838 4p

International Journal of Industrial Organization 21 (2003) 625–653

www.elsevier.com / locate / econbase

T he effects of mergers: an international comparison

*Klaus Gugler, Dennis C. Mueller , B. Burcin Yurtoglu, Christine Zulehner

Department of Economics, University of Vienna, BWZ, Bruennerstr. 72, A-1210 Vienna, Austria

Received 12 December 2001; received in revised form 5 July 2002; accepted 22 July 2002

Abstract

This paper analyzes the effects of mergers around the world over the past 15 years. We utilize a large panel of data on mergers to test several hypotheses about mergers. The effects of the mergers are examined by comparing the performance of the merging firms with control groups of nonmerging firms. The comparisons are made on profitability and sales. The results show that mergers on average do result in significant increases in profits, but reduce the sales of the merging firms. Interestingly, these post merger patterns look similar across countries. We also did not find dramatic differences between mergers in the manufacturing and the service sectors, and between domestic and cross-border mergers. Conglomerate mergers decrease sales more than horizontal mergers. By separating mergers into those that increase profits and those that reduce them and by then examining the patterns of sales changes following the mergers, we determine the effects of mergers on efficiency and market power. Our results suggest that those mergers that decrease profits and efficiency account for a large proportion. However, we can also identify mergers that increase profits by either increasing market power or by increasing efficiency. The first conclusion seems to be a more likely explanation for large companies, whereas the latter is likely to be true for small firms.  2002 Elsevier Science B.V. All rights reserved.

JEL classification: G34; L2

Keywords: Mergers; Acquisitions; International comparison

*Corresponding author. Tel.:1 43-1-4277-37484; fax:1 43-1-4277-37498. E-mail addresses: [email protected](K. Gugler), [email protected](D.C.

Mueller), [email protected](D.C. Mueller), [email protected](B.B. Yurtoglu), [email protected](C. Zulehner).

0167-7187 / 02 / $ – see front matter 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0167-7187(02)00107-8

626 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

1 . Introduction

The past century saw five great merger waves—one at its beginning, and successive waves at the ends of the 1920s, 1960s, 1980s and 1990s. While much of the earlier merger activity was confined to North America and Great Britain, the most recent wave has engulfed all of the major industrial countries of the world. And, as befits a global economy, it has been composed of an increasing percentage of cross-border acquisitions. What have been the causes of these great bursts of merger activity? What have been their effects? In this paper we focus largely on the second question, but the answers that we give to it will also shed light on the first. We confine our analysis to mergers taking place in the last two decades, but include in it mergers from around the world including also cross-border acquisi- tions.

1The hypotheses as to why mergers occur can be grouped into three broad categories. Of these, the first two presume that the managers of the merging companies seek to maximize profits or shareholder wealth. Under this assumption any merger must be expected to either increase the market power of the merging companies or reduce their costs. The third set of hypotheses includes those that posit other managerial goals than profits, as say the growth of the firm, or quasi-irrational behavior as might occur because managers are overcome by hubris.

From the point of view of the theory of the firm, it is important to determine whether mergers are best explained by one of the hypotheses from the third category, or by a hypothesis that presumes profit maximization. Ifall mergers are consistent with profit maximization, then corporate governance structures can be assumed to be designed in such a way as to align shareholder and managerial interests. If, on the other hand, a large fraction does not appear to increase shareholder wealth, corporate governance structures must be assumed to be deficient in bringing about such an alignment. We attempt to determine whether mergers increase market power or efficiency by examining their impacts on company sales and profits. In this way, we seek to determine to what extent mergers fall into each of these three categories.

The paper proceeds as follows. In the following section we present some predictions regarding the effects of mergers. Section 3 describes the methodology used to determine the effects of mergers. Our data base is described in Section 4. Sections 5 and 6 present our overall findings and those that are specific to the question of whether mergers increase efficiency or market power. In Section 7 we compare our findings with those previously reported in the literature. The sample is divided according to the mergers’ effects on market power and efficiency in Section 8. Conclusions are drawn in the final section.

