Capítulo 2. El sonido como generador de identidad
3.6. Temporalización del proyecto
3.6.1. Secuenciación de las actividades
The second part of the empirical analysis relates industry characteristics to the profitability of acquisitions. Therefore, the dependent variable in Table 4.5 is the three-day cumulated ab-normal acquirer announcement return (AcquirerCAR). There is a large body of literature, which analyzes acquirer announcement returns and links them to a number of deal, firm and industry characteristics (see, for example, Moeller, Schlingemann, and Stulz, 2004). I include all standard control variables and the industry structure variables needed to test the prediction of Hypothesis 2 that acquirer announcement returns are lower for homogeneous industries than for heterogeneous industries, because the latter should exhibit fewer defensive transac-tions. Additionally I also test Hypothesis 3, which states that (1) acquirer returns of large firms are smaller because they are more likely to undertake defensive acquisitions and (2) acquirer returns are smaller for large targets because they offer a higher defensive value for the acquirer. By splitting the sample into four size quartiles, I can analyze the profitability of acquisitions undertaken by small, medium and large firms separately (models (4) to (7)). The size quartiles are assigned to firms according to their size quartile in the prior year in the complete Worldscope database. For example, in the first size quartile are all acquirers, which have a market capitalization less than the 25th percentile of all firms listed in the Worldscope database at the end of the last year. The number of observations in each size quartile shows that small firms undertake considerably fewer acquisitions than medium sized or large firms.
Insert Table 4.5 here
The results in Table 4.5 only partly support Hypothesis 2. If I only control for year and coun-try fixed effects, the coefficient for HomogeneousInduscoun-try is negative and significant at the 5% level. When controlling for other industry, deal and firm characteristics the coefficient for
HomogeneousIndustry remains effectively unchanged but it is only statistically significant at the 10% level. The economic significance is rather small. On average acquirer returns are lower by 0.22% in homogenous than in mixed firm size industries. The difference between the coefficients for HomogeneousIndustry and HeterogeneousIndustry is only significantly different from zero for model (1), where I do not control for other industry, deal and firm characteristics. Therefore, I only find limited support for Hypothesis 2. This result suggests that defensive acquisitions do possibly cluster in homogeneous firm size industries, as pre-dicted by Gorton et al, but that other effects, for example, empire building or hubris, which probably do not cluster by industry, overlay the influence of industry structure.
I include three additional industry level variables to measure the acquisition activity in the acquirer’s and the target’s industry and the competitive situation in the acquirer’s industry. In times of high acquisition activity, measured by the number of transactions in the acquirer's three-digit SIC industry over the prior two years (#DealsIndustry)123 the acquirer announce-ment returns decrease. These results are in line with Duchin and Schmidt (2007), who find that within merger waves acquirer announcement as well as buy and hold returns are lower than outside merger waves. However, their interpretation of this finding is different. They argue that empire-building managers try to hide their intentions by undertaking inefficient acquisitions inside merger waves, whereas I propose that these lower acquirer announcement returns are mainly caused by defensive transactions. This kind of (inefficient) acquisitions should be carried out mostly inside merger waves, because during these times firms are ex-posed to a particularly strong takeover threat. In such situations, quiet-life managers are likely to consider acquisitions in order to secure the independence of their firms and the associated private benefits. In Section 4.4, I will test this hypothesis by categorizing firms according to their prior acquisition activity and discuss why I find other explanations less convincing.
123 Results are robust if one uses aggregated transaction volume (AcqVolumeIndustry) instead of number of transactions.
As measure for the acquisition activity in the target’s industry, I use the liquidity index pro-posed by Schlingemann, Stulz, and Walkling (2002). The index measures the turnover of as-sets in the target’s industry and high values imply stronger competition for these asas-sets.
Therefore, I expect a negative effect of LiquidityIndex on acquirer announcement returns.
However, the coefficient is mostly positive and always statistically insignificant. It seems that LiquidityIndex has no explanatory power above what is already explained by #DealsIndustry, even though the correlation between both measures is rather moderate (Pearson correlation coefficient: 0.4).
