The intra-industry effects of proxy contests

In this study, we examine the intra-industry effect of proxy contests. Proxy contests convey the information of common industrial risks or expected competitive relationship change. We find significant negative abnormal returns in the group of competitors of target firms with negative abnormal returns, and such negative abnormal returns become larger for similar-size competitors. In contrast, there are no significant abnormal returns for competitors of target firms with positive abnormal returns. These findings are consistent with the information-based theory but not the competitive theory.

Over the years, researchers have studied the accounting and market performance of target firms in proxy contests and conclude that proxy contests improve the target firms' competitive performance (Cohn et al. 2016;DeAngelo and DeAngelo 1989;Dodd and Warner 1983;Fos 2016;Laudano 2004;Mulherin and Poulson 1998). An important implication is that proxy contests can potentially change the competitive landscape of industries. However, it remains unclear how investors interpret the information content of proxy contests and react to the industry's competition change.
To the best of our knowledge, this is the first study to investigate the intra-industry effect of proxy contests which hasn't been extensively explained in the existing literature.
Proxy contests potentially have two effects on target firms' competitors. On the one hand, the operational inefficiency and poor performances normally are the triggers of proxy contests (Austin 1965;Cohn et al. 2016;Mukherjee and Varela 1993;Sridharan and Reinganum 1995). The information-based theory posits that proxy contests often reveal common industrial risks. Investors interpret such risks as pertaining to peer competitors, leading to the industry contagion effect. On the other hand, the competitive theory suggests that the stock market reaction is affected by the potential change of the industrial competitive landscape. A large portion of the proxy contests are the attempts to mitigate the agency problem stemming from inefficient management (Borstadt and Zwirlein 1992), potentially to improve the target firms' performances. According to the competitive theory, the positive stock market reaction to target firms is interpreted as the market's anticipation of the target firms potentially improving their profitability and seizing more market share from their competitors (Borstadt et al. 1992;Fos 2016). In contrast, when investors expect that the proxy contest firms may underperform their competitors in the post-contest periods, the stock market has no reaction or even a negative reaction to the proxy contest announcements. For example, Mukherjee and Varela (1993) find that the lower post-contests profit margins of some targeted firms lead to their underperformance.
The stock market reaction to proxy contest firms has an impact on the stock market reaction to their competitors, which has different predictions under the informationbased theory versus the competitive theory. When the cumulative abnormal returns (CARs) of targeted firms are negative, in line with the information-based theory, a negative stock return of the target firm is a signal that common industrial risks also affect the competitors, inducing a negative stock market reaction to the competitors. In contrast, under the competitive effect theory, negative stock returns of target firms are interpreted as an expectation of their weakened competitive ability and underperformance in the future, leading to a positive stock market reaction to their competitors.
On the other hand, when the CARs of target firms are positive, according to the information-based theory, the market reaction to the competitors is trivial since no new common risk is revealed. In contrast, in line with the competitive effect theory, positive CARs of the target firms are mainly the consequence of an expected strengthened competitive advantage of target firms over their competitors, inducing negative stock returns to the competitors.
Using a sample of proxy contests between 1998 and 2008, we show that target firms, on average, experience positive risk-adjusted abnormal returns upon the announcements of proxy contests, consistent with prior studies. However, we find a striking cross-sectional difference in the market reactions to the target firms. 56% of the target firms have significantly positive CARs over the (−10, +10) period, while the remaining 44% of the target firms have negative CARs in the same event window. The interesting question is determining which effect (the information-based intra-industry effect or the competitive intra-industry effect) influences the market reaction to the competitors of target firms.
We first examine the stock returns to all the competitors and find significantly negative CARs of the competitors over the (−10, +10) period. Then we split the competitors of the target firms into two groups: one with negative CARs and another one with positive CARs. In the negative-CARs target firm group, the CARs of the competitors is significantly negative. While in the positive-CARs target firm group, the CARs of the competitors is not significant. These results are consistent with the information-based theory.
To further disentangle these two effects, we add firm size to the analysis as a determinant of the intra-industry effect. Given the fact that similar-size competitors have more features in common and more direct competition with the target firms than other competitors, the information-based effect and the competitive effect are expected to be stronger in similar-size competitors. Overall, the average CAR of similar-size competitors over the (−10, +10) period is significant and more negative than that of all competitors. When we split the similar-size competitors into the positive-and negative-CARs target firm groups, we find a difference in the competitors' average CAR. The average CAR is −2.38% (−4.18%) and significant for the competitors whose sizes are within ± ten (five) percent of the negative-CARs target firms, more negative than the average CAR of all competitors of the same target firms. In contrast, there is no significant average CAR for the competitors whose sizes are within ± ten (five) percent of the positive-CARs target firms. Both CARs of the similar-size competitors are consistent with the prediction of information-based theory, not competitive theory. Therefore, these results support a stronger information-based intra-industry effect to the similar-size competitors.
In line with the information-based theory, the more information asymmetry there is, the more new information should be conveyed from the proxy contests and the stronger the intra-industry effect. To test this inference, we split the target firms into quintiles based on size, considering that the smaller-size firms have more information asymmetry than large-size firms. Overall, we find that the competitors' CARs are significantly negative in the two smallest target firm groups and insignificant in other groups. We then split the target firms into the negative-and positive-CARs groups and find similar patterns. The findings also support the information-based theory.
Lastly, we conduct a regression of the competitors' abnormal returns on the targetrelated factors, competitor-related factors, and industry factors. The regression results reveal that when the target firms have positive abnormal returns, the stock market reaction to the competitors is primarily driven by the target-related factors. In contrast, when the target firms have negative abnormal returns, the stock market reaction to the competitors is mainly affected by the competitor-related factors and industry factors.
Our findings contribute to the large body of literature on proxy contests. While extant research finds that the market reacts positively to the proxy contests (Borstadt and Zwirlein 1992;Cohn et al. 2016;Faleye 2004), we find a negative market reaction to the targeted firms' competitors. Moreover, we disentangle the information-based intra-industry effect and competitive intra-industry effect. In a broader view, our findings enrich the corporate governance research by showing the impact of target firms' expected corporate governance change on the market reaction to their competitors.
The remainder of the paper is structured as follows. Section 2 provides a literature review and hypotheses development. Section 3 describes the data and variables. Section 4 provides empirical results. And section 5 concludes.

