity, the rules established by the Comissão de Valores Mobiliários, which may subject such acquisition to its previous authorization in each case." BBR, Braz.Bus.Rev (Engl.ed., Online),Vitória, v. 7, n. 1, Art. 1, p. 1-22, jan -apr.2010www.bbronline.com.brI did not find any changes that can modify the rules or interpretation of CVM Instruction 299/1999 so as to undermine the theoretical support presented by Gabrielli & Saito (

table . Table 1 : Indicators for calculating the F_Score
Piotroski (2000)007)s & Galdi (2007), who adapted and tested the method proposed byPiotroski (2000)for the Brazilian market, concluding that financial information in Brazil can help investors to value companies.The impact of fundamental analysis can be verified by the differences in abnormal returns of companies classified as winners and losers according to accounting indicators.This study is based on the fourth-quarter financial disclosures of firms listed on the Bovespa, obtained from the Economática database.I used these figures to calculate financial indicators for profitability, capital structure and operational efficiency (Table1), and then employed these to calculate the F_Score, to classify firms as winners or losers according to the BBR, Braz.Bus.Rev (Engl.ed., Online), Vitória, v. 7, n. 1, Art. 1, p. 1-22, jan -apr.2010 www.bbronline.com.brwinners and losers Lopes & Galdi (2007)inancial indicators, I followed the proposal ofLopes & Galdi (2007), principally in relation to cash and cash equivalent.Thus, ROA, ∆ROA, CTA (cash and cash equivalents to total assets) and ACCRUAL indicate the firms' Lopes & Galdi (2007)ted fromPiotroski (2000)andLopes & Galdi (2007).NIit = net income of company i in period 1; Stock repurchases and fundamental analysis BBR, Braz.Bus.Rev (Engl.ed., Online),Vitória, v. 7, n. 1, Art. 1, p. 1-22, jan -apr.2010  www.bbronline.com.br  expected rate of return of asset i and t  the year of the repurchase announcement;