When do vertical mergers create value?
This paper studies the market reaction to vertical mergers and explores the many rationales for vertical integration proposed in the industrial organization literature. Abnormal returns for vertical merger announcements are positive until the late 1990s, and turn negative afterward. Acquirers suffer most of the losses. We find support for the most fundamental insight in the industrial organization literature, namely, that vertical mergers generate the greatest value when undertaken in imperfectly competitive markets. We find some evidence to support ideas of asset and site specificity, that is, creating value when market exchange is difficult. We do not find support for information-based or price uncertainty theories.If markets were competitive, with no frictions, then transactions between firms could be efficiently executed with arm's length contracts. However, market frictions can lead to a rationale for integration and mergers. This insight can be traced back to Stigler (1950) and its implications are explored in numerous subsequent studies. In this paper, we classify mergers into vertical, horizontal, and conglomerate deals. We study the market reaction to these deals and, in particular, vertical mergers' announcements and examine how they are related to the underlying rationales for vertical integration modeled in the large industrial organization literature. Although conglomerate mergers and horizontal mergers have been the focus of many studies, ours is one of the few papers that examine vertical mergers.
A significant obstacle to the study of vertical mergers had been the identification of vertically related transactions. Horizontal mergers are more easily classified and were studied by Eckbo (1983, 1985) in the 1980s and, more recently, by Fee and Thomas (2004), Shahrur (2005), and Gugler and Siebert (2007). Early researchers (e.g., Spiller, 1985) hand collected data on vertical deals and consequently had relatively small samples. Recently, Fan and Goyal (2006) used the Benchmark Input Output tables compiled by the Bureau of Economic Analysis to develop a classification system for vertical deals. Our measures are similar to the ones used by Fan and Goyal (2006), but we extend their analysis in several important ways. Fan and Goyal (2006) examine market reactions to the announcement of different deals between 1962 and 1996 and find that, on average, vertical mergers are associated with significant positive wealth effects. (1) We extend the sample to a later period ending in 2002. Consistent with Fan and Goyal (2006), we find that vertical mergers are indeed associated with positive wealth effects until 1998. However, in line with Moeller, Schlingemann, and Stulz (2005), we find that vertical deals after 1998 are associated with losses to acquirers. (2)
More importantly, we complement Fan and Goyal's (2006) findings by linking market reactions to vertical deals to the underlying rationales for vertical integration. A large body of work in industrial organization, discussed in detail in the next section, indicates that vertical integration can generate value in the presence of imperfect competition. Although assumptions, details, and specifications vary across different models, the essential result that emerges is that due to the possibility of rationing inputs, shutting out competitors, and price discrimination, vertical integration is value enhancing in the presence of imperfect competition. Consistent with this theory, we find that vertical deals in noncompetitive environments are associated with higher returns relative to other vertical deals. This is not necessarily true for horizontal or conglomerate mergers. If the increased ability to shut out rivals is a source of gains in vertical mergers, then this should be reflected in losses to rivals upon the announcement of mergers. Indeed, we find that vertical mergers are detrimental to competitors of the target, as well as of the acquirer firms.
Another potential benefit of vertical integration may arise in the presence of asset or site specificity. Williamson (1983) is the first to discuss this issue, Perry (1989), among others, confirms that when firms need to invest in assets that are specialized and market exchanges are difficult, vertical integration may lead to efficient investments. In line with Caves and Bradburd (1988) and others, we use research and development (R&D) expenditures to capture asset specificity and find that vertical deals, in which both the target and acquirer are R&D intensive, are associated with higher total returns. These gains are seen only in vertical deals and not in horizontal or conglomerate mergers. (3) To gauge site specificity, we try another proxy, namely, the geographical distance between the firms in question. This variable has no effect on vertical mergers but increases the returns to horizontal deals.
Grossman and Hart (1986) demonstrate that with incomplete contracts, vertical integration may provide better investment incentives. The empirical industrial organization evidence on this matter, which is all industry specific, is mixed (note a survey by Lafontaine and Slade, 2007). We use analyst coverage to proxy for information opacity and the difficulty of arm's length contracting. We find no evidence to support the view that vertical integration leads to higher returns in the presence of information problems. However, horizontal mergers do seem to be motivated by information asymmetries.
Finally, we test to see if in the presence of price uncertainty, vertical integration may be associated with efficient production decisions and, therefore, higher value. We use the volatility of the producer price index in the acquirer and target industries to capture price uncertainty and find no effect on the value generated from vertical or any other type of merger.
In summary, we find that vertical mergers appear to generate the greatest returns when dominant firms integrate. There is weak evidence to suggest that vertical integration generates value when assets are specialized and no evidence that information problems or price uncertainty present opportunities for value-maximizing vertical integration.
The rest of the paper is organized as follows. The next section discusses the theoretical rationales for value maximizing vertical integration and develops our hypotheses. Section II describes our measures of vertical integration, and Section III describes the data and discusses the wealth effects of mergers. Section IV presents our main empirical results and Section V provides our conclusions.
I. Value Determinants of Vertical Mergers and Hypothesis
In this section, we briefly review the literature that theoretically and empirically document the value of vertical integration. This will lead to our testable hypotheses.
The theory of vertical integration can be traced back to Stigler (1950) and, in a more specific context, to Williamson (1971). Several papers, starting with Stigler (1950), indicate that in the presence of noncompetitive markets, vertical integration may be beneficial. Some theoretical studies explore the role of dominant firms. Riordan (1998), for example, finds that if there is a dominant firm in an industry with a cost advantage relative to a competitive fringe, then it will tend to vertically merge backward. This increases the dominant firm's capacity at the expense of the fringe. Both output and input prices increase in the degree of vertical integration. Although a vertical merger may lead to a decrease in profits for some of the firms in the fringe, profits increase for the dominant firm.
Other models consider "market foreclosure" or the idea that successful vertical integration may benefit the merging firms while driving competitors out of the market. Salinger's model (1988) is probably one of the best known. Salinger (1988) assumes an oligopoly in two successive stages of production. This creates the well known phenomenon of "double mark-ups" (i.e., in each stage, a noncompetitive structure increases prices further away from marginal costs). Vertical integration can eliminate the successive mark-ups in these stages of production and lower prices to consumers, but it can also drive an independent supplier or a few of them out of the market, thus raising the prices of an intermediate good. Either effect can dominate. Without going into the details of the model, it should be clear that the effects of integration described here are a function of the ability of one firm to affect the prices and production of others. In other words, an oligopoly setting is a necessary condition for this rationale for vertical integration.
Hart et al. (1990) come up with another rationale for vertical integration. They assume a setup with two potential suppliers (or upstream firms) and two potential buyers (or downstream firms). The equilibrium concept is a Nash equilibrium. Each firm decides whether to integrate or not (when they integrate they share profits) and then uncertainty about marginal costs, which drives the model, is resolved. The paper goes on to analyze numerous cases depending on the efficiency of the suppliers and the various parameters in the model.
Under some circumstances, there will be full integration. Sometimes some firms will exit and sometimes the industry will remain optimally unintegrated. Again, all results are based upon oligopoly equilibrium arguments. That is, the firms that merge affect the other firms in all stages of production. In Salinger (1988) and Hart et al. (1990), the firms that integrate exploit their market positions, and their rivals see their prices go up or they are shut out of the market. The clear implication from these models is that in oligopolistic settings, the merging firms should see improvements, whereas their rivals are at least weakly worse off.
Other papers, such as Ordover, Saloner, and Salop (1990) find that vertical integration may be value maximizing if it raises rivals' costs. In an oligopolistic setting, a firm may buy its suppliers. Thereby, it can increase the costs of the remaining independent firms. Chen and Rirordan (2007) point out that in a noncompetitive environment, vertically integrated firms can use exclusive contracts to exclude an equally efficient upstream firm and effect a downstream cartelization. DeFontenay and Gans (2005) compare the case of an upstream monopoly to a case where few competitors exist and focus on incentives created by the outcome of a possible bargaining process between upstream and downstream firms (Klein and Murphy, 1997).
The details of the outcome vary across models and depend on assumptions related to demand functions or the specification of downstream competition and existing externalities. The efficiency consequences are also debated. However, collectively, these models suggest that noncompetitive markets with the possibility of rationing, shutting out competitors, elimination of externalities, and with the opportunity for exclusive contracts and price discrimination may generate at least a private rationale for vertical integration (see also Perry, 1978, and a survey by Perry, 1989). Also, the nonmerging firms generally lose out. Therefore, if vertical mergers do not generate value in a noncompetitive environment, it will cast some doubt on a large set of theories in the industrial organization literature. This very view is also expressed in the extensive empirical survey by Lafontaine and Slade (2007) who state, "It is always the link in the chain that has market power, whether it be monopoly or monopsony power that benefits from integration. Thus, absent market power at some stage in the chain, the above motives (all the motives we discussed) cannot be relied upon to explain the data."
Perhaps more revealing, Lafontaine and Slade (2007) suggest that the market-related motives for vertical integration are unique for this type of integration. In a noncompetitive horizontal setting, the outcome is ambiguous, whereas the same situation can lead to "unambiguously beneficial" vertical mergers. Fee and Thomas (2004) and Shahrur (2005) point out that in the case of horizontal mergers, efficiencies can lead to the observed beneficial effect, even in the case of competitive industries. Similarly, Gugler and Siebert (2007) demonstrate that for the semiconductor industry, and, in particular, for the memory and microcomponents market, horizontal mergers and research joint ventures lead to an increased market share for the combined firm, consistent with efficiency gains. We focus on vertical mergers, but because we do classify mergers as horizontal, vertical, and other, we can try, to some extent, to verify the findings in the horizontal mergers' literature as well.
The vast majority of the empirical industrial organization literature on vertical mergers covers one industry and sometimes includes industry-specific proxies, such as paper capacity for the pulp and paper industry (Ohanian, 1994) or the number of rooms for the hospitality industry (Kehoe, 1996). Our work focuses on some general properties that may make vertical integration advantageous. Therefore, we seek proxies that cut across industries. (4)
Most of our tests consider mergers where both the buyer and the target have market power. Spiller (1985) finds that gains from mergers and the division of gains depend upon whether or not only one side has market power. He tests his ideas, including the impact of vertical integration on systematic risk, on a small sample of vertical mergers. As we study vertical mergers across several industries, the proxies of market power used in our tests are general so as to be meaningful across all industries studied. We expect these noisy proxies of market power to pick up large gains to vertical integration in the presence of strong noncompetitive forces (i.e., when both the merging firms are dominant). However, weather these noisy proxies are powerful enough to pick up weaker cases of market power, that is, when only one of the merging firms is dominant, is an empirical question.
The discussion in the previous paragraphs naturally leads to our first hypothesis:
H1: Vertical mergers should be more successful in a noncompetitive environment. Rivals should be worse off.
The presence of transaction costs may affect the optimality of vertical integration as pointed out by Williamson (1971), Klein, Crawford, and Alchian (1978), Perry (1989), and others. Joskow (1985) follows Williamson (1971) in identifying two types of specificity that are relevant to our tests: 1) site specificity: buyer and seller are in a "cheek by jowl" relationship with one another, reflecting an ex ante decision to minimize inventory and transportation expenses, and 2) physical asset specificity: when one or both parties to the transaction make investments in equipment and machinery that involve design characteristics specific to the transaction and which have lower values in alternative uses. The latter idea is that when firms need to invest in assets that are specialized and when the market exchange of these assets is costly, vertical integration may align the incentives of the parties involved and may lead to efficient investments (Joskow, 1988). Lieberman (1991) studies plant-level and firm-level integration in the chemical industry and also concludes that the probability of vertical integration increases with asset specificity (proxied by fixed costs and the existence of gas fueled plants) and input variability.
The importance of asset specificity is also noted by Caves and Bradburd (1988) who remark, "The chief empirical predictors of vertical integration coming from the transaction cost model are small numbers of transactors on both sides of the market ex ante and the prevalence of transaction specific assets and switching costs that create ex post lock in problems with arm-length transactions." Caves and Bradburd (1988) use R&D intensity to proxy for asset specificity and indicate that it significantly affects the probability of mergers. Further support for the use of R&D intensity is provided by Levy (1985) who says, "Firms producing research intensive products are expected to vertically integrate because the nature of the new technologies is difficult to predict." Masten, Meehan, and Snyder (1989) and Anderson and Scmittlein (1984) are among other researchers who use R&D as a proxy for asset specificity.