1 We shall not distinguish between mergers and takeovers, but rather simply refer to all as mergers.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 627

2 . Predictions

We would like to derive predictions regarding the consequences of mergers from a model, which is applicable to all forms of mergers, to all industries and in all countries. Unfortunately, we know of nogenerally accepted model of firm behavior that meets these criteria. The most commonly used oligopoly model in industrial organization—the Cournot model with identical firms and homogeneous products—has the awkward feature of predicting that no horizontal mergers take

2place, even when they raise industry profits. Nevertheless, both economic theory and our economic intuition allow us to make some predictions about the effects of mergers on sales and profits.

The term ‘market power’ connotes the ability to control price. Any merger that increases a firm’s market power must, therefore, increase its ability to control (raise) the price of its products. If the firm faces a downward sloping demand schedule, and it takes advantage of its increase in market power by raising price, both its output and sales will fall, if it is maximizing profits, as it will then be

3operating in the elastic portion of its demand schedule. Of the three types of mergers, increases in market power are perhaps most likely

to follow horizontal mergers, but are also possible with vertical and conglomerate mergers. Conglomerate mergers can increase the degree ofmultimarket contact between the merging firms and their rivals. High multimarket contact raises the costs of cutting price in any given market and thus can facilitate more cooperative

4behavior thereby effectively increasing the merging firm’s market power. A vertical merger can also increase multimarket contact, if the merging

companies’ rivals are also vertically integrated. A vertical merger can also increase market power by raising entry barriers and thus effectively lowering the merging firm’s elasticity of demand. Thus, we hypothesize that an increase in market power is a possible consequence of all three types of mergers. For any merger that increases market power, we predict an increase in profits and a decline in sales.

A merger is said to have increased the efficiency of the merging firms, if it reduces their costs. A fall in marginal costs should cause a profit-maximizing

5company to lower its price and thus lead to an increase in both sales and profits. If mergers take place because managers pursue growth rather than profits, or out

2 See Salant et al. (1983). 3 We find the assumption that firms in our sample face negative sloped demand schedules reasonable

because: (1) they do not tend to operate in atomistically competitive industries, (2) many do sell differentiated products, and (3) even firms selling seemingly homogeneous products like cement differ from their rivals in terms of location, speed of delivery and similar factors that give them some control over price.

4 For a development and testing of the theory of multimarket contact, see Scott (1993). 5 A merger might increase efficiency without changing marginal costs by, say, reducing fixed costs.

In this case we should expect an increase in profits and no change in sales. We shall not test separately for this case.

628 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 1 Possible consequences of mergers

DP . 0 DP , 0

DS . 0 1 3 Efficiency increase Market power reduction (?)

DS , 0 2 4 Market power increase Efficiency decline

of managerial hubris, neither efficiency nor market power need change. We shall, however, hypothesize that these sorts of mergers lead to declines in efficiency because of the transaction costs of bringing two companies together. We thus predict for mergers that decrease efficiency decreases in both profits and sales.

The various possible consequences of mergers are depicted in Table 1. Since all horizontal mergers eliminate competition between the merging companies, any horizontal merger that meets our test for an increase in efficiency must do so because itsnet effect is to increase efficiency. The same can, of course, be said for the other types of mergers. All of our tests are thus of the net effects of mergers. In addition to the three possible consequences already discussed, the table includes a fourth—profits decline, but sales increase. Since this outcome is the mirror image of cell 2, we have labeled it ‘Market power reduction,’ but the question mark indicates our uncomfortableness with this categorization. No profit-maximizing manager would undertake a merger because she wanted to increase the amount of competition her firm faces. Sales increases coupled with profit declines might be observed, if the managers were sales or growth maximizers. Thus, both the motivation behind and the consequences of mergers falling into cell 3 are more difficult to identify than for the other three cells.