Literature review and hypotheses development
The proxy contest research is not new; however, the intra-industry effects of proxy contests have not been examined in prior studies. Consequently, we review the major papers relevant to the development of this study under two headings: studies related to (1) stock market reaction and performance outcome of proxy contests, and (2) the intraindustry effects.

Stock market reaction and performance outcome of proxy contests
Prior studies show that the proxy contest is often an attempt to discipline inefficient management. However, the stock market reaction and performance outcome of proxy contests is mixed. On the one hand, there is evidence that proxy contest firms receive positive abnormal returns, and the post-contest performances are improved as expected. Dodd and Warner (1983) examine proxy contests for board seats between 1962 and 1978. They find significantly positive abnormal returns of the target firms upon the announcement of these contests, although most of the time dissidents fail to gain control of the board. Dodd and Warner attribute the positive market reaction to the improved corporate performance brought forth (perhaps prompted) by the proxy contests. Studying proxy contests during 1978-1985, DeAngelo and DeAngelo (1989 obtain similar results with abnormal returns in a range from 2.94% to 3.84% around the proxy contest announcement day. They show that proxy contests often result in the management change even when the initial contest has failed. They find that stockholder wealth gains are most noticeable when dissidents can force the sale or liquidation of the firm. Consequently, target firms tend to outperform the market upon the announcements of proxy fights even though their accounting performance tends to be poorer than average before the proxy contests (Borstadt and Zwirlein 1992;Fos 2016;Mukherjee and Varela 1993). Mulherin and Poulson (1998) look at a longer period from 1979 to 1994 and find that proxy contests create value, especially for firms that are acquired. They also find that even for firms that are not acquired, management turnover has a positive effect on shareholder wealth. Laudano (2004) surveys the research on proxy contests and concludes that "the cumulative research on proxy contests supports the contention that such contests are an effective tool for disciplining inefficient managers and implementing corporate changes." Laudano argues that proxy contests increase shareholder wealth regardless of their outcomes.
On the other hand, there is also evidence that proxy contest firms have inferior performance after the contests. Ikenberry and Lakonishok (1993) identify negative post-event abnormal returns for proxy contest firms. Mukherjee and Varela (1992) find that unsuccessful contest firms often suffer losses and successful contest firms show a higher rate of bankruptcy in the long term than other firms. Mukherjee and Varela (1993) further point out that the underperformance of contest firms is due to their lower profit margin than matching firms after the contests. Fos (2016) finds that proxy contests that aim at changing capital structure and governance do not lead to higher firm values.