Several studies also test the other notion of specificity, site specificity. Most of these studies again focus on specific industries and use the distance between the merging entities as a proxy for site specificity. Spiller (1985) suggests that "site specific assets can increase the viability of a vertical merger ... for example, a plant which, if located near to another reduces the profitability from selling to or buying from another firm." He finds that distance negatively affects stock returns in mergers. In contrast, Levy (1985) observes no correlation between distance and returns but finds that R&D intensity affects returns. (5) We follow this body of work and use R&D intensity to capture asset specificity and distance to proxy for site specificity. Thus, we formulate our second hypothesis:
H2: Vertical mergers should be associated with higher returns when the target and/or the acquirer have specific assets. Asset specificity is proxied by R&D intensity, and site specificity is proxied by geographical distance.
Some of the studies of vertical integration focus on incomplete contracts and the incentives they create. If arm's length contracts are harder to write, enforce, and monitor on the outside relative to contracts within the firm, then vertical integration can be efficient. Grossman and Hart (1986) suggest that investment incentives may differ as a result of the allocation of control rights ex post. Therefore, depending on relatedness, vertical integration may provide the correct investment incentives. Hughes and Kao (2001) model more directly the advantage gained by information sharing between upstream and downstream firms in noncompetitive markets. If vertical integration is an efficient tool in a world of asymmetric information and incomplete contracts, then vertical mergers should be associated with higher returns when contracting is harder. We use the degree of information opacity of the target and acquiring firm to capture instances where arm's length contracting is less effective and formulate our next hypothesis:
H3: Vertical mergers should be associated with higher returns when there is less public information available about the target and the acquirer.
Another rationale for vertical integration is price uncertainty. Several models discuss various manifestations of this idea. Carlton (1977) suggests that if price uncertainty exists, firms will charge a premium to account for the possibility of not being able to sell their entire output. If integration can provide a higher probability of usage, then vertical integration will provide cost savings. More recently, Baker, Gibbons, and Murphy (2002) discuss incentives in relational contracts for upstream and downstream parties. One of their results concerns prices. Low prices create incentives for downstream consumers to renege on contracts, whereas high prices create incentives for upstream producers to renege. Therefore, they conclude that a large price dispersion (a big difference between the two outcomes in their model) encourages firms to seek vertical integration, where temptations to renege on contracts do not exist. The popular press discusses price uncertainty as well, but is much less clear about the issues.
Therefore, we propose:
H4: Vertical mergers should be associated with higher returns when there is greater price uncertainty in the acquirer and the target industries.
We now proceed to discuss our empirical strategy and our findings.
II. Measuring Vertical Integration
The most important empirical barrier to the analysis of vertical merger is the identification of mergers as horizontal, vertical, or conglomerate. We follow Fan and Goyal (2006) and Acemoglu et al. (2009) and use the industry commodity flow information in the Use Table of Benchmark Input-Output Accounts for the US Economy compiled by the Bureau of Economic Analysis. The Use Table is a matrix containing the value of commodity flows between each pair of roughly 500 private sector intermediate input/output industries. If we denote by [a.sub.ij] the dollar value of i's output required to produce industry j's total output, and then divide [a.sub.ij]. by the dollar value of industryj's total output, the resulting fraction, denoted by [v.sub.ij], represents the dollar value of industry i's output required to produce one dollar's worth of industryj's output. For example, if industry [v.sub.ij] buys from industry i $500 worth of goods, and industryj then sells $2,000 worth of output, v0 is 500/2000 = 0.25. We use 1/2([v.sub.ij] + [v.sub.ji]) to capture the potential for vertical integration between industries i and j (this is similar to Fan and Goyal, 2006; Acemoglu et al., 2009). We use the average of the coefficients for vertical relation as very often industries sell in both directions, and ignoring one direction might bias our analysis. We convert the acquirer and the target Center for Research in Security Prices (CRSP) Standard Industrial Classification (SIC) code to initial offering (IO) industry codes and classify mergers as vertically related if the corresponding vertical coefficient is larger than a certain cutoff. For robustness, we consider three alternative cutoffs: 1%, 5%, and 10%. (6)
We use primary segments for our analysis. Some firms have multiple segments. Therefore, a merger may appear unrelated when viewed from the perspective of primary segments but may actually be related if viewed from the perspective of secondary segments. If this is the case, we may be counting too few vertical mergers. However, primary segments generally provide the bulk of the business for most firms. For example, Shahrur (2005) finds that 90% of targets and 77% of bidders in his sample are single-segment firms. Moreover, for 95% of targets and 90% of bidders, more than 75% of their business is derived from their primary segment. Fan and Goyal (2006), who perform robustness checks on pairs of primary and secondary segments, find that their results are essentially unchanged with this specification. Thus, it seems that using primary segments to classify vertical mergers does not qualitatively change the results. Because reporting of nonprimary segments is not without its problems (Bens, Berger, and Monahan, 2009), we use primary segments for our classification. In other words, the term "pure vertical" implies mergers with a high vertical potential in their primary segments. This is also in the spirit of Fee and Thomas (2004) who define horizontal mergers as a "bidder and a target with at least one segment each with the same four-digit SIC code" and it also follows Fan and Goyal (2006).
A merger transaction is classified as horizontal if both the acquirer and the target are in the same industry as captured by the four-digit CRSP SIC code. If the four-digit SIC industry has a high vertical relation coefficient with itself (i.e., the industry uses a high fraction of its own output), then a horizontal merger can also be classified as vertically related. We refer to such transactions as mixed horizontal vertical mergers. Conversely, pure horizontal mergers are mergers classified as horizontal mergers that are not vertically related. Similarly, pure vertical mergers are those that are classified as vertically related but are not horizontal. Finally, if a transaction is neither horizontal nor vertical, it is classified as a conglomerate merger.
As previously mentioned, we use three cutoffs (1%, 5%, and 10%) to classify a merger as vertically related. We use three different definitions of industry, four-digit CRSP SIC codes as mentioned before, two-digit SIC codes, as well as Fama and French (1988) industry classifications. This analysis leads to nine different classifications of mergers into the different types (i.e., vertical, horizontal, mixed, and conglomerate). (7) We use these different classifications to test the strength and robustness of our results.
III. Merger Sample Description and Wealth Effects
Our data on acquisitions are from the Securities Data Company's US Mergers and Acquisitions Database. We select mergers and acquisitions announcements in which both the target and acquirer are US publicly listed firms with announcement dates from 1979 to 2002. We consider only completed deals and exclude leaseback offers (LBOs), spin-offs, recapitalizations, self-tender and exchange offers, repurchases, minority stake purchases, acquisitions of remaining interest, and asset sales. We further require that the acquirer and target price data exist in the CRSP and that we are able to calculate the vertically related coefficient. Our final sample consists of 1,692 transactions.
Table I and Figure 1 illustrate the number of acquisitions, as well as the proportion of mergers, that are classified as vertically related over the years. When we use four-digit SIC codes to classify the target and acquirer, we find that about 9.73% of all deals are vertically related according to the strictest definition of vertical relatedness (10% cutoff'). If we use this strict definition, only 5.25% of deals can be classified as pure vertical mergers. As expected, the percentage of deals classified as vertical increases with lower cutoffs. With a 1% cutoff, 39% of deals are classified as vertically related and 21.58% as pure vertical deals. A similar trend is evident when we use Fama-French industry classifications (see Table I, Panel B, as well as Figure 2). As Fama-French industries are more broadly defined relative to the four-digit SIC, a larger fraction of the vertically related deals are also classified as horizontally related. Consequently, the fraction of deals classified as pure vertical is smaller relative to the four-digit industry classification. Using a 1% cutoff, only 9.88% of deals are classified as pure vertical when we use the Fama-French classification compared to 21.58% using the four-digit SIC. This suggests that the use of different industry classifications and cutoffs may have an impact on the categorization of mergers. (8)
We report most of our main results using the four-digit SIC classification, which is more precise, and a 1% cutoff. We also report the impact on these results as we move to stricter cutoffs and to Fama-French industry classifications. Another feature of Table I, which is consistent with most other work on mergers, is the presence of significant clusters over time in merger activities including merger waves in the late 1980s and the late 1990s. (9)
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
We consider acquirer and target wealth effects using standard event study methodology. We estimate cumulative abnormal returns (CARs) over the (-1, +1) day window using CRSP value-weighted index returns with the parameters estimated over the 255-day estimation period that ends 46 days prior to the initial merger announcement. (10) The total or aggregate return is the weighted return of the acquirer and the target, where the weights are relative market capitalization 10 days prior to the announcement.
We find that pure vertical mergers are associated with positive total returns of 1.1% (Table II). The total return to vertical deals is significantly lower than the 4.7% earned by pure horizontal mergers. The better performance of horizontal mergers is consistent with Fee and Thomas (2004), who cover a similar time period. The average total return to vertical deals however, is also lower than the 1.9% earned by conglomerate mergers though this underperformance is only significant with the 1% cutoff and not with the 5% cutoff. As this result is in contrast to the findings of Fan and Goyal (2006), we examine it further by splitting our sample into transactions taking place before and after 1998.
There appears to be a decline in returns for all types of mergers after 1998, consistent with the findings of Moeller et al. (2005). Consistent with Fan and Goyal (2006), the average return to vertical deals prior to 1998 is positive and significant (2.3%). The returns decline to a -1.1% after 1998. The decline in returns is not confined to just vertical deals as the return to horizontal mergers also declines from 6.1% before 1998 to 1.7% in the later period. Moeller et al. (2005) report that the decline in merger performance after 1998 is due to an increase in target size, as well as the lower book to market ratio of the acquirers. Consistent with these results, we find that vertical deals in the post- 1998 period tend to be by acquirers with lower book to market ratios and involve larger targets (Table II, Panel C). Further, there is negative skewness in acquirer returns for vertical deals that is also consistent with the findings of Moeller et al. (2005) for all deals. (11) Therefore, it appears that the decline in performance of vertical deals is part of the broader trend studied by Moeller et al. (2005). (12)
When we decompose total returns into target returns and acquirer returns, we find that in line with all of the prior literature starting from Bradley, Desai, and Kim (1988), including Andrade, Mitchell, and Stafford (2001) and, more recently, Moeller et al. (2005), most of the value accrues to the targets. (13) Not surprisingly, target returns are fairly similar across the different type of mergers and across the different time periods.
One other issue that may explain the apparent relative worsening of the performance of vertical deals in the later period is that, on average, there were relatively more of them after 1998. If we consider, for example, pure vertical deals at a 5% cutoff, they account for 11.7% of the sample in 1981-1997 versus 17.2% in later years. A larger number of deals may indicate decreasing marginal returns. This may, in turn, explain the negative returns to vertical deals after 1998. We also looked at deals one industry at a time. There is a concentration of vertical deals in two-digit SIC code 28, Chemicals and Allied Products. We find that after 1998, SIC code 28 accounted for about 10% of all nonvertical deals and 22.54% of all vertical deals. This concentration of vertical deals in SIC code 28 in not restricted to the post-1998 period. However, whereas vertical deals in SIC code 28 had a mean return of -2.2% in the post-1998 period, these deals had a mean return of +2.6% prior to 1998. This deterioration in the returns to SIC code 28 after 1998 is confined to vertical deals. Nonvertical deals in SIC code 28 earned 2.4% until 1998 and +1.3% post-1998. The poor performance of vertical deals in SIC code 28 in the post-1998 period may have contributed to the overall poor performance of vertical deals post-1998. (14)
Next, we control for firm and deal characteristics. We control for the mode of payment by including a dummy variable (Anystock) that takes a value one if stock is used for payment. As cash deals are associated with higher returns in most other work, we expect the coefficient of Anystock to be negative and significant. We control for acquirer size by including acquirer total assets in the year prior to the announcement of the deal. (15) We include the ratio of target size to acquire size (TarSize_AcqSize) to control for the relative size of the transaction. As small targets relative to the acquirer are likely to have lower wealth effects, we expect the coefficient of relative size to be positive. Naturally, we include a dummy for pure horizontal deals, pure vertical deals, and mixed horizontal vertical deals. Conglomerate is the default.
Table III contains the ordinary least square (OLS) regressions. The findings are qualitatively similar to the univariate results discussed earlier. Pure horizontal mergers feature significantly higher returns over the entire period, and in particular prior to 1998 (Column 3). This is consistent with the findings of Shahrur (2005) and Fee and Thomas (2004) who interpret the positive outcome of horizontal mergers as the result of increases in efficiency. The coefficient of pure vertical mergers is not significant for the whole sample, implying that pure vertical mergers are not much different from conglomerate mergers. However, for the period after 1998, pure vertical mergers are associated with significantly lower total returns after controlling for size, year dummies, and the type of transaction. Note that returns for horizontal mergers were positive and significant prior to 1998, but insignificantly different from zero afterward.