3 . Methodology

To determine whether a merger has increased profits or not, we must predict what the profits of the two merging firms would have been in the absence of the merger. Determining this counterfactual is one of the most difficult aspects in empirical investigations of the effects of mergers. Even if we observe a 20% increase in profits after a merger, it might be a failure because the two firms’ profits would have increased by 40% had they not merged. Most studies that try to address the counterfactual problem do so by choosing a benchmark group of firms and predicting that the performance of the merging firms would have followed that of the benchmark group. In event studies the benchmark group is usually the companies in the market portfolio, in studies like ours that use accounting data, the benchmark firms have been chosen from the same industry(ies) as that (those) of

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 629

6the two merging firms. We adopt the same methodology and assume that the merging companies’ profits and sales would have changed in the same way as the median firm’s profits and sales in their respective two-digit ISIC industries

7changed. We chose the two-digit level of aggregation as opposed, say, to the four-digit

level, because in many countries our samples of possible benchmark companies were too small to be able to identify a nonmerging company for every four-digit industry. Although a two-digit industry’s performance will track one of its four-digit industries with some error, we see no reason to expect a systematic bias from this choice of benchmark. Moreover, it has the possible advantage of not being contaminated by the merger itself, as would occur if the merger sys- tematically changed the sales or profits of all other firms in the merged company’s four-digit industry(ies).

Consider first the problem of predicting the merging companies’ sales. Define: S as the sales of the acquiring company in yeart 1 n; S as the sales of theGt1n Dt acquired company in yeart; S as the predicted sales of the combined companyCt1n in yeart 1 n; S as the sales of the median firm in the industry of the acquiringIGt1n company in yeart 1 n; andS as the sales of the median firm in the industry ofIDt1n the acquired company in yeart 1 n.

The predicted sales for the combined company in yeart 1 n is estimated as follows:

S SIGt1n IDt1n ]] ]]S 5 S 1 S (1)Ct1n Gt21 DtS SIGt21 IDt

The sales of the acquiring company are projected relative to its sales in the year prior to the merger, the sales of the acquired company are projected relative to its sales in the year of the merger.

It often happens, of course, that companies make several acquisitions over short spans of time. To allow for this possibility we amend Eq. (1) to take into account mergers occurring after timet. If, for example, a firm made one acquisition int and another int 1 2, then the amended formula for predictingS would lookCt1n like the following (for n > 2)

6 See the studies in Mueller (1980a), McDougall and Round (1986), Ravenscraft and Scherer (1987), and Healy et al. (1992).

7 Our control group excludes firms that made mergers in the periodt 2 1 to t 1 5, wheret is the year of the merger. In the small fraction of cases where no control group was available for the respective industry and country or country group, we take the median firm in the whole manufacturing / service sector of the respective country / country group. We also tried using the firms at the first and third quartiles of the distribution of firm sizes rather than the median, and matching merging firms to the nearest quartile. This did not produce a substantive change in our findings, so we report only the one set of results.

630 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

S S SIGt1n IDt1n IDt1n ]] ]] ]]S 5 S 1 S 1 S (2)Ct1n Gt21 Dt Dt12S S SIGt21 IDt IDt12

Many firms both acquire and sell assets. We also need to account for the effects of spin- and sell-offs on the merging companies’ sales. We do so by treating these sales symmetrically to acquisitions. Namely, we subtract the sales of any part of a company sold or spun-off during the 5 years after a merger, again scaling the sales of the spun-off unit by the changes in sales for the median firm in its industry.

If, for example, a firm made one acquisition in yeart, another in yeart 1 2 and spins or sells off a company in yeart 1 3, the final formula for predictingSCt1n would then be (forn > 3)

S S S SIGt1n IDt1n IDt1n ISt1n ]] ]] ]] ]]S 5 S 1 S 1 S 2 S (3)Ct1n Gt21 Dt Dt12 St13S S S SIGt21 IDt IDt12 ISt13

where S denotes the sales spun or sold off by the acquiring company in yearSt13 t 1 3 and S is the sales of the median firm in the industry of the divestedISt1n

8company in yeart 1 n. Our methodology for determining the effects of mergers on sales is to compare

the predicted value for the merged company’s sales in yeart 1 n after adjusting for all mergers and spin-offs as obtained using Eq. (3) with the actual level of sales of this company.