Intra-industry effects
Intra-industry effects have been documented in many major corporate events studies, such as bankruptcy (Jorion and Zhang 2007;Lang and Stulz 1992), dividends reduction or omission (Impson 2005), stock split announcements (Tawatnuntachai and D'Mello 2002), stock repurchases (Otchere and Ross 2002), and accounting restatements (Gleason et al. 2008). Lang and Stulz (1992) examined the intra-industry effects of bankruptcy announcements. They report that there is an intra-industry contagion effect when the bankruptcy conveys negative prospects of factors common to the industry; however, there also exists a competitive effect of bankruptcy when competing firms snatch a market share from distressed firms. Both effects have been observed in other studies. Aharony and Swary (1996) provide evidence of the information-based contagion effects of bank failures. Jorion and Zhang (2010) find that bond rating downgrades have two opposing effects, the contagion effect, and the competition effect, on the industry rivals.
Looking at the dividend reductions and omissions announcements of ten utility companies, Impson (2005) finds an intra-industry systemic risk effect in the electric utility industry in response to dividend omissions and decreases. Gleason et al. (2008) find that some accounting restatements cause investors to reassess the financial statement information previously released by non-restating firms. Sometimes, the intraindustry effect can also be positive. Tawatnuntachai and D'Mello (2002) show that favorable information conveyed by stock split announcements transfers to non-splitting firms within the same industry. Otchere and Ross (2002) investigate stock repurchases and find that share buyback announcements signal positive information about the values of both announcers and rivals.
Intra-industry studies have also found that firm size is a key determinant of the market reaction. For example, Collins et al. (1987) use firm size to proxy the amount of information and the number of informed traders when investigating the information content of prices concerning earnings. They find that the size of the announcing-firm within the industry has a direct impact on the magnitude of the intra-industry effect. Gonen (2003) finds a positive relationship between the intra-industry effect of a corrective disclosure and the industry position of the firm measured by its relative size within the industry. Tawatnuntachai and D'Mello (2002) find that the interaction of the CARs of stock-splitting firms and their relative size position in the industry has a significantly positive effect on the CARs of non-splitting firms. Gleason et al. (2008) find that larger firms in an industry have more pronounced intra-industry effects due to revenue reinstatements. However, as the firm size increases, they have more resources to reposition themselves and thus convey less industry information. In extending this line of research, we examine how target firm size and competitor firm size affect the outcome of intra-industry analyses.

Hypotheses development
The possible presence of the intra-industry effect of proxy contests can be explained by the information-based theory and/or the competitive theory. According to the information-based theory, the proxy contest is a signal of a common risk shared by the targets and their competitors. It predicts a negative stock market reaction to the competitors of proxy contest targets. From the perspective of competitive theory, the stock market reaction to the competitors is contingent on the expectation of the changing competitive relationship between the target firm and the competitors. Given the fact that the abnormal return to the target firms is positive overall, the market expectation is that in general proxy contests improve the target firms' competitive advantage over their competitors. Consequently, in line with both theories, we predict a negative market reaction to the competitors overall. This gives rise to the first hypothesis.
H1. The stock market reaction to the competitors of the target firms of proxy contests is negative.
To disentangle the information-based theory and the competitive theory, we further split the competitors into two groups: the competitors to the target firms with negative CARs and the competitors to the target firms with positive CARs. When the CARs for the target firms are negative, there are also negative CARs for the competitors due to the revealed common risk by the proxy contests. Therefore, we propose the second hypothesis in line with the information based theory.
H2a. The stock market reaction to the competitors of the target firms with negative CARs is negative.
In contrast, based on the competitive theory, the positive CARs for the competitors are due to the expectation that the target firms will lose their competitive advantage to the competitors. Thus, we propose the alternative second hypothesis in line with the competitive theory.
H2b. The stock market reaction to the competitors of the target firms with negative CARs is positive.
According to the information-based theory, when the CARs of the target firms is positive, consistent with the expectation on the discipline role of proxy contests on the target firms, no new information of industry risk is revealed. As a result, there is no significant stock market reaction to the competitors. Therefore, we propose the third hypothesis.
H3a. The stock market reaction to the competitors of the target firms with positive CARs is not significantly different than zero.
Alternatively, according to the competitive theory, positive CARs of the target firms can be interpreted as the market expectation of the competitive advantage of target firms over their competitors. Therefore, the market reaction to the competitors should be the oppositenegative. This gives rise to the alternative third hypothesis.
H3b. The stock market reaction to the competitors of the target firms with positive CARs is negative.
As previously mentioned that firm size is a key determinant of the market reaction of intra-industry effect, we propose that when the competitors are of a similar size as the target firms, the effect of common risk is expected to be stronger than that in other groups, as the similar-size competitors and the target firms have more in common. Therefore, when the CARs of the target firms are negative, we expect that the negative stock market reaction to the similar-size competitors is larger than that of all competitors. This is consistent with the information-based theory which gives rise to the fourth hypothesis.
H4. The stock market reaction to the competitors having a similar size as the target firms with negative CARs is more negative than that to other competitors.
On the other hand, the expected effect on the targets firms with positive CARs is more in line with the competing theory. Specifically, when the competitors are of similar sizes as the target firms, the effect of the changing competitive landscape is also expected to be stronger than that in other groups, as the similarsize competitors and the target firms are directly competing. Therefore, when the CARs of the target firms are positive, we expect that the negative stock market reaction to the similar-size competitors is larger than that of all competitors. Consistent with the competitive theory, this gives rise to the fifth hypothesis.
H5. The stock market reaction to the competitors having a similar size as the target firms with positive CARs is more negative than that to other competitors.
In view of the information-based theory, the intra-industry effect depends on the extent of information asymmetry. When there is little information asymmetry, not much new information is revealed by the proxy contest. In contrast, when there is large information asymmetry, more new information is conveyed. We use firm size to proxy for the degree of information asymmetry. The small size target firms normally have more information asymmetry than large size target firms. This reasoning links the size of target firms to the intra-industry effect of proxy contests, leading to our sixth hypothesis.
H6. The intra-industry effect of the smaller size target firms is stronger than that of the larger size target firms due to greater information asymmetry.
According to the information-based theory, the intra-industry effect exists because of new information from the target firms, competitors and the industry are revealed, implying that the target-related factors, competitors-related factors, and industry factors are determinants of the competitors' CARs. When CARs for the target firms are negative, the information is more related to the common risk to the competitors and the industry, competitors-related factors and industry factors should have more weight than target-related factors in affecting the competitors' CARs. Therefore, we propose the seventh hypothesis.
H7. The intra-industry effect is mainly affected by the competitors-related factors and industry factors when the CARs of the target firms are negative.
On the other hand, when CARs for the target firms are positive, there is not much new information related to the industry risk revealed, instead, the target-related factors are relatively more important in affecting the competitors' CARs. This gives rise to the eighth hypothesis.
H8. The intra-industry effect is mainly affected by the target-related factors and competitors-related factors when the CARs of the target firms are positive.
The next section describes the data and methodology used in our empirical tests of these hypotheses.