In line with the evidence in Table II, we find that most of the returns accrue to the target and that target returns are not significantly different across merger types. It is the significantly lower acquirer returns that drive the outcomes. In Appendix A, we re-estimate Table III by adding controls for industry. We first include dummies for the two-digit SIC codes of the acquirer (results are reported in Panel A). The results are qualitatively similar. Horizontal mergers have a higher total return over the entire sample period, a positive return prior to 1998, and an insignificant return after 1998. The returns for vertical mergers are insignificant for the entire period but turn negative after 1998. This is similar to the results reported in Table III (which do not include industry dummies). There is a small difference for mixed deals.
In Appendix A, Panel B, we control for the industry of the target in the same fashion. Results are similar, except that the negative returns for vertical mergers are not statistically significant. We also examine the effect of introducing a dummy to capture technology deals. (16) This dummy takes a value of one if the either the acquirer or the target belongs to SIC code 35 (computer equipment), 36 (electrical equipment), or 73 (business services). As seen in Panel C of Appendix A, the results are very similar to Table III that has no industry controls.
In summary, the basic results do not change when we add controls. Horizontal mergers are associated with the largest gains and the performance of all mergers declines after 1998 as has been already documented by Moeller et al. (2005). There are more vertical mergers in the later period, which might have led to a decline in performance. This trend does not seem to be only industry or tech driven; however, we do observe much poorer performance of vertical mergers in the chemical industry.
IV. Determinants of Vertical Mergers and Returns
In this section, we test our hypotheses. It may be the case that although vertical mergers overall do poorly, certain types of vertical mergers make economic sense.
A. Market Share and Industry Concentration
As discussed in Section I, several theories proposed by Williamson (1971), Salinger (1988), Ordover et al. (1990), Hart et al. (1990), Riordan (1998), Chert and Rirodan (2007), DeFontenay and Gans (2005), and many others suggest that vertical integration can generate value in noncompetitive industries. These observations are summarized in Hypothesis 1.
To study the impact of acquirer and target industry market structures, we calculate target and acquirer market shares. Target market share is defined as the sales of the target company divided by industry total sales in the year prior to the announcement of the merger. Industry is defined by a four-digit SIC classification and encompasses all firms with data in Compustat. Similarly, the acquirer market share is the share of the acquirer divided by total industry sales. Because the effect of noncompetitive forces is likely to be strongest in the tails of the distribution, we create a dummy (High Share) that takes a value one when both the acquirer and the target are in the top decile of all observations. In other words, High Share dummy takes a value one when both the target and the acquirer are dominant players in their respective industries. The top deciles include targets with market shares larger than 10% and acquirers with market shares in excess of 40% in their respective industries.
The results for this estimation are reported in Table IV. We estimate the model separately for each type of merger, as the impact of market shares on total returns is likely to vary by merger type. Consistent with Hypothesis 1, we find that when both the acquirer and the target are dominant players in their respective industries, vertical mergers are associated with higher total returns. The high market share is positive for all mergers, but it is significant (at 9% and 7%, respectively) only for vertical mergers. We note that separately, the acquirer market share and the target market share are not significant suggesting that vertical mergers add value only if both firms have a large market position. This finding supports the set of models discussed earlier in Riordan (1998), Salinger (1988), Hart et al. (1990), Ordover et al. (1990), and Chen and Rirodan (2007).
The fact that one-sided market power does not lead to positive returns in our tests seems not to support one sided models such as Spiller (1985). We should note, however, that our proxies for market power are noisy. Thus, it may be that we simply cannot detect the impact of one-sided market power. Therefore, we conclude that either there are no gains to one-sided market power or that our tests do not have enough power to detect these gains if they do exist.
Our result that vertical mergers have higher returns when the target and acquirer are dominant firms in the industry is fairly robust. When we use Fama-French industry classifications (and a similar cutoff of 1%), we continue to find that the high share dummy is positive and significant (Table IV, Panel B). (17) However, with a 5% cutoff for classifying vertical deals, the high share dummy, although still positive, loses significance. As discussed earlier, with these stricter criteria, there are fewer deals that are classified as pure vertical and the small number of observations may account for the loss of significance. The other control variables have signs consistent with most other work on mergers. Stock mergers are worse and relative size helps. We notice that the market share variables are not significant for horizontal deals, supporting the results of Fee and Thomas (2004) and Shahrur (2005), as well as the comment by Lafontaine and Slade (2007) reported earlier. This strengthens the view that our findings are due to rationale mentioned in the vertical integration models rather than to general issues relating to market power.
We also examine the distribution of total returns to targets and acquirers. (18) Consistent with the previous discussion, the high share dummy is not significant in explaining target and acquirer returns except for vertical deals where it significantly increases target returns (Table V). This suggests that the targets have greater bargaining power in situations where there are gains to be divided up in vertical mergers.
If the gains to vertical merger are based on the industrial organization theories discussed earlier, they should be more noteworthy in concentrated industries. Dominant firms should have a greater ability to impose costs on rivals in such cases. To examine this idea further, we also include target and acquirer industry Herfindahl indices, calculated at the four-digit SIC level. An industry is defined as a concentrated industry if the Herfindahl index is in the top quartile for all observations. In Table VI, we include the Herfindahl dummy and an interactive dummy, which characterizes firms with large market shares operating in concentrated industries. Vertical integration between dominant acquirers and targets, both operating in concentrated industries, significantly increases total returns. CARs in these cases are 7.5% higher than in other vertical mergers controlling for firm and deal characteristics. No such increase in returns can be observed for other types of mergers. As we can see in Panel B, this result is robust to using Fama-French industries to characterize the companies in our sample. (19) The overall picture is similar to the previous tables (and consistent with work on horizontal mergers). These positive effects do not occur if the merging firms are relatively small and unable to influence the price of the intermediate or final good. (20)
None of the industrial organization theories discussed previously predicts how gains should be distributed between the acquirer and the target. As argued earlier, in a large fraction of mergers, most of the gains are captured by the targets. This is the case in our sample as well. Our tests consider mergers between dominant firms and still find that most gains accrue to the target reinforcing the view of scarcity of targets relative to potential acquirers.
If the gain in vertical mergers, when the acquirer and target are large firms, is due to the increased ability to impose costs on rival firms through rationing or price discrimination, then other firms in the industry should experience negative returns at the announcement of the deal. As discussed earlier, in models such as Salinger (1988), Hart et al. (1990), or Chen and Riordan (2007), as well as many other models of integration, rival firms should be affected negatively. To test this implication, in Table VII, we calculate abnormal returns for all firms in the acquirer's industry, except the acquirer, over the -1 to 1 day window surrounding the announcement of the transaction. We then calculate the market value-weighted average returns for these other firms in the acquirer industry. Similarly, we calculate the average abnormal return to all other firms in the target firms industry. In the target industry, two effects are possible. If we follow the models strictly, other firms in the acquirer industry may be shut out or see price increases, in which case rivals can expect a negative market reaction upon announcement of the merger. Alternatively, the acquisition may increase the likelihood that other firms in the target industry will be acquired. Therefore, the announcement may have a positive effect on other firms in the target industry. This information effect has been documented extensively in the literature (Song and Walkling, 2000; Akhigbe, Borde, and Whtye, 2000). (21)
For vertical transactions, we find that the returns to rivals in the acquirer industry are negative and significantly more negative than returns for horizontal and mixed transactions (for the median, although not for the mean). These results are consistent with the theories we previously covered. The returns of other firms in the target industry are positive, but as argued earlier, it may be that the targets are now more desirable.
Table VIII includes deal characteristics. The results are qualitatively similar. The table supports foreclosure and raising rival costs theories. In particular, we find that the average returns to competing firms in both the acquirer and the target industries are negative and significant at 9% and 10%, respectively, when the acquirer and the target have high market share in vertical deals. (22) The non-competitive realm is where the theories should have most bite, as argued extensively earlier in the paper, and that is what we find here. There is no significant correlation between the market share of the acquirer and the target and the returns to rival firms for other types of mergers. This provides some support for the idea that gains in vertical mergers arise from an increased ability to impose costs on rivals.
We should note that Fee and Thomas (2004) and Shahrur (2005) find that in horizontal mergers, rivals' returns are generally positive. Our findings and their results allow us to contrast the motives for horizontal mergers (efficiency, according to their interpretation, or the one in Gugler and Siebert, 2007) with the noncompetitive flavor of successful vertical mergers in our sample. We should note that in 2.3% of all vertical deals, both the acquirer and the target have large shares. In comparison, in only 0.8% of the horizontal mergers in our sample are both the acquirer and target large. This further supports the idea that horizontal and vertical mergers are undertaken for different reasons.
In summary, our findings thus far agree with the initial notion, going back to Williamson (1971), Salinger (1988), Ordover et al. (1990), Hart et al. (1990), Riordan (1998), Chen and Rirodan (2007), DeFontenay and Gans (2005), and many others that vertical integration will be value enhancing in noncompetitive environments. We can also offer support for other models that follow the transaction costs hypotheses. The empirical industrial organization literature (Lieberman, 1991) suggests that high concentrations increase the probability of integration. In our work, we do not have a similar measure, but we can see that the viability of mergers, as measured by market reactions, increases with market power and concentration. It is interesting to note that both the acquirer and the target concentrations matter, confirming the idea that noncompetitive industries are the source of gains in this type of merger.
As noted, the previous tests (in particular Tables IV and V) that included shares of both the acquirer and the target seem to indicate that one-sided market power does not matter. To test this notion further (Spiller, 1985), we identify mergers in which the target has high market power and the acquirer does not and vice versa, based upon definitions presented earlier in this section. We find that one-sided target market power has no significant effect on total returns, target returns, or acquirer returns for all merger types. Our result, that positive returns in vertical deals is seen only in mergers of dominant firms, is not surprising as these deals should be associated with the largest gains from shutting out their rivals. It is quite likely that the same tests, based on noisier proxies that span across industries, are not able to detect the smaller gains in vertical deals with one-sided market power.
B. Asset Specificity, R&D Expenditures, and Geographic Proximity
As noted in our discussion leading to Hypothesis 2, we follow Joskow (1985) in empirically distinguishing between asset specificity and site specificity. In the presence of asset specificity, vertical integration may facilitate the alignment of incentives of the two parties and ensure efficient investment. Consequently, Hypothesis 2 suggests that asset-specific vertical deals should be associated with higher total returns. We follow Caves and Bradburd (1988), Masten et al. (1989), Levy (1985), and Anderson and Schmittlein (1984) and use R&D expenditures normalized by sales to proxy for asset specificity. If the acquirer, the target, or both engage in high levels of R&D, vertical integration may reduce the inefficiencies associated with market exchanges of these assets. Empirically, we classify targets and acquirers as high R&D if they are in the top deciles of R&D for the sample. This includes targets with a ratio of 25% or more of R&D over sales and acquirers with a 40% or higher R&D to sales ratio.
For site specificity, we follow studies such as Levy (1985) and Masten et al. (1989) and use geographic proximity between the target and acquirer as a proxy. In particular, we obtain the location of the target and acquirer headquarters from Compustat. We use this location to create a "SameMSA" dummy that takes on a value one if the target and acquirer are in the same MSA.