Besides the acquisition activity, I also measure the influence of competition in the acquirer's industry. As a proxy for competitive situation the median net profit margin in the acquirer's three-digit SIC industry (MedianNetProfitMargin)124 is used. Giroud and Mueller (2009) show that competition in the product market mitigates managerial agency problems. There-fore, I expect lower acquirer announcement returns in less competitive industries. Consistent with this hypothesis a negative coefficient for MedianNetProfitMargin is estimated. Indeed, it seems that managers in non-competitive industries undertake more shareholder value destroy-ing acquisitions. However, the effect is economically small and not statistically significant in all size quartiles. The coefficients for the complete sample, models (2) and (3), imply that an increase of MedianNetProfitMargin by one standard deviation decreases acquirer announce-ment returns by about 13 basis points. Results are comparable if I use the Herfindahl index or the number of firms in the industry (#Firms) as proxies for the competitive situation (not re-ported), however, significance levels are generally lower.125 It seems that product market competition has a positive but relatively small influence on acquirer announcement returns.
124 The industrial organization literature uses the industry net profit margin commonly as an empirical proxy for the Lerner Index. The intuition is that monopolists and oligopolists can set prices in excess of marginal costs, which yields higher margins.
125 The Herfindahl index is likely to be biased in my sample because it is based solely on data from World-scope, which contains mostly public firms. Therefore, the Herfindahl index is overestimated for industries with a large number of private firms, since the market share of these firms is ignored. The median net profit
As predicted by Hypothesis 3 I find that large firms tend to undertake fewer profitable acqui-sitions than small firms. Announcement returns for large acquirer are about 0.5 percentage points smaller than for medium sized acquirer and about 1.5 percentage points smaller than for small acquirer (see model (2)). Coherent with this result the coefficient for LogEquity-Market implies that an increase of LogEquityLogEquity-Market by one standard deviation decreases ac-quirer announcement returns by about 0.5 percentage points.
The second prediction of Hypothesis 3 that the relative size of the target and acquirer has a negative effect on abnormal returns does not seem to hold given the significantly positive coefficients for RelativeSize in model (2) and (3). However, when splitting the sample into size quartiles (models (4) to (7)) I find a significantly positive effect of RelativeSize for small and medium sized acquirers but a significantly negative effect for large acquirers. Together with the first result that large firms undertake acquisitions, which are less profitable, I inter-pret this finding as evidence that only firms of a certain size are able to undertake defensive transactions and that the defensive value of the acquisition increases with target size. In other words, managers of larger firms seem to be willing to pay inefficiently high premia and this willingness increases with the size of the target, because the acquisition of a larger target of-fers better protection from future takeover attempts. Since large firms are more likely to pay excessive premia, especially for relatively large targets, we observe lower average announce-ment returns for this group of firms. Already Moeller, Schlingemann, and Stulz (2004) find a significantly negative effect of firm size on acquirer announcement returns and that the effect of relative size of target and acquirer depends on the size of the acquirer (for small firms posi-tive, for large firms negative).126 Moeller et al. do not find a clear explanation for the negative size effect but suggest that maybe managers of large firms are more frequently prone to
margin should be less affected from this problem, since the median profitability of firms in the sample is in-fluenced by all firms in the industry independent of their inclusion in the Worldscope database.
126 The latter observation could explain why prior studies have found different results for the relative size coef-ficient. For instance, Asquith, Bruner, and Mullins (1983) find a positive relationship but in Travlos (1987) it is (insignificantly) negative.
bris. Another possible explanation could be that empire-building managers in large firms with substantial free cash flows and few investment opportunities prefer to undertake acquisitions instead of paying out the excess cash flow to shareholders. If this hypothesis would be true, one should expect large firms to hold more excess cash. However, Moeller et al. do not find evidence that larger firms hold more excess cash than smaller firms.
The other results from these regressions are mostly in line with the prior literature (see, for example, Fuller, Netter, and Schlingemann, 2002; Moeller, Schlingemann, and Stulz, 2004). I find that acquisitions of public targets yield significantly lower abnormal acquirer returns whereas tender offers lead to significantly higher acquirer returns. These results hold for most specifications. Pure equity as well as pure cash acquisitions have lower abnormal returns;
however, the respective coefficients are in most specifications insignificant. The other deal and firm specific control variables are almost always insignificant.