Data and methodology
We obtain a list of proxy contests from the Security Data Corporation's (SDC) database. The SDC database contains 737 domestic proxy contest initiations from January 1988 through December 2008. We then match the sample with the Center for Research in Security Prices (CRSP) database by CUSIP and Ticker Symbol and get 647 matches (470 by CUSIP and 177 by Ticker Symbol). For each firm, we use the number of shares outstanding and the closing price at the end of the year before the proxy contest initiation to calculate the market capitalization. Table 1 reports descriptive statistics for the pooled sample of 647 proxy contest initiations between 1988 and 2008. There is a trough in proxy contest initiations between 2000 and 2004, and a peak toward the end of the period, with over fifty initiations per year during 2006-2008. While we are hesitant to draw any conclusions about the "typical" size of target firms, we note that targets are much larger during the 2005-2008 period than any other period, especially compared to the late 1990s (a period of great market return).
We follow Lang and Stulz's (1992) method and use the primary four-digit SIC code in CRSP to identify the industry competitors of our target sample. We then obtain company financial information from COMPUSTAT. To study the effect of a proxy contest announcement on its industry competitors, we form an equal-weighted portfolio of all firms in the same industry.
The main testing variable in our study is the CARs around the proxy contest announcements. Abnormal returns are computed using a standard event-study Market capitalization = shares outstanding * closing price. Market capitalization is calculated on the last trading day of the year before the proxy contest and is in millions. The industry competitors are defined as the firms with the same 4-digit SIC code. The sample is from 1988 to 2008 methodology following Brown and Warner (1985). Any non-trading event day has been converted to the next trading day. For each security j, the market model is used to calculate the abnormal return (AR) for event day t as follows: where R jt is the rate of return on security j for event day t, and R mt is the rate of return on the CRSP equally-weighted market portfolio on event day t. 1 Market model parameters are estimated using the window of [−301, −46] relative to the proxy contest announcements. The coefficients α j and β j are the ordinary least squares estimates of the intercept and slope, respectively, of the market model regression.
The daily abnormal returns are summed to get the cumulative abnormal returns (CARs) from day T 1 before the proxy contest announcement to day T 2 after the announcement date. The cumulative abnormal returns for firm j (CAR j ) from day T 1j to day T 2j is defined as: We accumulate abnormal returns at various intervals around the announcement date. For a sample of N securities, the mean CARs is defined as: The test statistic described by Brown and Warner (1985) is the mean standardized cumulative abnormal return (SAR jt ). To compute this statistic, we obtain an estimate of standard deviation, S jt. The value of S jt is: where S j 2 residual variance for security j from the market model regression D j number of observations during the estimation period R mt rate of return on the market index at date t of the event period R m mean rate of return on the market index during the estimation period The SARs are accumulated, and t-test statistics for the sample of N securities are employed.

Market reaction to the target firms
In this section, we first present the market reaction to the proxy context target firms to establish how in general investors interpret the initiations. In Table 2, we report target firms' abnormal returns in different window periods relative to the proxy-initiation announcement day. With the equal-weighted CRSP index benchmark, the cumulative abnormal return (CAR) for the period of (−10, +10) is 3.47%, significant at the 1% level. The average abnormal returns on the announcement day (t = 0) is 1.15% and on day t + 1 is 0.55%. Similarly, with the value-weighted CRSP index benchmark, the cumulative abnormal return (CAR) for the period of (−10, +10) is 4.28%, significant at the 1% level. The average abnormal returns on the announcement day (t = 0) is 1.20% and on day t + 1 is 0.59%. The daily abnormal returns during the period of (−4, −1) are all significantly positive and the CARs of (−5, −1) are 1.5% (equal-weighted benchmark) and 1.7% (value-weighted benchmark). The results are consistent with the possibility of information leakage of a forthcoming proxy contest. The post-contest CARs of the target firms in Table 2 are also consistent with the previous research, e.g., abnormal return of 4.27% in the month of the proxy contest reported for a smaller sample from 1968 to 1987 by Ikenberry and Lakonishok (1993). In general, these results support the notion of the discipline role of proxy contests. The left half of Table 2 also reveals that 361 (56%) target firms have positive CARs in the window of (−10, 10) around the announcement day, while 275 (44%) target firms have negative CARs during this period with the equal-weighted method. The results suggest that such CAR asymmetry may come from the different expectation of the impact of proxy contests on the competitive position of the target firms, as well as the interpretation of the information conveyed by the proxy contests. The value-weighted results are in the right half of Table 2 and show similar results to the equal-weighted results.