Consistent with Hypothesis 2, we find that when the target and acquirer are both high R&D firms, vertical mergers are associated with higher total returns (Table IX). Such deals are associated with 4% higher returns relative to other vertical mergers. Further, there is no such gain to high R&D in other types of mergers. This suggests that as in the case of oligopolies, when both the target and the acquirer are at the high end of the asset specificity range, returns are significantly higher. Because R&D is a somewhat crude measure of what we are trying to gauge, we see the impact when we focus on the more extreme cases and when both partners are high on this scale. It also seems that acquirer and target R&D are significant (sometimes positive and sometimes negative) for various types of mergers. The interpretation may have to do with the ability to integrate specialized firms. However, it seems, on margin, that mutual specificity is important and helpful to vertical deals only. (23)
The result that vertical integration is associated with higher total returns when both the targets and the acquirers are high R&D firms holds when we use different industry classifications (e.g., Fama-French [See Panel B] or two-digit SIC codes). The result is less significant when we use stricter classification criteria, such as a 5% cutoff to classify vertical mergers. As discussed earlier, the number of deals classified as pure vertical drops with the stricter criteria and this may account for the lack of significance. The results are also sensitive to our definition of what is high R&D. If we broaden our group of high R&D from the top decile to the top quartile, High R&D dummy is positive, but no longer significant. This suggests that only when R&D is substantial do we observe the R&D effect. However, this is also the gist of the theory. (24)
As noted, site specificity is usually proxied by geographical distance, following papers such as LeW (1985), Spiller (1985), or Masten et al. (1989). We find no significant difference between vertical mergers that occur within the same MSA and others that span different regions. Consistent with the evidence in Kedia, Panchapagesan, and Uysal (2008), we find that acquirers in proximate deals make higher returns in all types of mergers except vertical mergers (Table X). Kedia et al. (2008) interpret the higher acquirer returns in proximate deals as evidence of information advantages. Thus, it is possible that geographic proximity is not a good proxy for site specificity in this context. In summary, our findings thus far offer some support for the transaction costs approach. (25)
An additional prediction of transaction cost theories is that wealth effects will be greater during periods of economic shocks in the presence of relationship-specific assets. That is, when contracting is more difficult, mergers should be more useful. Empirically, we should find that wealth gains are positively related to measures of asset specificity interacted with measures of economic shocks at the industry level. (26) In Appendix B, we interact the high R&D variable with a dummy for the years the economy was in a recession. We obtain the recession years data from the National Bureau of Economic Research (NBER) business cycles data. In particular, 1980, 1981, 1982, 1990, 1991, and 2001 are coded as recession years. We find that when we use Fama-French industries (as opposed to SIC codes), there is evidence that high-R&D firms are associated with greater total returns in years of recession.
C. Information Asymmetry
To test Hypothesis 3, regarding the relationship between merger returns and available information about the target and the acquirer, we gather data from the Institutional Brokers Estimate System (IBES) about the number of analysts who provide earning estimates for the target and the acquirer in the year prior to the merger announcement. The dummy Target Infodum takes a value of one if the target has no analyst coverage. Similarly, the dummy Acquirer Infodum takes a value of one if the acquirer has no analyst coverage. Finally, the dummy High Information Asymmetry takes a value one when both the acquirer and the target have no analyst coverage.
There is no evidence that information asymmetry, as captured by analyst coverage, has any impact on total returns in vertical deals as seen in Table XI. If we interpret geographical proximity in an informational light, our results from the previous section support this view as well. However, deals with a horizontal element seem to be associated with higher returns in the presence of high information asymmetry. We suggest that when both the acquirer and the target have information problems with respect to the financial markets, but operate in the same industry, it is likely that they are able to evaluate each other's prospects properly. It may not be possible for acquirers and targets that span different industries to resolve their information problems via integration. The only positive finding we can report for vertical mergers is if one considers R&D to be a measure of informational asymmetry. In that case, one can interpret the results in Table XI as providing some support for the information asymmetry idea in vertical mergers, as well. (27)
D. Price Uncertainty
Finally, we examine the role of price uncertainty in making vertical integration valuable. We gather data from the Bureau of Labor Statistics on the Producer Price Index (PPI) for the target and acquirer industries. The PPI program measures the average change over time in the selling prices received by domestic producers for their output. As a measure of price uncertainty, we compute the variance of the monthly PPI in the 36 months prior to the merger. In unreported regressions, we observe that merger returns are unrelated to price uncertainty. The coefficient of our uncertainty measures shifts between positive and negative for different classifications of vertical integration and is never significant. Price uncertainty is also never significant when explaining returns in other type of mergers. This does not support the results of papers such as Lieberman (1991) for the chemical industry. If we could find a relationship, we could support models such as Carlton (1977) that suggest that vertical integration allows firms to protect themselves from output price fluctuations. Naturally, it may be that our crude proxy is responsible for the lack of significance.
V. Conclusions
In this paper, we explore the market reactions to vertical mergers and correlate the returns with predictions based on industrial organization theories. We document a trend of declining merger returns over the 1990s consistent with prior work, such as Moeller et al. (2005). This is also reflected in the returns to vertical deals, which were positive prior to 1998 in line with the finding of Fan and Goyal (2006) and declined afterward with negative and significant returns to acquirers.
We find support for the most fundamental insight in the industrial organization literature, that is, that vertical deals between partners with market power, especially in concentrated industries, can be successful. In our sample, such deals are associated with significantly higher returns. This is not true for horizontal mergers. This latter result is consistent with several recent studies suggesting that horizontal mergers are not initiated for anticompetitive motives. We find less support for other suggested reasons for vertical integration. In particular, we find only some support for the view that vertical mergers generate value when firms invest in specialized assets making market exchanges difficult. There is little evidence to support the view that information-based contracting problems or price uncertainty, at least as captured in this paper, generate a value-maximizing rationale for vertical integration. However, informational problems and close proximity can increase returns to horizontal mergers. Our results, along with those of Moeller et al. (2005), suggest that mergers undertaken for the right reasons can be beneficial. Our findings also suggest that although horizontal mergers seem to work by increasing efficiencies, vertical mergers are best when they take advantage of noncompetitive environments.
Appendix A: Effect of Industry Dummies on Total Returns The table displays the results of an OLS regression with dummies for the two-digit SIC included. Panel A includes dummies for the two-digit SIC of the acquirer. Panel B includes dummies for the two-digit SIC of the target. The classification of deals is based on the four-digit SIC from CRSP with a 1% cutoff for vertical classification. Pure horizontal (vertical) dummy takes a value of one if the transaction is only a horizontal (vertical) transaction. The horizontal vertical dummy takes a value of one if the deal is both a vertical and a horizontal transaction. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy if stock is used for payment in the transaction and Acq_Assets is the total assets of the acquirer in the year prior to the announcement. High Tech dummy takes a value of one if either the target or the acquirer are in two-digit SIC codes 35, 36, or 73. The p-values are in parentheses. Panel A. Total Returns Tarsize AcgSize 0.028 0.032 0.008 (0.00) (0.00) (0.47) AnyStock -0.029 -0.015 -0.048 (0.00) (0.06) (0.00) Acq_Assets 0 0 0 (0.44) (0.57) (0.62) Pure horizontal dummy 0.033 0.046 0.003 (0.00) (0.00) (0.87) Pure vertical dummy -0.006 0.001 -0.023 (0.42) (0.9) (0.05) Horizontal vertical dummy -0.003 0.005 -0.013 (0.69) (0.65) (0.29) High tech dummy Constant 0.034 0.017 0.045 (0.00) (0.17) (0.00) Year effects Yes Yes Yes Industry effects Yes Yes Yes Years included All obs <1998 [greater than or equal to] 1998 Observations 1579 1013 566 [R.sup.2] 0.13 0.14 0.15 Panel B. Total Returns Tarsize AcgSize 0.027 0.03 0.009 (0.00) (0.00) (0.41) AnyStock -0.028 -0.013 -0.05 (0.00) (0.1) (0.00) Acq_Assets 0 0 0 (0.64) (0.42) (0.66) Pure horizontal dummy 0.032 0.044 0.009 (0.00) (0.00) (0.58) Pure vertical dummy -0.002 0.003 -0.014 (0.8) (0.75) (0.25') Horizontal vertical dummy -0.002 0.007 -0.014 (0.85) (0.52) (0.29) High tech dummy Constant 0.032 0.017 0.041 (0.00) (0.18) (0.00) Year effects Yes Yes Yes Industry effects Yes Yes Yes Years included All obs < 1998 [greater than or equal to] 1998 Observations 1579 1013 566 [R.sup.2] 0.13 0.16 0.11 Panel C. Total Returns Tarsize AcgSize 0.028 0.033 0.01 (0.00) (0.00) (0.33) AnyStock -0.03 -0.016 -0.048 (0.00) (0.03) (0.00) Acq_Assets 0 0 0 (0.51) (0.45) (0.57) Pure horizontal dummy 0.029 0.041 0.006 (0.00) (0.00) -0.68 Pure vertical dummy -0.006 0.003 -0.021 (0.38) (0.74) (0.04) Horizontal vertical dummy -0.006 0.004 -0.019 (0.41) (0.71) (0.09) High tech dummy -0.011 -0.008 -0.013 (0.05) (0.25) (0.13) Constant 0.041 0.022 0.042 (0.00) -0.07 (0.01) Year effects Yes Yes Yes Industry effects No No No Years included All obs < 1998 [greater than or equal to] 1998 Observations 1579 1013 566 [R.sup.2] 0.09 0.1 0.08 Appendix B: Interaction with Business Cycles The dependent variable is total returns measured over the [-1,1] window. Panel A uses the four-digit SIC with a 1% cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a 1% cutoff for classification. Separate results for are reported for the subgroup: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Tarsize-Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value of one if there is any stock payment for the transaction. Acq-Assets is the size of the acquirer. Target (acquirer) R&D is the ratio of the target (acquirer) R&D to sales. High R&D dummy takes a value of one if both the target and the acquirer have R&D to sales in the top 10%. High R&D * Recession dummy takes a value of one if both the acquirer and the target are high R&D (as defined earlier) and it is a year of recession. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with 1% Cutoff Pure Pure Vertical Horizontal Mixed TarSize AcgSize 0.046 0.121 0.003 (0.00) (0.03) (0.81) AnyStock -0.05 -0.024 -0.049 (0.00) (0.58) (0.01) Acq_Assets 0 0 0 (0.36) (0.87) (0.35) tar_rndsales -0.002 0.05 0.002 (0.51) (0.9) (0.06) acq-rndsales -0.038 -0.04 0.004 (0.08) (0.94) (0.00) High RND dummy 0.04 -0.126 -0.001 (0.11) (0.59) (0.97) Highrnd*Recession dummy 0.003 0.053 -0.027 (0.96) (0.87) (0.58) Constant 0.022 -0.057 0.047 (0.45) (0.79) (0.65) Observations 305 129 247 [R.sup.2] 0.2 0.19 0.41 Year effects Yes Yes Yes Panel A. Panel B. FF Industries Four-Digit with a 1% Cutoff SIC with 1% Cutoff Pure Pure Conglomerate Vertical Horizontal TarSize AcgSize 0.036 0.038 0.052 (0.00) (0.02) (0.00) AnyStock -0.027 -0.02 -0.016 (0.00) (0.17) (0.34) Acq_Assets 0 0 0 (0.46) (0.6) (0.93) tar_rndsales -0.007 -0.032 0.014 (0.63) (0.00) (0.89) acq-rndsales -0.002 -0.088 -0.002 (0.00) (0.58) (0.05) High RND dummy 0.008 0.049 0.005 (0.77) (0.45) (0.95) Highrnd*Recession dummy -0.163 0.206 -0.05 (0.00) (0.02) (0.69) Constant -0.003 0.051 -0.022 (0.94) (0.5) (0.82) Observations 630 136 330 [R.sup.2] 0.16 0.3 0.14 Year effects Yes Yes Yes Panel B. FF Industries with a 1% Cutoff Mixed Conglomerati TarSize AcgSize 0.019 0.039 (0.06) (0.00) AnyStock -0.053 -0.028 (0.00) (0.00) Acq_Assets 0 0 (0.84) (0.23) tar_rndsales 0.002 -0.008 (0.06) (0.58) acq-rndsales 0.004 0.009 (0.00) (0.72) High RND dummy 0.003 -0.006 (0.85) (0.86) Highrnd*Recession dummy -0.051 -0.236 (0.19) (0.00) Constant 0.025 -0.004 (0.79) (0.93) Observations 416 416 [R.sup.2] 0.35 0.18 Year effects Yes Yes
We thank Bill Christie (Editor) and an anonymous referee .for many very useful comments on two revisions of this manuscript. We also thank seminar participants at NYU, Binghamton Universiity and INSEAD for many usefid comments, as well as participants at an FMA session 2010 and, in particular, the discussant, Josh Spizman. The usual disclaimer applies. Kedia and Ravid thank the Whitcomb Center at Rutgers Business School. Ravid also thanks the Sanger Foundation for financial support.
References
Acemoglu, D., S. Johnson, and T. Mitton, 2009, "Determinants of Vertical Integration: Financial Development and Contracting Costs," Journal of Finance 63, 1251-1290.
Akhigbe, A., S.E Borde, and A.M. Whyte, 2000, "The Source of Gains to Targets and Their Industry Rivals: Evidence Based on Terminated Merger Proposals," Financial Management 29, 101-118.