Projecting the levels of profits is a little more difficult, because they can take on negative and zero values. Taking ratios of profits at different points in time may introduce significant errors. We shall, therefore, use changes in the ratios of profits to total assets to predict changes in the profits of the merging companies. Define: P as the profits of the acquiring company in yeart 1 n; P as the profits ofGt1n Dt the acquired company in yeart; P as the predicted profits of the combinedCt1n company in yeart 1 n; P as the profits of the median firm in the industry ofIGt1n the acquiring company in yeart 1 n; P as the profits of the median firm in theIDt1n industry of the acquired company in yeart 1 n; K as the assets of theGt1n acquiring company in yeart 1 n; K as the assets of the acquired company in yearDt t; K as the assets of the median firm in the industry of the acquiring companyIGt1n in year t 1 n; and K as the assets of the median firm in the industry of theIDt1n acquired company in yeart 1 n.

We can now compute the projected change in the returns on the acquirer’s assets from yeart 2 1 to t 1 n using again the changes observed for the median (in terms of profitability) company in its industry. Call this projected changeD .IGt21,t1n

8 Two biases might occur: If sales data are missing on additional mergers fromt to t 1 5 we underestimate projected sales, if sales data are missing on spin- or sell-offs fromt to t 1 5 we overestimate projected sales. Additional mergers occur more often than divestitures, while divestitures are larger on average. Thus, the two biases potentially offset each other.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 631

P PIGt1n IGt21 ]] ]]D 5 2 (4)IGt21,t1n K KIGt1n IGt21

If the median firm in the acquirer’s industry earned a 0.10 return on assets in t 2 1, and a 0.11 return int 1 n, then we would predict that the acquiring firm’s returns on assets would increase by 0.01.

Defining D for the acquired firm’s industry analogously toDID t,t1n IG t21,t1n gives us the following formula for predicting the profits of the combined company in year t 1 n.

K KIGt1n IDt1n ]] ]]P 5 P 1 K D 1 P 1 K D (5)Ct1n Gt21 Gt21 IGt21,t1n Dt Dt IDt,t1nK KIGt21 IDt

The profits of the combined company in yeart 1 n are predicted to be the profits of the acquirer int 2 1, plus the predicted growth in its profits fromt 2 1 to t 1 n, plus the profits of the acquired firm int, plus the predicted growth in its profits from t to t 1 n. Eq. (5) can be modified to take into account additional acquisitions and spin-offs in the same way that Eq. (1) was. Thus, if we take the same example from above where a firm made one acquisition in yeart, another in yeart 1 2 and spins or splits off a company in yeart 1 3, the final formula for predictingPCt1n is then (for n > 3)

K KIGt1n IDt1n ]] ]]P 5 P 1 K D 1 P 1 K DCt1n Gt21 Gt21 IGt21,t1n Dt Dt IDt,t1nK KIGt21 IDt

(6) K KIDt1n ISt1n ]] ]]1 P 1 K D 2 P 2 K DDt12 Dt12 IDt12,t1n St13 St13 ISt13,t1nK KIDt12 ISt13

whereP are the profits spun or sold off in yeart 1 3, K are the assets ofS t13 ISt1n the median firm in the industry of the spun- or sold-off company in yeart 1 n, K are the assets of the spun- or sold-off company in yeart 1 3, andDSt13 ISt 1 3,t 1 n is the projected change in the returns on the spun- or sold-off company’s assets

9from year t 1 3 to t 1 n.