All industry competitors
Next, we examine the CARs of the competitors in the same industry when proxy contests are announced. The results are presented in Table 3.
In Panel A of Table 3, we report the CARs of all competitors with equal-weighted and value-weighted methods, respectively. The first three columns show the CARs with the equal-weighted competitors' portfolios and the next three columns show the CARs with the equal-weighted CRSP index benchmark. The cumulative abnormal returns (CAR) with equal-weighted benchmark are −0.03% for the period of (−10, +10) and − 0.17 percentage for the period of (−5, 0); both are statistically significant. For the CARs with a value-weighted method in the next three columns, all the CARs are not statistically significant. These results are consistent with our first hypothesis that the stock market reaction to the competitors of the target firms of proxy contests is negative overall. In addition, the difference of CARs between the equal-weighted method and The table includes all target firms of proxy contests in our sample (out of the 647 proxy contest samples, some proxy fights have been dropped because they do not meet the requirement of a minimum of 120 security returns in the estimation period.) The abnormal return (AR) is the market adjusted return in percentage. Event day is the proxy contest announcement day. The market index benchmark is CRSP equal-weighted index and value-weighted index respectively. Estimation period ends 46 trading days before the event date. Minimum estimation length is 120 trading days. Maximum estimation length is 255 trading days. Estimate method is OLS. Number denotes the number of abnormal returns available to compute the average abnormal return. The symbols *, **, and *** denote statistical significance at the 0.10, 0.05 and 0.01 levels, respectively, using a generic two-tail test.  The table includes all competitors of target firms of proxy contests in our sample (out of the 647 proxy fight samples, some proxy fights have been dropped because they do not meet the requirement of a minimum of 120 security returns in the estimation period.) The abnormal return (AR) is the market adjusted return in percentage. Event day is the proxy contest announcement day. The market index benchmark is CRSP equal-weighted index and value-weighted index respectively. Estimation period ends 46 trading days before the event date.
Minimum estimation length is 120 trading days. Maximum estimation length is 255 trading days. Estimate method is OLS. Number denotes the number of abnormal returns available to compute the average abnormal return. The symbols *, **, and *** denote statistical significance at the 0.10, 0.05 and 0.01 levels, respectively, using a generic two-tail test.
the value-weighted method implies that the size of the firms has an impact on the magnitude of CARs.
To disentangle the information-based theory and competitive theory, we divide the proxy target firms into two subsamples, targets with positive CAR group and targets with negative CAR group, and compute abnormal returns of competitors with similar size in each group respectively. The results are reported in Panel B and Panel C of Table 3. Panel B shows the CARs of the competitors of target firms with negative CARs. With the equal-weighted method, the CARs in the window periods (−5, 1), (−5, 0), (−5, 5), (−10,-1) and (−10, +10) are all negative and significant, with the largest CAR of −0.34% in the (−10, +10) period. With the value-weighted method, the CARs are of the same sign and show a significant and larger magnitude in the same window periods. The results support the prediction by the information-based theory that the stock market reaction to the competitors is negative when the CARs of the target firms are negative. The results do not support the competitive theory which predicts the stock market reaction to the competitors is positive given that the CARs of the target firms are negative. These results are consistent with our second hypothesis, H2a, that the stock market reaction to the competitors of the target firms with negative CARs is negative and is not consistent with the alternative hypothesis H2b.
Panel C shows the CARs for the competitors of target firms with positive CARs. The CARs in the window periods (−5, 1), (−5, 0), (−5, 5), (−10,-1) and (−10, +10) are insignificant, with either the equal-weighted or value-weighted method. The results again support the prediction of negative stock market reaction to the competitors by the information-based theory, while conflicts with the competitive theory prediction of positive stock market reaction to the competitors, under the condition that the CARs of the target firms are positive. These results are consistent with our third hypothesis, H3a, that the stock market reaction to the competitors of the target firms with positive CARs is not significantly different from zero and is not consistent with the alternative hypothesis H3b.
In sum, we find empirical evidence for the information-based theory but no competitive theory in the market reaction to the competitors based on two target CARs groups.