Alexandridis, D., D. Petmezas, and N.G. Travlos, 2010, "Gains from Mergers and Acquisitions around the World: New Evidence," Financial Management 39, 1671-1695.
Anderson, E. and D.C. Schmittlein, 1984, "Integration of the Sales Force: An Empirical Examination," Rand Journal of Economics 15,385-395.
Andrade, G., M. Mitchell, and E. Stafford, 2001, "New Evidence and Perspectives on Mergers," Journal of Economic Perspectives 15, 102-120.
Baker, G., R. Gibbons, and K.J. Murphy, 2002, "Rational Contracts and the Theory of the Firm," Quarterly Journal of Economics 117, 39-84.
Bens, D., P. Berger, and S. Monahan, 2011, "Discretionary Disclosure in Financial Reporting: An Examination Comparing Internal Firm Data to Externally Reported Segment Data," Accounting Review 86, 417-450.
Berger, P. and E. Ofek, 1995, "A Diversification Effect on Firm Value," Journal of Financial Economies 37, 39-65.
Bradley, M., A. Desai, and E. Kim, 1988, "Synergistic Gains from Corporate Acquisitions and Their Division between the Stockholders of Target and Acquiring Firms," Journal of Financial Economics 21, 3-40.
Campa, J.M. and S. Kedia, 2002, "Explaining the Diversification Discount," Journal of Finance 57, 1731-1762.
Carlton, D., 1977, "Uncertainty, Production Lags, and Pricing," American Economic Review 67, 244-249.
Caves, R.E. and R.M. Bradburd, 1988, "The Empirical Determinants of Vertical Integration," Journal of Economic Behavior and Organization 9, 265-279.
Chen, Y. and M.H. Riordan, 2007, "Vertical Integration, Exclusive Dealing and Ex-Post Cartelization," Rand Journal of Economics 38, 1-21.
Davis, N., 2008, "Chemical Industry Mergers and Acquisitions: A Sign of Things to Come," ICIS.com, May 9.
DeFontenay, C. and J. Gans, 2005, "Vertical Integration in the Presence of Upstream Competition," Rand Journal of Economics 36, 544-572.
Eckbo, E., 1983, "Horizontal Mergers, Collusion and Stockholder Wealth," Journal of Financial Economics 11, 241-273.
Eckbo, E., 1985, "Mergers and the Market Power Doctrine: Evidence from the Capital Markets," Journal of Business 58, 325-349.
Fama, E. and K. French, 1988, "Permanent and Temporary Components of Stock Price," Journal of Political Economy 96, 246-273.
Fan, J. and V. Goyal, 2006, "On the Patterns and Wealth Effects of Vertical Mergers," Journal of Business 79, 877-902.
Fee, E. and S. Thomas, 2004, "Sources of Gains in Horizontal Mergers: Evidence from Customer, Supplier and Rival Firms," Journal of Financial Economics 73,423-460.
Fishman, M., 1988, "A Theory of Preemptive Takeover Bidding," Rand Journal of Economics 19, 88-101.
Grossman, S.J. and O. Hart, 1986, "The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration," Journal of Political Economy 94, 691-719.
Gugler, K. and R. Siebert, 2007, "Market Power vs. Efficiency Effects of Mergers and Joint Research Ventures: Evidence from the Semiconductor Industry," Review of Economics and Statistics 89, 645-659.
Hart, O., J. Tirole, D. Carlton, and O. Williamson, 1990, "Vertical Integration and Market Foreclosure," Brookings Papers on Economic Activity. 205-286.
Hoberg, G. and G. Phillips, 2010, "Product Market Synergies and Competition in Mergers and Acquisitions: A Text-Based Analysis," Review of Financial Studies 23, 3773-381l.
Hughes, J.S. and J.L. Kao, 2001, "Vertical Integration and Proprietary Information Transfers," Journal of Economics and Management Strategy 10, 277-299.
Joskow, P 1985, "Vertical Integration and Long-Term Contracts: The Case of Coal-Burning Electric Generating Plants," Journal of Law Economics and Organization l, 33-80.
Joskow, P, 1988, "Asset Specificity and the Structure of Vertical Relationships: The Empirical Evidence," Journal of Law, Economics, and Organization 4, 95-117.
Kedia, S., V. Panchapagesan, and V. Uysal, 2008, "Geography and Acquirer Returns," Journal of Financial Intermediation 17, 256-275.
Kehoe, M.R., 1996, "Franchising Agency Problems and the Cost of Capital," Applied Economics 28, 1485-1493.
Klein, B., R.G. Crawford, and A.A. Alchian, 1978, "Vertical Integration, Appropriable Rents, and the Competitive Contracting Process," Journal of Law and Economics 21, 297-336.
Klein, B. and K.M. Murphy, 1997, "Vertical Integration as a Self Enforcing Contractual Arrangement," American Economic Review 87, 415-420.
Kohers, N. and T. Kohers, 2001, "Takeovers of Technology Firms: Expectations vs. Reality," Financial Management 30, 35-54.
LaFontaine, E and M. Slade, 2007, "Vertical Integration and Firm Boundaries: The Evidence," Journal of Economic Literature 135, 629-685.
Lang, L.H.E and R.M. Stulz, 1994, "Tobin's Q, Corporate Diversification, and Firm Performance," Journal of Political Economy 102, 1248-1280.
Levy, D.T., 1985, "The Transactions Cost Approach to Vertical Integration: An Empirical Examination," Review of Economics and Statistics 67, 438-445.
Lieberman, M.A., 1991, "Determinants of Vertical Integration: An Empirical Test," Journal of Industrial Economics 39, 451-466.
Maksimovic, V. and G. Phillips, 2002, "Do Conglomerate Firms Resources Inefficiently?" Journal of Finance 57, 721-767.
Masten, S.E., J.W. Meehan, Jr., and E.A. Snyder, 1989, "Vertical Integration in the US Auto Industry: A Note on the Influence of Transaction Specific Assets," Journal of Economic Behavior and Organization 12, 265-273.
Moeller, S.B., F.P. Schlingemann, and R.M. Stulz, 2005, "Wealth Destruction on a Massive Scale? A Study of Acquiring Firm Returns in the Recent Merger Wave," Journal of Finance 60, 757-782.
Ohanian, N.K., 1994, "Vertical Integration in the US Pulp and Paper Industry, 1900-1940," Review of Economics and Statistics 76, 202-207.
Ordover, J.A., G. Saloner, and S.C. Salop, 1990, "Equilibrium Vertical Foreclosure," American Economic Review 80, 127-142.
Perry, M.K., 1978, "Vertical Integration: The Monopsony Case," American Economic Review 68, 561-570.
Perry, M.K., 1989, "Vertical Integration: Determinants and Effects," in R. Schmallensee and R. Willig, Eds., Handbook of Industrial Organization, North Holland, ND, 185-250.
Qiu, L. and W. Zhou, 2007, "Merger Waves: A Model of Endogenous Mergers," Rand Journal of Economics 38, 214-226.
Rajah, R., H. Servaes, and L. Zingales, 2000, "The Cost of Diversity: The Diversification Discount and Inefficient Investment," Journal of Finance 55, 35-80.
Ravid, S.A. and M. Spiegel, 1999, "On Toeholds and Bidding Contests," Journal of Banking and Finance 23, 1219-1242.
Rhodes Kropf, M. and S. Viswanathan, 2004, "Market Valuation and Merger Waves," Journal of Finance 59, 2685-2718.
Riordan, M, 1998, "Anticompetitive Vertical Integration by a Dominant Firm," American Economic Review 88, 1232-1248.
Salinger. M., 1988, "Vertical Mergers and Market Foreclosure," Quarterly Journal of Economics 103, 345-356.
Schoar, A.A., 2002, "The Effect of Corporate Firm Diversification on Firm Productivity," Journal of Finance 62, 2379-2403.
Shahrur, H., 2005, "Industry Structure and Horizontal Takeovers: Analysis of Wealth Effects on Rivals, Suppliers and Corporate Customers," Journal of Financial Economics 76, 61-98.
Shenoy, L, 2008, "An Examination of the Efficiency, Foreclosure, and Collusion Rationales for Vertical Takeovers," Tulane University Working Paper.
Song, M. and R. Walkling, 2000, "Abnormal Returns to Rivals of Acquisition Targets: A Test of the Acquisition Probability Hypothesis," Jouf7Tal of Financial Economics 55, 143-171.
Spiller, P., 1985, "On Vertical Mergers," Journal of Law Economics and Organization 1, 285-312.
Stigler, G., 1950, "Monopoly and Oligopoly by Merger," American Economic Review 40, 23-34.
Villalonga, B., 2004, "Diversification Discount or Premium? New Evidence from the Business Information Tracking Series," Journal of Finance 59, 479-506.
Williamson, O., 1971, "The Vertical Integration of Production: Market Failure Considerations," American Economic Review 61, 112-123.
Williamson, O., 1983, "Credible Commitments Using Hostages to Support Exchange," American Economic Review 73, 519-540.
(1) These positive synergies are in contrast to the sizable literature demonstrating that conglomerate mergers destroy value. Berger and Ofek (1995) and kang and Stulz (1994) were the first to find that conglomerates trade at a discount. There are several papers that examine the source of this discount including Rajah, Servaes, and Zingales (2000) and Schoar (2002). Further work by Maksimovic and Phillips (2002), Campa and Kedia (2002), and Villalonga (2004) argues that the observed diversification discount does not imply that diversification in itself destroys value.
(2) Results are similar if we use other cutoffs in the mid to late 1990s, as in Fan and Goyal (2006). The idea of merger waves with different characteristics is explored in Rhodes Kroft and Viswanathan (2004) and Qiu and Zhou (2007), among others.
(3) There is a variety of other measures used for asset specificity in industrial organization studies (Lafontaine and Slade, 2007). However, the vast majority &these studies focus on one industry and measures could be more industry specific.
(4) Similarly, in a wide-ranging study covering vertical integration in 93 countries, Acemoglu, Johnson, and Mitton (2009) focus on the few variables that are general and vary from country to country.
(5) Along with being a proxy for site specificity, geographical distance has been used by several papers to proxy for monitoring cost and information advantages. Kedia, Panchapagesan, and Uysal (2008) find that geographically proximate deals have higher returns relative to distant deals due to information advantages.
(6) These are similar cutoffs to Fan and Goyal (2006). We should also remember that we are measuring input divided by output. As such, even a small percentage may indicate a much higher degree of vertical integration.
(7) In other words, a merger may be classified according to Fama and French (1988) industries and a 1% cutoff, Fama and French and a 5% cutoff, or Fama and French and a 10% cutoff, then, in addition, the four-digit SIC 1% cutoff, the four-digit 5% cutoff, etc. Some of these classifications are nested, but others may yield different sets of firms in each bucket.
(8) Although moving from a 1% to a 5% cutoff changes the deals classified as vertical, it does not appear to affect firm characteristics of the vertical sample. For example, for the 1% sample, the average acquirer has assets of $6,206 (in millions), whereas for the firms in the 5% cutoff, the average assets are similar at $5,986 (in millions).
(9) See Moeller et al. (2005) and Andrade, Mitchell, and Stafford (2001).
(10) We also estimated a (-5, +5) window. Results are similar, however. Given the many different classifications we already present, we do not include these tables. They are available upon request.
(11) The 5th and 95th percentile returns for acquirers in vertical deals are -0.203 and +0.108, respectively. This is in line with the distribution of returns in nonvertical deals that are also negatively skewed. In particular, for nonvertical deals, the 5th and 95th percentile returns are 0.225 and +0.134.
(12) From Panel A of Table II, it appears that the sharp decline in merger performance in 1998 is seen for all groups of mergers except conglomerate mergers. In particular, a t-test to examine whether the average return to conglomerates from 1979 to 1997 is different from the average return from 1998 to 2002 cannot be rejected. It is interesting that conglomerates, at least as defined by the 1% cutoff, do not appear to reflect the broader trend of declining performance documented by Moeller et al. (2005). However, a study of this is not the focus of this paper.
(13) Although in other countries the gains may be more evenly distributed (Alexandridis, Petmezas, and Travlos, 2010). The authors ascribe the discrepancy to different levels of competition in the market for corporate control.
(14) It is not entirely clear why this is the case. Moeller et al. (2005) do not find an industry explanation for their large loss deals. However, trade publications in the chemical industry suggest that the industry has been "transformed" by acquisitions over the last two decades but that "acquisitions made at high multiples at or near the top of the business cycle, however, in some instances destroyed rather than enhanced value. Companies such as France's Rhodia and Switzerland's Clariant were forced to exit the fine chemicals market" (Davis, 2008).
