O trabalho utiliza um estudo de evento para examinar os retornos de ações relacionados a emissões públicas por empresas brasileiras listadas na BOVESPA, realizadas entre 1992 e 2002, buscando determinar como o mercado reagiu antes, durante e depois da data do anúncio da emissão. Após utilizar a metodologia convencional de mensuração de retornos anormais por OLS, foram utilizados modelos ARCH e GARCH, que levam em consideração a heteroscedasticidade condicional da volatilidade dos retornos anormais, em mais de 70% da amostra, após a constatação da presença desses processos nos resíduos originais. Os resultados mostram que (1) há evidências de insider information antes da data do anúncio, (2) que ocorrem retornos anormais negativos na data do anúncio e (3) que, no período de um ano após as emissões, as ações das empresas que captaram recursos via underwriting tiveram retornos negativos após ajuste ao risco e ao mercado.

The paper uses an event study to examine stock returns associated to public equity issues carried out by Brazilian firms listed on BOVESPA, from 1992 to 2002, with the purpose of determining how the market reacted before, during, and after the issue announcement dates. After utilizing the conventional method for measuring abnormal returns by OLS, we used ARCH and GARCH models, which take into account the conditional heteroskedastic volatility of the abnormal returns in more than 70% of the sample, since the presence of these processes was found in the original residuals. The results show that (1) there is evidence of insider information prior to the announcement date, (2) abnormal returns occurred on the announcement date, and (3) within one year after the issues, the shares of issuing firms had negative returns after adjusting for risk and market effects.

We have carried out an event study to investigate stock returns associated with the announcement of equity issues by Brazilian firms between 1992 and 2003 in order to determine market reaction before, during, and after the issue announcement. After measuring abnormal returns by OLS, we used ARCH and GARCH models over 70% of the sample. Our results are remarkably consistent with most of the international empirical literature. Some previous empirical findings have turned up abnormal returns before the announcement date, interpreted as signs of insider information. This evidence also appears in our study as we found an average cumulative abnormal return of -0.01 three weeks before the announcement. With respect to the announcement date, the evidence reported in the literature is virtually unanimous in showing negative abnormal returns, meaning that stock issues convey pessimistic information to the market. Our study confirms these findings with an average -0.03 cumulative abnormal return on the first three days following the announcement. Finally, the empirical literature has also collected evidence of long-term negative abnormal returns after the issues, which we also confirm, with an abnormal return of -0.28 after one year following the announcement. The results also show that ARCH/GARCH estimation of abnormal returns is superior to OLS estimation.

O trabalho relata um estudo de evento para investigar a reação do mercado antes, durante e após os anúncios de emissões de empresas brasileiras, entre 1992 e 2003. Após mensurar os retornos anormais por mínimos quadrados, utilizou-se modelos ARCH e GARCH para 70% da amostra. Os resultados são consistentes com a literatura internacional. Alguns resultados empíricos antecedentes evidenciaram retornos anormais anteriores ao anúncio, interpretados como indícios de informação privilegiada. Tal evidência também aparece no presente estudo, onde se encontrou um retorno anormal cumulativo médio de -0,01 a três dias antes do anúncio. Com relação à data do anúncio, a literatura existente é virtualmente unânime em relatar retornos anormais negativos, significando que emissões de ações transmitem informações pessimistas ao mercado. O presente estudo confirma tais achados com um retorno anormal cumulativo médio de-0.03 nos três primeiros dias após os anúncios. A literatura empírica também acumulou evidências de retornos anormais negativos no longo prazo após as emissões de ações, o que também foi confirmado na presente pesquisa, com um retorno anormal cumulativo médio de -0,28, um ano após os anúncios. Os resultados também mostram que a estimação dos retornos anormais por ARCH e GARCH é superior àquela realizada por mínimos quadrados.

We have carried out an event study to investigate stock returns associated with the announcement of equity issues by Brazilian firms between 1992 and 2003 in order to determine market reaction before, during, and after the issue announcement. After measuring abnormal returns by OLS, we used ARCH and GARCH models over 70% of the sample. Our results are remarkably consistent with most of the international empirical literature. Some previous empirical findings have turned up abnormal returns before the announcement date, interpreted as signs of insider information. This evidence also appears in our study as we found an average cumulative abnormal return of -0.01 three weeks before the announcement. With respect to the announcement date, the evidence reported in the literature is virtually unanimous in showing negative abnormal returns, meaning that stock issues convey pessimistic information to the market. Our study confirms these findings with an average -0.03 cumulative abnormal return on the first three days following the announcement. Finally, the empirical literature has also collected evidence of long-term negative abnormal returns after the issues, which we also confirm, with an abnormal return of -0.28 after one year following the announcement. The results also show that ARCH/GARCH estimation of abnormal returns is superior to OLS estimation.