Similar-size industry competitors
In this section, we examine whether the intra-industry effect of proxy contests on similar-size competitors is different than that of all competitors. Similar-size competitors have a lot more common features and competing relationships with the target firms than other competitors. We define similar-size industry competitors as firms with market capitalizations that are within ±10% of that of the target firms in the same industry. We then form an equally-weighted portfolio of the industry competitors for each proxy contest firm. The results are reported in Table 4. With a ±10% threshold, we list 767 competitors for 129 target firms with negative CARs in the left half of Panel B. For the target firms that have positive CARs, we list 995 competitors for 186 target firms in the left half of Panel C. Alternatively, we define similar-size competitors as firms with a market capitalization within ±5% of that of the target firms in the same industry; the number of competitors that match the target drops slightly. With ±5% threshold, we list 395 competitors for 95 target firms with negative CARs in the right half of Panel B, and 522 competitors for 146 target firms with positive CARs in the right half of Panel C.
The intra-industry effect on all similar-size competitors is reported in Table 4 Panel A. In the period of (−10, +10), the similar-size industry competitors experience significant CARs of −1.38% when the competitors are firms within ±10% of that of the target firms, and CARs of −2.12% when the competitors are firms with market capitalizations within ±5% of that of the target firms.
In Panel B of Table 4, we examine the intra-industry effect for similar-size competitors of the target firms with negative CARs. In the period of (−10, +10), the similarsize industry competitors experience significant CARs of −2.38% when the competitors are firms within ±10% of that of the target firms, and CARs of −4.13% when the competitors are firms with market capitalizations within ±5% of that of the target firms. These results indicate a significant intra-industry effect of proxy contests when target firms have negative CARs.
In contrast to the results in Panel B, we see no evidence of an intra-industry effect to the similar-size competitors of the target firms with positive CARs in Panel C. The daily abnormal returns for competitors in the same industry cluster around zero with no statistical significance during the event windows. The positive abnormal returns earned by the target seems to be limited to target firms only. Therefore, there is no intraindustry effect found on the similar-size competitors when the market reaction to the target firms is positive.
In sum, the CARs is the largest in the similar-size competitors of the target firms with negative CARs and insignificant in the similar-size competitors of target firms with positive CARs. These results are consistent with the hypothesis H4 that the stock market reaction to the competitors having a similar size as the target firms with negative CARs is more negative than that to other competitors, while the results are not consistent with hypothesis H5. This finding supports the information-based theory while conflicts with the competitive theory.

Impact of the target firm's industry position on the intra-industry effect
Prior studies suggest the importance of firm size on intra-industry effects (Gleason et al. 2008;Gonen 2003;Tawatnuntachai and D'Mello 2002). We next explore how the intra-industry systemic risk effect might vary for different target firm sizes.
We compute each target firm's relative size within its industry by dividing the target firm's market capitalization by the median industry market capitalization. Then we rank each target firm by placing the ratios into five quintiles (1 represents target firms with the lowest relative market size, and 5 represents target firms with the highest relative market size). The results are reported in Table 5.
In Panel A of Table 5, we provide the CARs of the competitors of all target firms during the (−10, +10) period around proxy contest announcements. The CARs for the competitors is −1.00% in the smallest target firms' relative size quintile and − 0.41% in the second smallest target firms' quintile; both are significant at the 1% level. The CARs are not significant in the larger target firms' relative size quintiles.
In Panel B of Table 5, we provide the CARs of the competitors of target firms with negative CARs during the (−10, +10) period around proxy contest announcements. The CARs for the competitors is −1.38% in the smallest target firms' relative size quintile and − 1.83% in the second smallest target firms' relative size quintile, both are larger than that for all competitors. The CARs in the third smallest quintile is −1.23% and significant but the fourth and fifth quintile results are insignificant.
In Panel C of Table 5, we find that the CARs during the (−10, +10) period around proxy contest announcements is −1.13% and significant at the 5% level in the smallest target firms' relative size quintile, −0.21% and significant at the 10% level in the second smallest target firms' relative size quintile. The CARs are insignificant in other quintiles.
In sum, smaller target firms appear to be more likely to have a negative impact on the stock market's reaction to their competitors, while the largest targets tend not to influence such a reaction. When the target firms have negative CARs, the market reaction to the competitors is the most negative. The results are consistent with the  For each proxy contest firm, we define its competitors as the firms with the same 4-digit SIC code and market capitalization within ± 10 % (the left of the panels) or ± 5 % (the right of the panels) of that of the proxy contest target firm. Then we form an equal-weighted competitor portfolio for each proxy contest firm. The competitors are divided into two groups. In Panel A, the firms are the competitors of the targets with a negative cumulative abnormal return in the (−10, −1) and (−10, +10) day period around the proxy contest announcement date. In Panel B, the firms are the competitors of the targets with a positive cumulative abnormal return in the (−10, −1) and (−10, +10) day period around the proxy contest announcement date. The abnormal return (AR) is the market model residual in percentage. Event day is the proxy contest announcement day. Event study uses CRSP daily data. Market index is CRSP equal-weighted index. Estimation period ends 46 trading days before the event date. Minimum estimation length is 120 trading days. Maximum estimation length is 255 trading days. Estimate method is OLS. The numbers in pair denote the portfolios' number/the competitors' number available to compute the average abnormal return respectively. The symbols $,*, **, and *** denote statistical significance at the 0.10, 0.05, 0.01 and 0.001 levels, respectively, for differences in two panels.
hypothesis H6 that the intra-industry effect of the smaller size target firms is stronger than that of the larger size target firms. This finding supports the information-based theory but not the competitive theory.