(15) We also estimate a specification where we use log assets. The results are qualitatively similar. We discuss some implications of this specification later.
(16) For a specific analysis &technology deals, see Kohers and Kohers (2001).
(17) When we use the log asset specification, the result for the Fama-French classification still holds, but the level of significance drops to 12% for the four-digit SIC classification.
(18) Other factors, such as the number of competing bids, are likely to be important in determining the share of the target in the overall returns. However, these variables are not likely to be correlated with the type of merger.
(19) We also observe a negative and significant coefficient for the acquirer Herfindahl index, but only in one classification. We are not sure how to interpret this coefficient.
(20) When we use the log asset specification results become more significant.
(21) For related theories, see, for example, Fishman (1988) and Ravid and Spiegel (1999). See also Shenoy (2008) for impact on rivals in vertical takeovers.
(22) If we use the log assets specification, the results are significant only at the 11% and 14% levels, respectively, with very similar coefficients.
(23) With log assets specification the results for Fama-French classification continue to hold but for the four-digit SIC classification the significance level drops to 13%.
(24) These results are not reported in the paper for brevity but are available from the authors upon request. The two cutoffs we used were: 1) high R&D is any firm that reports an R&D expense and 2) high R&D is any firm in the top 25% of R&D in the sample. Both these are milder criteria when defining R&D and include many more firms as R&D intensive. We find no evidence that high R&D as captured by the above criteria is associated with higher total returns. The evidence suggests that only when R&D is very high, as captured by our initial proxy of the top 10% of the sample, is there any evidence that vertical mergers are associated with higher returns. These tables are available upon request.
(25) For a different and new perspective on these issues, see Hoberg and Phillips (2010).
(26) We thank an anonymous referee for suggesting this testing.
(27) We thank an anonymous referee for suggesting this interpretation.
Simi Kedia, S. Abraham Ravid, and Vicente Pons *
* Simi Kedia is an Associate Professor of Finance and Economics at Rutgers Business School. Newark and New Brunswick, NJ. S. Abraham Ravid is a Professor of Finance and Economics at Rutgers Business School Newark and New Brunswick, NJ. Vicente Pons is affiliated with Exotix Limited in London. UK.
Table I. Panel A. Time Trends in Vertical Mergers, Four-Digit SIC Classification The table displays the fraction of deals classified as vertical over the time period under examination. Panel A uses the target and acquirer four-digit SIC from CRSP, whereas Panel B uses the Fama-French (1988) industry classification as the basis for classification. Transactions are classified as vertically related if the acquirer and target are vertically related and as pure vertical if they are vertically related, but not horizontally related. The cutoff refers to the percentage of vertical relatedness used in the table. Panel A. Industry is hour-Digit SIC Year Num Pure Vertical Pure Vertical Vertical Related Vertical Related Cutoffs 10% 5% 1979 5 0.000 0.000 0.000 0.000 1980 8 0.250 0.250 0.250 0.250 1981 50 0.000 0.060 0.000 0.080 1982 34 0.029 0.059 0.059 0.147 1983 34 0.059 0.059 0.088 0.118 1984 81 0.049 0.049 0.062 0.099 1985 85 0.082 0.094 0.118 0.129 1986 97 0.022 0.041 0.062 0.103 1987 72 0.056 0.153 0.056 0.181 1988 68 0.074 0.103 0.118 0.191 1989 50 0.120 0.220 0.140 0.240 1990 51 0.098 0.137 0.118 0.157 1991 42 0.048 0.214 0.143 0.333 1992 26 0.039 0.039 0.231 0.269 1993 36 0.083 0.083 0.278 0.389 1994 57 0.035 0.105 0.123 0.298 1995 88 0.057 0.125 0.148 0.318 1996 109 0.018 0.055 0.119 0.248 1997 122 0.016 0.066 0.123 0.303 1998 162 0.043 0.074 0.161 0.309 1999 131 0.099 0.137 0.214 0.428 2000 128 0.031 0.109 0.109 0.320 2001 107 0.065 0.103 0.168 0.374 2002 53 0.057 0.094 0.208 0.396 Total 1,696 0.0525 0.0973 0.1297 0.2606 1979-1997 1,115 0.0597 0.1006 0.1177 0.2028 1998-2002 581 0.059 0.1034 0.172 0.3654 Panel A. Industry is hour-Digit SIC Year Pure Vertical Vertical Related Cutoffs 1% 1979 0.200 0.200 1980 0.375 0.375 1981 0.060 0.180 1982 0.059 0.265 1983 0.147 0.206 1984 0.162 0.247 1985 0.224 0.247 1986 0.124 0.227 1987 0.181 0.375 1988 0.250 0.338 1989 0.180 0.340 1990 0.255 0.353 1991 0.262 0.571 1992 0.385 0.462 1993 0.278 0.417 1994 0.211 0.421 1995 0.182 0.375 1996 0.229 0.395 1997 0.279 0.484 1998 0.259 0.432 1999 0.298 0.534 2000 0.148 0.391 2001 0.224 0.486 2002 0.264 0.509 Total 0.2158 0.3868 1979-1997 0.2128 0.3409 1998-2002 0.2386 0.4704 Panel B. Time Trends in Vertical Mergers/Fama-French Classification Year Num Pure Vertical Pure Vertical Vertical Related Vertical Related Cutoffs 5% 1% 1979 5 0.000 0.000 0.000 0.200 1980 8 0.250 0.250 0.375 0.375 1981 50 0.000 0.080 0.060 0.180 1982 34 0.000 0.121 0.000 0.242 1983 34 0.000 0.121 0.000 0.212 1984 81 0.025 0.099 0.062 0.247 1985 85 0.059 0.131 0.167 0.250 1986 97 0.052 0.104 0.083 0.229 1987 72 0.000 0.183 0.070 0.380 1988 68 0.045 0.194 0.149 0.343 1989 50 0.020 0.240 0.060 0.340 1990 51 0.02 0.157 0.118 0.353 1991 42 0.000 0.333 0.048 0.571 1992 26 0.039 0.269 0.077 0.462 1993 36 0.139 0.389 0.139 0.417 1994 57 0.054 0.306 0.125 0.429 1995 88 0.058 0.322 0.069 0.379 1996 109 0.046 0.248 0.092 0.395 1997 122 0.066 0.306 0.174 0.488 1998 162 0.044 0.315 0.076 0.440 1999 131 0.070 0.434 0.124 0.543 2000 128 0.064 0.325 0.095 0.397 2001 107 0.056 0.374 0.103 0.486 2002 53 0.076 0.396 0.094 0.509 Total 1,696 0.0476 0.2625 0.0988 0.3899 1979-1997 1,115 0.045947 0.202789 0.098316 0.341684 1998-2002 581 0.062 0.3688 0.0984 0.475 Table II. Returns and Type of Merger Transactions This table reports average returns measured over the [-1,1] window. The classification is based on the four-digit SIC from CRSP with a 1 % (5%) cutoff for vertical classification in Panel A (B). Pure horizontal (vertical) deals are those where the transaction is only a horizontal (vertical) transaction. Mixed horizontal vertical deals are those deals where there is both a vertical and a horizontal transaction. Conglomerate transactions are those that are neither horizontal nor vertical. Panel C displays the average target size in millions of dollars and the acquirer book-to-market for vertical and nonvertical deals for the two time periods. The t-statistic represents the difference in the average value for vertical and nonvertical deals. Total Pure Pure Mixed Conglomerate Returns Vertical Horizontal Horizontal (4) (1) (2) Vertical (3) Panel A. 1 % Cutoff Total returns 1979-1997 0.023 *** 0.061 *** 0.034 *** 0.021 *** 1998-2002 -0.011 0.017 ** -0.02 ** 0.013 *** All 0.011 ** 0.047 *** 0.009 0.019 *** Acquirer returns 1979-1997 -0.002 0.036 * 0.011 -0.002 *** 1998-2002 -0.038 *** -0.019 -0.056 *** -0.005 All -0.016 *** 0.019 -0.019 ** -0.007 ** Target returns 1979-1997 0.187 *** 0.207 *** 0.164 *** 0.187 *** 1998-2002 0.24 *** 0.21 *** 0.24 *** 0.26 *** All 0.207 *** 0.209*** 0.198 *** 0.210 *** Panel B. 5% Cutoff Total returns 1979-1997 0.024 *** 0.057 *** 0.028 ** 0.022 *** 1998-2002 -0.004 0.01 ** -0.02 ** -0.008 All 0.012 ** 0.043 *** 0.003 0.018 *** Acquirer returns 1979-1997 -0.006 0.034 ** 0.003 -0.006 *** 1998-2002 -0.028 *** -0.027** -0.058 *** -0.013 All -0.016 *** 0.016 -0.028 *** -0.008 *** Target returns 1979-1997 0.18 *** 0.190 *** 0.170 *** 0.190 *** 1998-2002 0.26 *** 0.21 *** 0.25 *** 0.25 *** All 0.219 *** 0.194 *** 0.211 *** 0.207 *** Total t-test for t-test for t-test for Returns Diff in Diff in Diff in 1 and 4 2 and 4 3 and 4 Total returns 1979-1997 0.44 3.63 *** 1.65 1998-2002 -2.28*** 0.21 -2.92 *** All -1.67* 3.13 *** -1.26 Acquirer returns 1979-1997 1.92* 2.11 ** 1.86 * 1998-2002 -2.37** -0.66 -3.41 ** All -1.65* 2.01 ** -1.81 * Target returns 1979-1997 0.02 0.83 -1.17 1998-2002 -0.69 -1.08 -0.69 All -0.17 -0.04 -0.63 Total returns 1979-1997 0.37 3.99 *** 0.72 1998-2002 1.04 0.13 -2.48 *** All -1.0 3.26 *** -2.21 ** Acquirer returns 1979-1997 0.11 4.66 ** 1.13 1998-2002 -0.95 -0.77 -2.89 *** All -1.19 2.88 *** -2.62 *** Target returns 1979-1997 0.38 0.05 -0.62 1998-2002 0.44 -1.15 -0.10 All 0.59 -0.74 0.20 Years Before 1998 Years [greater than or equal to] 1998 Panel C. Characteristics of b'ertical and Nonvertical Deals Target size Vertical deals 441.8 854.5 Nonvertical deals 295.6 1096.9 t-statistic 1.64 0.48 Acquirer book-market Vertical deals 0.47 0.337 Nonvertical deals 0.54 0.43 t-statistic 2.02 ** 1.49 * Significant at the 0.10 level. ** Significant at the 0.05 level. *** Significant at the 0.01 level. Table III. OLS Regression of Total, Acquirer, and Target Returns The classification of deals is based on the four-digit SIC from CRSP with a 1 % cutoff for vertical classification. The pure horizontal (vertical) dummy takes a value one if the transaction is only a horizontal (vertical) transaction. The coefficient of Acq_assests has been multiplied by 106. The horizontal vertical dummy takes a value one if the deal is both a vertical and a horizontal transaction. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy if stock is used for payment in the transaction, and Acq_Assets is the total assets of the acquirer in the year prior to the announcement. The dummies indicate whether a deal is; nure horizontal_ nure vertical_ or mixed horizontal and vertical. The n-values are in parentheses. Panel A. Total Returns TarSize AcgSize 0.029 0.033 0.012 (0.000) (0.00) (0.25) AnyStock -0.031 -0.016 -0.05 (0.000) (0.03) (0.00) Acq_Assets -0.09 -0.196 -0.079 (0.52) (0.42) (0.63) Pure horizontal dummy 0.031 0.043 0.008 (0.001) (0.00) (0.62) Pure vertical dummy -0.006 0.002 -0.022 (0.35) (0.78) (0.04) Horizontal vertical dummy -0.008 0.003 -0.021 (0.30) (0.79) (0.06) Constant 0.036 0.019 0.034 (0.001) (0.12) (0.02) Year dummies Yes Yes Yes Years included All obs < 1998 [greater than or equal to] 1998 Num of observations 1,579 1,013 566 [R.sup.2] 0.09 0.1 0.07 Panel B. Acquirer Returns TarSize AcgSize 0.02 0.022 0.016 (0.00) (0.00) (0.13) AnyStock -0.029 -0.01 -0.056 (0.00) (0.2) (0.00) Acq_Assets 0.086 0.022 0.072 -0.544 (0.93) (0.68) Pure horizontal dummy 0.031 0.046 -0.003 (0.001) (0.00) (0.85) Pure vertical dummy -0.006 0.005 -0.026 (0.38) (0.52) (0.02) Horizontal vertical dummy -0.009 0.009 -0.032 (0.220 (0.39) (0.01) Constant -0.005 -0.013 0.02 (0.92) (0.27) (0.21) Year dummies Yes Yes Yes Years included All obs < 1998 [greater than or equal to] 1998 Num of observations 1,580 1,013 566 [R.sup.2] 0.073 0.07 0.09 Panel C. Target Returns TarSize AcgSize -0.054 -0.034 -0.151 (0.000) (0.00) (0.00) AnyStock -0.106 -0.058 -0.166 (0.000) (0.00) (0.00) Acq_Assets 0.66 0.78 0.501 (0.060) (0.16) (0.31) Pure horizontal dummy 0.013 0.027 -0.013 (0.56) (0.29) (0.78) Pure vertical dummy -0.014 -0.016 -0.007 (0.42) (0.39) (0.81) Horizontal vertical dummy -0.006 -0.029 0.031 (0.76) (0.21) (0.34) Constant 0.191 0.22 0.421 (0.097) (0.00) (0.00) Year dummies Yes Yes Yes Years included All obs < 1998 [greater than or equal to] 1998 Num of observations 1,579 1,013 566 [R.sup.2] 0.091 0.07 0.14 Table IV. Total Returns and Target and Acquirer Market Shares The dependent variable is total returns measured over the [-1,1] window. Panel A uses the four-digit SIC with a 1% cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a I% cutoff for classification. The coefficient of Acq_Assets has been multiplied by 106. Separate results are reported for subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are classified as mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value one if there is any stock payment for the transaction. Acq_Assets is the size of the acquirer. Target (acquirer) market share is the ratio of target (acquirer) sales to industry sales (the same four-digit SIC as the target [acquirer]). The high share dummy takes a value of one if both the target and the acquirer have high market shares. Target (acquirer) is defined as having high market share if it is in the top 10%. The top 10% includes targets with greater then 10% of the market share and acquirers with greater than 40% of the market share. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Vertical Pure Horizontal Mixed Tarsize AcgSize 0.048 0.118 0.012 (0.00) (0.04) (0.45) AnyStock -0.051 -0.029 -0.038 (0.00) (0.49) (0.07) Acq_Assets -0.212 -0.808 0.92 (0.36) (0.84) (0.56) Target market share -0.013 -0.177 0.042 (0.77) (0.55) (0.77) Acquirer market share -0.018 -0.041 -0.017 (0.55) (0.83) (0.85) High share dummy 0.065 0.143 0.047 (0.089) (0.57) (0.71) Constant 0.021 -0.052 0.038 (0.48) (0.81) (0.76) Observations 305 129 247 [R.sup.2] 0.197 0.19 0.13 Year dummies Yes Yes Yes Panel A. Four-Digit SIC with a Panel B. FF Industries with 1% Cutoff a 1% Cutoff Conglomerate Pure Vertical Pure Horizontal Tarsize AcgSize 0.032 0.041 0.052 (0.00) (0.017) (0.00) AnyStock -0.028 -0.022 -0.022 (0.00) (0.16) (0.2) Acq_Assets -0.09 -0.124 0.115 (0.50) (0.63) (0.73) Target market share 0.025 -0.033 -0.106 (0.3) (0.6) (0.31) Acquirer market share -0.007 -0.041 -0.033 (0.67) (0.29) (0.54) High share dummy 0.023 0.084 0.04 (0.25) (0.066) (0.58) Constant -0.004 0.05 0.047 (0.93) (0.52) (0.68) Observations 630 136 330 [R.sup.2] 0.131 0.211 0.14 Year dummies Yes Yes Yes Panel B. FF Industries with a 1% Cutoff Mixed Conglomerate Tarsize AcgSize 0.025 0.032 (0.03) (0.00) AnyStock -0.048 -0.029 (0.00) (0.00) Acq_Assets 0.049 -0.26 (0.95) (0.19) Target market share 0.032 0.035 (0.67) (0.20) Acquirer market share -0.007 -0.004 (0.9) (0.83) High share dummy 0.049 0.034 (0.53) (0.'13) Constant 0.024 -0.005 (0.82) (0.92) Observations 416 416 [R.sup.2] 0.13 0.162 Year dummies Yes Yes Table V. Acquirer and Target Returns and Target and Acquirer Market Shares-Only Important Coefficients Are Reported The dependent variable is total returns measured over the [-1,1 ] window. Panel A uses the four-digit SIC with a 1 % cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a 1% cutoff for classification. Separate results are reported for subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. The table reports partial results. Variables that were included in the regression but have not been reported are 1) Tarsize Acqsize: the ratio of target size to acquire size, 2) AnyStock: a dummy that is equal to one if there is any stock payment for the transaction, and 3) Acq_Assets: the size of the acquirer. Target (acquirer) market share is the ratio of target (acquirer) sales to industry sales (the same four-digit SIC as the target [acquirer]). The high share dummy takes a value of one if both the target and the acquirer have high market shares. Target (acquirer) is defined as having high market share if it is in the top 10%. The top 10% includes targets with greater then 10% of the market share and acquirers with greater than 40% of the market share. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Vertical Pure Horizontal Mixed Acquirer returns Target market share -0.02 -0.357 -0.079 (0.66) (0.25) (0.61) Acquirer market share 0.011 0.01 -0.048 (0.71) (0.96) (0.63) High share dummy 0.018 0.16 0.1 (0.65) (0.54) (0.46) [R.sup.2] 0.15 0.18 0.14 Target returns Target market share -0.139 0.215 -0.097 (0.33) (0.63) (0.77) Acquirer market share -0.065 0.256 -0.255 (0.49) (0.36) (0.23) High share dummy 0.269 -0.144 0.239 (0.028) (0.71) (0.41) [R.sup.2] 0.188 0.2 0.17 Observations 305 129 247 Year dummies Yes Yes Yes Panel A. Four-Digit Panel B. FF Industries SIC with a 1% Cutoff with a 1% Cutoff Pure Pure Conglomerate Vertical Horizontal Acquirer returns Target market share -0.014 -0.031 -0.152 (0.56) (0.65) (0.16) Acquirer market share 0.001 -0.011 -0.018 (0.92) (0.79) (0.75) High share dummy 0.018 0.034 0.053 (0.35) (0.47) (0.47) [R.sup.2] 0.079 0.222 0.13 Target returns Target market share 0.098 -0.231 0.09 (0.20) (0.23) (0.66) Acquirer market share 0.221 -0.103 0.284 (0.00) (0.37) (0.01) High share dummy -0.053 0.341 -0.138 (0.4) (0.014) (0.34) [R.sup.2] 0.153 0.256 0.19 Observations 630 136 330 Year dummies Yes Yes Yes Mixed Conglomerate Acquirer returns Target market share -0.026 -0.016 (0.74) (0.53) Acquirer market share 0.003 0.007 (0.96) (0.69) High share dummy 0.059 0.019 (0.48) (0.37) [R.sup.2] 0.13 0.112 Target returns Target market share -0.141 0.129 (0.45) (0.1) Acquirer market share -0.083 0.202 (0.56) (0.00) High share dummy 0.277 -0.021 (0.16) (0.75) [R.sup.2] 0.19 0.183 Observations 416 416 Year dummies Yes Yes Table VI. Total Returns and Industry Concentration The dependent variable is total returns measured over the [-1,1 ] window. Panel A uses the four-digit SIC with a 1 % cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a 1% cutoff for classification. The coefficient of Acq_Assets has been multiplied by 106. Separate results for are reported for subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value of one if there is any stock payment for the transaction. Acq_Assets is the size of the acquirer. Target (acquirer) Herfindahl is the sales Herfindahl index for the respective four-digit SIC. Target (acquirer) market share is the ratio of target (acquirer) sales to industry sales (the same four-digit SIC as the target [acquirer]). The high share in concentrated industry dummy takes a value of one if both the target and the acquirer have high market shares and both belong to concentrated industries. Target (acquirer) is defined as having high market share if it is in the top 10%. Target (acquirer) industry is defined as being concentrated if the Herfindahl index is in the top 25%. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Pure Vertical Horizontal Mixed Tarsize AcgSize 0.051 0.12 0.011 (0.00) (0.04) (0.52) AnyStock -0.052 -0.03 -0.04 (0.000) (0.47) (0.06) Acq_Assets -0.31 1.59 1.19 (0.2) (0.71) (0.47) Target Herfindahl 0.039 -0.021 -0.042 (0.21) (0.93) (0.67) Acquirer Herfindahl -0.067 0.306 0.078 (0.046) (0.22) (0.39) Target market share -0.04 -0.313 0.062 (0.40) (0.31) (0.69) Acquirer market share 0.018 -0.16 -0.063 (0.59) (0.41) (0.56) High share in concentrated 0.074 0.117 0.048 Industry dummy (0.055) (0.64) (0.7) Constant 0.075 -0.308 0.035 (0.013) (0.24) (0.78) Observations 305 129 247 [R.sup.2] 0.211 0.22 0.13 Year dummies Yes Yes Yes Panel A. Panel B. FF Industries Four-Digit with a 1% Cutoff SIC with a 1% Cutoff Pure Pure Conglomerate Vertical Horizontal Tarsize AcgSize 0.032 0.044 0.052 (0.00) (0.012) (0.00) AnyStock -0.029 -0.025 -0.023 (0.00) (0.12) (0.17) Acq_Assets -0.08 -0.17 0.157 (0.56) (0.51) (0.64) Target Herfindahl -0.024 0.013 -0.071 (0.25) (0.79) (0.26) Acquirer Herfindahl 0.034 -0.057 0.157 (0.17) (0.33) (0.03) Target market share 0.042 -0.04 -0.082 (0.12) (0.54) (0.45) Acquirer market share -0.024 -0.011 -0.103 (0.26) (0.82) (0.1) High share in concentrated 0.016 0.096 0.029 Industry dummy (0.43) (0.047) (0.69) Constant 0 0.054 0.048 (0.99) (0.50) -0.68 Observations 630 136 330 [R.sup.2] 0.134 0.218 0.15 Year dummies Yes Yes Yes Panel B. FF Industries with a 1% Cutoff Mixed Conglomerate Tarsize AcgSize 0.025 0.032 (0.03) (0.00) AnyStock -0.048 -0.03 (0.00) (0.001) Acq_Assets 0.055 -0.23 (0.95) (0.25) Target Herfindahl 0.04 0 (0.42) (0.99) Acquirer Herfindahl -0.02 0.032 (0.67) (0.27) Target market share 0.005 0.038 (0.95) (0.22) Acquirer market share 0.001 -0.02 (0.99) (0.42) High share in concentrated 0.043 0.028 Industry dummy (0.59) (0.222 Constant 0.02 -0.008 -0.85 (0.86) Observations 416 416 [R.sup.2] 0.13 0.163 Year dummies Yes Yes Table VII. Returns to Rival Firms in Acquirer and Target Industries This table reports summary statistics of the average value-weighted cumulative abnormal returns (expressed in percentages) earned by all other firms in the acquirer's four-digit SIC (Panel A) and all other firms in the target's four-digit SIC. The CARs are calculated over the [-1,1] around the merger announcement. The average values are reported for subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Mixed and horizontal combines pure horizontal and mixed transactions. The numbers in parentheses report the values of a t-test and a Wilcoxon sign-rank test for the difference between the mean and medians, respectively. Panel A. Rival Panel B. Rival Returns-Acquirer Returns-Target Industry (%) Industry (%) Mean Median Mean Median Pure vertical 0.054 -0.16 0.219 0.13 Pure horizontal 0.51 0.45 0.51 0.45 Mixed 0.31 0.26 0.31 0.26 Mixed and 0.38 0.36 0.38 0.36 horizontal Conglomerate -0.078 -0.015 -0.013 -0.02 Test for difference (0.52) (0.46) (0.92) (0.83) between vertical and conglomerate Test for difference (1.25) (2.27) ** (0.65) (0.59) between vertical, mixed and horizontal Number of Observations Pure vertical 295 Pure horizontal 127 Mixed 245 Mixed and 372 horizontal Conglomerate 589 Test for difference between vertical and conglomerate Test for difference between vertical, mixed and