Regression of the competitors' CARs on the target-related, competitor-related and industry factors
To examine the factors that may drive the market's response to the competitors, we perform a regression analysis of the competitors' CARs based on the target-related and competitors-related factors. The ordinary least squares method is used to estimate the following model: The competitors are defined as the firms with the same 4-digit SIC code. This sample includes all competitors of the proxy contest target firms with negative cumulative abnormal returns in the (−10, +10) day period around the proxy contest announcement date. For each proxy contest firm, we form an equal-weighted competitor portfolio using all firms in the same industry. As the proxy of the proxy contest firm's industry position, the size ratio is the ratio of the proxy contest firm's size to the median of its industry competitors' size. All size ratios are ranked into 5 quintiles. Quintile 1 includes the lowest ratio, and quintile 5 includes the highest ratio. Size is measured by the market capitalization on the last trading day of the year before the proxy contest and equal to shares outstanding * closing price. Event day is the proxy contest announcement day. Any non-trading event date has been converted to the next trading date. Windows is (−10, +10) day. The abnormal return (AR) is the market model residual in percent. Event study uses CRSP daily data. Market index is CRSP equally-weighted index. Estimation period ends 46 trading days before the event date. Minimum estimation length is 120 trading days. Maximum estimation length is 255 trading days. Estimate method is OLS. The average CARs for the proxy contest firms and their competitors in each quintile are reported in Panel B. The symbols $,*, **, and *** denote statistical significance at the 0.10, 0.05, 0.01 and 0.001 levels, respectively, using a generic two-tail test.
The definitions of the regression variables are in Appendix Table 8. Except for the market capitalization, the values are measured year-end in the fiscal year t-1 of the proxy fight announcement day. Since the previous tests show evidence of the information-based theory but not the competitive theory, we predict the signs of the coefficients of independent variables in the view of the information-based theory only. For all competitors, the target firms' CARs should have a positive relationship with the competitors' CARs since proxy contests is a signal of common risk. The target firms' size is negatively related to the competitors' CARs since more new information is conveyed by smaller firms which have more serious information asymmetry. The proxy contest (Harris and Raviv 1991) reduces the cost of debt so that the target firm's debt is expected to have a positive relationship with the competitors' CARs. The target firm's ROA is a measure of its profitability and is expected to be positively related to the competitors' CARs in the same industry. The target firm's book-to-market is either negatively or positively related to the target's CAR and the competitors' CARs. The competitors' debt and size are negatively associated with their growth and CARs. The competitors' ROAs are expected to be positively related to the competitors' CARs. The competitors' book-to-market ratios are either negatively or positively related to the competitors' CARs.
Industry mean ROA is expected to be positively related to the competitor's CAR since it reflects the industry average profitability. Industry mean book-to-market is expected to be either negatively or positively related to the competitors' CAR. Table 6 provides the results of the regression analysis.
In the first column of Table 6, we report the regression results for all competitors. The coefficient of the target firms' CARs is positive and significant, indicating a positive relationship between the target firms' CARs and the competitors' CARs. The coefficients of target-related factors (target firm's debt ratio, target firm's ROA and target firm's book-tomarket ratio), competitors-related factors (competitors' log total assets and competitors' return on assets), and industry factors (industry's mean return on assets) are all significant. The results show that the competitors' CARs are affected by target-related, competitorsrelated and industry factors.
To further examine the determents of the competitors' CARs, we split the competitors into two groups: the competitors of target firms with negative CARs and the competitors of target firms with positive CARs. The regression results are reported in column 2 & 3 of Table 6.
For the competitors of target firms with negative CARs, among the target-related factors, the coefficient of the target firm's ROA is 0.021 and marginally significant at 10%. Two competitors-related factors-competitors' size and ROA-have significant and negative coefficients. The coefficient of the industry factor, industry mean book-tomarket ratio, is 0.033 and that of the industry mean book-to-market is 0.001; both are significant at the 5% level. These results are consistent with H7 that the intra-industry effect is more affected by the competitors-related factors and industry factors when the CARs of the target firms are negative.
For the competitors of target firms with positive CARs, the coefficients of four target-related factors (target firm's CAR, debt, ROA and book-to-market ratio) are significant; only one coefficient of the competitors-related factor (competitors' ROA) is significant, and none of the coefficients of the industry factor is significant. These The ordinary least squares method has been used to estimate the following model: Competitors' CAR (−10, +10) = α + β 1 TCAR + β 2 TDEBT + β 3 TSIZE + β 4 TROA + β 5 TBM + β 6 CEDBT + β 7 CSIZE + β 8 CROA + β 9 CBM + β 10 INDROA + β 11 INDBM Where TCAR is the target firms' CAR during the period of (−10, +10), TDEBT is the target firm's debt ratio, TSIZE is target firm's log total assets, TROA is the target firm's return on assets, TBM is the target firm's book-to-market ratio, CDEBT is the competitors' debt ratio, CSIZE is the competitors' log total assets, CROA is the competitors' return on assets, CBM is the competitors' book-to-market, INDROA is the industry mean return on assets, and INDBM is the industry mean book-to-market ratio. Except for the market capitalization, the values are in the t-1 fiscal year as the proxy contest announcement. The symbols *, **, and *** denote statistical significance at the 0.10, 0.05 and 0.01 levels, respectively, using a generic two-tail test. All proxy fight firms are divided into one group with a negative cumulative abnormal return (CAR) and another group with a positive CAR. The industry competitors in one group are all firms with the same 4-digit SIC code but are not identified as the competitors of proxy contest firms in another group within the period of (−180, +180). We form an equal-weighted competitors portfolio for each target firm. The abnormal return (AR) is the market model residual in percentage. Event day is the proxy fight announcement date. Any nontrading event date has been converted to the next trading date. Event study uses CRSP daily data. Market index is CRSP equal-weighted index. Estimation period ends 46 trading days before the event date. Minimum estimation length is 120 trading days. Maximum estimation length is 255 trading days. Estimate method is OLS. Number denotes the number of abnormal returns available to compute the average abnormal return. The symbols $,*, **, and *** denote statistical significance at the 0.10, 0.05, 0.01 and 0.001 levels, respectively, using a generic two-tail test.
results are consistent with H8 that the intra-industry effect is more affected by the target-related factors and competitors-related factors when the CARs of the target firms are positive. In sum, the competitors of the target firms with positive CARs have been affected primarily by the target-related factors. On the contrary, the competitors of the target firms with negative CARs have been affected mainly by the competitors-related factors and industry factors.