horizontal ** Significant at the 0.05 level. Table VIII. Returns to Rival Firms and Market Shares The dependent variable is the average value-weighted abnormal returns of all other firms in the same four-digit SIC as the acquirer and the target measured over the [-1,1] day window around the announcement of the deal. The coefficient of Acq_Assets has been multiplied by 106. Panel A displays the results for acquirer industry and Panel B for the target industry. Vertical deals were defined using the four-digit SIC with a 1% cutoff for classification. Separate results are reported for the subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Tarsize_Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value of one if there is any stock payment for the transaction. Acq_Assets is the size of the acquirer. Target (acquirer) market share is the ratio of target (acquirer) sales to industry sales (the same-four digit SIC as the target [acquirer]). The high share in concentrated industry dummy takes a value of one if both the target and the acquirer have high market shares and both belong to concentrated industries. Target (acquirer) is defined as having high market share if it is in the top 10%. The p-values are given in parentheses. Panel A. Acquirer Industry Pure Pure Mixed Conglomerate Vertical Horizontal TarSize AcgSize 0.618 -0.192 0.438 0.015 (0.24) (0.79) (0.32) (0.95) AnyStock -0.155 0.48 0.008 -0.176 (0.74) (0.37) (0.99) (0.58) Acq_Assets -1.39 -66.51 -7.47 1.26 (0.9) (0.19) (0.86) (0.85) Acquirer market share 2.271 -1.891 -1.325 -0.082 (0.12) (0.42) (0.59) (0.91) Target market share 3.838 -1.779 2.848 0.67 (0.07) (0.64) (0.46) (0.56) High share dummy -3.232 3.936 -2.79 0.123 (0.09) (0.22) (0.4) (0.9) Constant -0.307 0.108 -0.307 -0.693 (0.7) (0.9) (0.72) (0.25) Observations 295 127 245 589 [R.sup.2] 0.08 0.18 0.06 0.03 Year dummies Yes Yes Yes Yes Panel B. Target Industry Pure Pure Mixed Conglomerate Vertical Horizontal TarSize AcgSize 1.114 -0.176 0.437 -0.053 (0.02) (0.81) (0.33) (0.83) AnyStock -0.276 0.491 0.002 -0.802 (0.52) (0.36) (1) (0.02) Acq_Assets -7.29 -66.47 -7.39 0.289 (0.47) (0.19) (0.86) (0.96) Acquirer market share 1.257 -1.923 -1.323 0.086 (0.32) (0.41) (0.59) (0.91) Target market share 0.12 -1.758 2.84 -0.353 (0.96) (0.64) (0.46) (0.77) High share dummy -3.017 3.936 -2.757 0.604 (0.1) (0.22) (0.41) (0.57) Constant -0.405 0.094 -0.301 0.036 (0.56) (0.92) (0.73) (0.95) Observations 291 128 244 579 [R.sup.2] 0.08 0.18 0.06 0.05 Year dummies Yes Yes Yes Yes Table IX. Total Returns and Asset Specificity The dependent variable is the total returns measured over the [-1,1] window. Panel A uses the four-digit SIC with a 1% cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a 1% cutoff for classification. Separate results are reported for subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. The upper panel (R&D) reports the results of estimations that use R&D to capture asset specificity, whereas the bottom panel (Geographic proximity) uses geography to capture asset specificity. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value of one if there is any stock payment for the transaction. Acq_Assets is the size of the acquirer. Target (acquirer) R&D is the ratio of the targets (acquirer) R&D to sales. High R&D dummy takes a value one if both the target and the acquirer have high R&D. Target (acquirer) is defined as having high R&D if it is in the top 10%. The top 10% includes targets with R&D greater than 25% of sales and acquirers with R&D greater than 40% of sales. Same MSA dummy takes a value one when both the target and the acquirer are headquartered in the same MSA. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Pure Vertical Horizontal Mixed Conglomerate R&D Target R&D -0.002 0.055 0.002 (0.51) (0.89) (0.04) Acquirer R&D -0.038 -0.033 0.004 (0.083) (0.95) (0.00) High R&D dummy 0.04 -0.105 -0.006 (0.084) (0.58) (0.75) [R.sup.2] 0.203 0.19 0.41 Geographic proximity Same MSA dummy 0.006 0.104 0.021 (0.65) (0.05) (0.27) [R.sup.2] 0.19 0.21 0.13 Observations 305 129 247 Year dummies Yes Yes Yes Panel B. FF Industries with a 1% Cutoff Pure Pure Pure Vertical Vertical Horizontal Mixed R&D Target R&D -0.001 -0.035 0.02 (0.96) (0.003) (0.85) Acquirer R&D -0.002 -0.123 -0.002 (0.001) (0.44) (0.05) High R&D dummy -0.033 0.105 -0.014 (0.19) (0.085) (0.85) [R.sup.2] 0.147 0.263 0.14 Geographic proximity Same MSA dummy -0.006 0.011 0.034 (0.52) (0.64) (0.12) [R.sup.2] 0.13 0.19 0.14 Observations 630 136 330 Year dummies Yes Yes Yes Pure Vertical Conglomerate R&D Target R&D 0.002 -0.003 (0.02) (0.86) Acquirer R&D 0.004 -0.03 (0.00) (0.19) High R&D dummy -0.007 -0.042 (0.66) (0.25) [R.sup.2] 0.35 0.164 Geographic proximity Same MSA dummy 0.013 0.001 (0.34) (0.95) [R.sup.2] 0.13 0.15 Observations 416 416 Year dummies Yes Yes Table X. Acquirer and Target Returns and Asset Specificity-Only Important Coefficients Are Reported The dependent variable is total returns measured over the [-1,1 ] window. Panel A uses four-digit SIC codes, whereas Panel B uses Fama-French (1988) industries with a 1 % cutoff for classification. Separate results for are reported for different subgroup. The upper panel (R&D) reports the results of estimation that use R&D to capture asset specificity, whereas the bottom panel (Geographic proximity) uses geography to capture asset specificity. Variables that were included in the regression, but have not been reported are Tarsize Acgsize, Any stock, and Acq_Assets. Target (acquirer) R&D is the ratio of the target (acquirer) R&D to sales. High R&D dummy takes the value one if both the target and the acquirer have high R&D. Target (acquirer) is defined as having high R&D if it is in the top 10%. The top 10 percentile includes targets with R&D greater then 25% of sales and acquirers with R&D greater than 40% of sales. Same MSA dummy takes a value one when both the target and the acquirer are headquartered in the same MSA. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Horizontal Mixed Conglomerate Vertical R&D Acquirer returns Target R&D -0.001 0.141 0.003 -0.008 (0.71) (0.73) (0.01) (0.58) Acquirer R&D -0.032 -0.077 0.004 -0.001 (0.16) (0.89) (0.00) (0.002) High R&D dummy 0.014 -0.083 0.005 0.038 (0.56) (0.68) (0.82) (0.11) [R.sup.2] 0.155 0.17 0.41 0.096 Target returns Target R&D 0.009 -0.414 -0.001 0.042 (0.42) (0.47) (0.66) (0.36) Acquirer R&D -0.019 0.705 0.006 -0.004 (0.79) (0.4) (0.02) (0.030) High R&D dummy 0.057 -0.322 -0.006 -0.131 (0.45) (0.26) (0.91) (0.10) [R.sup.2] 0.176 0.21 0.21 0.131 Geographic proximity Acquirer returns Same MSA dummy 0.007 0.091 0.036 -0.018 (0.61) (0.1) (0.08) (0.04) [R.sup.2] 0.15 0.19 0.15 0.08 Target returns Same MSA -0.011 0.117 -0.048 -0.041 (0.8) (0.15) (27) (0.16) [R.sup.2] 0.17 0.21 0.17 0.12 Observations 305 129 247 630 Year dummies Yes Yes Yes Yes Panel B. FF Industries with a 1% Cutoff Vertical Horizontal Mixed Conglomerate R&D Acquirer returns Target R&D -0.03 0.031 0.003 -0.005 (0.015) (0.77) (0.00) (0.71) Acquirer R&D -0.072 -0.001 0.004 0.1 (0.67) (0.13) (0.00) (0.000) High R&D dummy 0.046 0.007 -0.001 -0.034 (0.47) (0.92) (0.94) (0.32) [R.sup.2] 0.261 0.13 0.35 0.161 Target returns Target R&D -0.036 0.15 -0.001 0.021 (0.33) (0.47) (0.83) (0.66) Acquirer R&D 0.774 -0.004 0.005 -0.049 (0.13) (0.0 (0.02) (0.48) High R&D dummy -0.346 -0.204 0.003 -0.12 (0.072) (0.15) (0.95) (28) [R.sup.2] 0.241 0.18 0.21 0.15 Geographic proximity Acquirer returns Same MSA dummy 2.633 0.031 0.024 -0.011 (0.27) (0.17) (0.09) (0.31) [R.sup.2] 0.23 0.13 0.13 0.11 Target returns Same MSA -0.069 0.012 -0.029 -0.018 (0.35) (0.79) (0.39) (0.61) [R.sup.2] 0.21 0.17 0.19 0.14 Observations 136 330 416 416 Year dummies Yes Yes Yes Yes Table XI. Total Returns and Information Asymmetry The dependent variable is total returns measured over the [-1,1] window. Panel A uses the four-digit SIC with a 1% cutoff for classification, whereas Panel B uses Fama-French (1988) industries and also a 1 % cutoff for classification. The coefficient of Acq_Assets has been multiplied by 106. Separate results for are reported for the subgroups: 1) pure vertical: transactions that are classified as only vertical, 2) pure horizontal, 3) mixed: transactions that are mixed vertical horizontal, and 4) conglomerate: transactions that are neither vertical nor horizontal. Tarsize Acgsize is the ratio of target size to acquire size. AnyStock is a dummy that takes a value of one if there is any stock payment for the transaction. Acq_Assets is the size of the acquirer. Target (acquirer) Infodum is a dummy that takes a value of one if the target (acquirer) has no analyst coverage. High Info Asy. dummy takes a value of one when both the target and the acquirer have no analyst coverage. All years for which data are available are included. The p-values are given in parentheses. Panel A. Four-Digit SIC with a 1% Cutoff Pure Pure Vertical Horizontal Mixed Conglomerate TarSize AcgSize 0.037 0.099 0.004 0.025 (0.000) (0.05) (0.8) (0.000) AnyStock -0.047 -0.027 -0.042 -0.026 (0.000) (0.47) (0.03) (0.000) Acq_Assets -0.218 1.17 0.034 -0.12 (0.33) (0.98) (0.43) (0.35) Target Infodum 0.005 -0.062 -0.066 -0.008 (0.69) (0.26) (0.02) (0.30) Acquirer Infodum 0.012 -0.034 -0.077 0 (0.33) (0.57) (0.01) (0.98) High Info Asy. dummy -0.014 0.029 0.081 0.01 (0.47) (0.69) (0.02) (0.43) Constant 0.021 0.079 0.123 0.007 (0.79) (0.7) (0.18) (0.85) Observations 350 154 267 808 [R.sup.2] 0.181 0.19 0.17 0.105 Year dummies Yes Yes Panel B. FF Industries with a 1% Cutoff Pure Pure Vertical Horizontal Mixed Conglomerate TarSize AcgSize 0.02 0.052 0.024 0.022 (0.096) (0.00) (0.02) (0.000) AnyStock -0.016 -0.019 -0.047 -0.029 (0.27) (0.2) (0.00) (0.000) Acq_Assets -0.118 0.16 0.019 -0.257 (0.62) (0.95) (0.83) (0.16) Target Infodum -0.003 -0.058 -0.028 -0.008 (0.88) (0.01) (0.11) (0.38) Acquirer Infodum 0.012 -0.049 -0.028 0.003 (0.53) (0.04) (0.13) (0.74) High Info Asy. dummy -0.002 0.056 0.03 -0.006 (0.95) (0.05) (0.18) (0.67) Constant 0.016 0.041 0.048 0.01 (0.79) (0.45) (0.65) (0.79) Observations 159 403 457 545 [R.sup.2] 0.169 0.14 0.14 0.124 Year dummies Yes Yes
![]() ![]() ![]() ![]() | |
Author: | Kedia, Simi; Ravid, S. Abraham; Pons, Vicente |
---|---|
Publication: | Financial Management |
Article Type: | Report |
Date: | Dec 22, 2011 |
Words: | 18931 |
Previous Article: | Financial distress in the Great Depression. |
Next Article: | Does the management's forecast of merger synergies explain the premium paid, the method of payment, and merger motives? |
Topics: |