Robustness check
One concern of our analysis is the overlapping competitors in both groups of targets. If one target firm with a positive market response and one target firm with a negative market response have the same SIC code and their proxy fight announcement day are close, they have the same competitors in the small window period. One may speculate the noise on the intra-industry effect from overlapping competitors for both target firm groups in a small period. To find out the extent of the noise effect, we construct an industry competitors sample by including all firms with the same four-digit SIC code but are not identified as the competitors of proxy contest firms in another group within the period of (−180,+180) around the announcement day.
The results are presented in Table 7. We find that the industry competitors of target firms with a negative market response experience significant average negative CARs of −0.25% in the (−10, +10) window. Comparing to the CARs of similar-size competitors of negative-CAR target firms (−4.13%) in the right half of Panel B of Table 4, the intraindustry effect of non-overlapping competitors is weaker but still statistically significant. In contrast, there are no significant CARs for industry competitors of positive-CAR target firms. The results shown in Table 7 are very similar to the results in Table 4 for competitors with the same four-digit SIC code, except that the magnitude of market reactions to the competitors of the target firms with negative CARs is smaller. The main reason is that in the robustness check, the competitors are defined as all firms with the same SIC with various sizes while in Table 4 the competitors are defined as firms with the same SIC code and a size within ±5% of the target firm. Given the finding that the intra-industry effect is most pronounced when the size of the target firm and that of the competitors are similar, it is not surprising to see that the CARs in this robustness check are much smaller than the CARs shown in Table 4. Therefore, the cluster of proxy contest firms with the same SIC code does not negate the basis of our findings.

Conclusion
We explore the intra-industry effect of proxy contests. Prior literature indicates that intra-industry reaction to proxy contests may convey the information of common industrial risks and/or expected competitive relationship change. Using a sample of 647 proxy contest initiations between 1988 and 2008, we find significant negative abnormal returns in the group of competitors of the target firms with negative market reaction, and the negative abnormal returns become greater for similar-sized competitors. In contrast, there are no significant abnormal returns in the group of competitors of the target firms with positive market reaction. Overall, our findings are in supportive of the information-based theory but not the competitive theory. The target firm's CAR during the period of (−10, +10).

TDEBT
The target firm's debt ratio.

TSIZE
The target firm's log total assets.

TROA
The target firm's return on asset.

TBM
The target firm's book-to-market ratio.

CDEBT
The competitor's debt ratio.

CSIZE
The competitor's log total assets.

CROA
The competitor's return on asset.

CBM
The competitor's book-to-market ratio.

INDROA
The industry's mean return on assets.

INDBM
The industry's mean book-to-market ratio.