4 . Data description

Our principal source of data is theGlobal Mergers and Acquisitions database of Thompson Financial Securities Data (TFSD ). This company collects merger and

9 Again, two biases occur which potentially offset each other: If the relevant profits data on additional mergers undertaken fromt to t 1 5 are missing and taken over profits are positive (which they are on average), we underestimate projected profits. If the relevant profits data on divestitures undertaken from t to t 1 5 are missing and spun or sold off profits are positive (which they are on average), we overestimate projected profits.

632 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

spin-off data using a variety of sources such as financial newspapers, Reuters Textline, the Wall Street Journal, Dow Jones, etc. The database covers all transactions valued at US$1 million or more. We define a merger as a transaction

10where more than 50% of the equity of a target firm is acquired. During the period 1981 to 1998, there were 69 605 announcements of such mergers. Our data for the United States begin in the late 1970s, for all other countries in the mid-1980s. Fig. 1 presents the total number of deals by completion year.

Table 2A summarizes the characteristics of completed mergers (see also Table 2B). From the nearly 70 000 announced mergers across the world, nearly 45 000 mergers were actually completed with almost half of these taking place in the United States. For the full sample, horizontal mergers make up 42% of all mergers, conglomerate mergers 54% and vertical mergers only 4%. To be defined as a vertical merger at least 10% of the sales (purchases) of the primary four-digit industry to which one of the merging companies belongs must go to (come from) the industry to which the other belongs. We use the 1992 input–output table for the US economy to make this determination. Horizontal mergers are defined as

Fig. 1. The Number of Completed Mergers and Divestitures, 1981–1998. Source: SDC Thompson Financial Securities.

10 Symmetrically we define a spin- or sell-off as a transaction where more than 50% of the equity are disposed off. We use the term ‘divestitures’ interchangeably.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 633

Table 2A Summary statistics on mergers and acquisitions from around the world from 1981 to 1998

Period: Until 1990 1991 / 92 1993 / 94 1995 / 96 1997 / 98 Whole period

United States of America Number of deals 8194 1965 2840 3782 4367 21 148 Average deal value (Mn $) 238.2 102.8 137.6 217.0 408.7 246.7 Cross border 3.4% 11.7% 13.9% 16.0% 16.7% 10.6% Horizontal 39.6% 47.4% 48.7% 49.3% 48.9% 45.2% Vertical 5.8% 4.9% 3.8% 2.8% 2.8% 4.3% Conglomerate 54.6% 47.7% 47.5% 47.9% 48.3% 50.5%

United Kingdom Number of deals 1,180 501 790 1138 1108 4717 Average deal value (Mn $) 217.3 113.1 60.6 135.0 212.1 158.3 Cross border 35.0% 30.3% 26.8% 27.4% 29.0% 29.9% Horizontal 31.6% 35.9% 34.7% 37.8% 41.2% 36.3% Vertical 4.7% 5.0% 3.5% 4.3% 3.6% 4.2% Conglomerate 63.7% 59.1% 61.8% 57.9% 55.2% 59.5%

Continental Europe Number of deals 986 2125 1996 2359 2129 9595 Average deal value (Mn $) 393.4 186.1 159.2 220.4 414.1 285.9 Cross border 53.8% 24.2% 26.6% 33.3% 39.8% 33.5% Horizontal 37.0% 43.8% 37.5% 35.8% 39.6% 38.9% Vertical 4.8% 3.5% 3.3% 3.2% 3.4% 3.5% Conglomerate 58.2% 52.7% 59.2% 61.0% 57.0% 57.6%

Japan Number of deals 172 88 61 151 174 646 Average deal value (Mn $) 513.2 456.0 198.1 783.3 169.4 464.9 Cross border 80.8% 72.4% 59.0% 34.4% 28.2% 52.6% Horizontal 33.7% 29.5% 36.1% 35.1% 42.0% 35.9% Vertical 4.7% 0.0% 3.2% 2.0% 4.0% 3.1% Conglomerate 61.6% 70.5% 60.7% 62.9% 54.0% 61.0%

Australia /New Zealand /Canada Number of deals 671 425 549 766 821 3232 Average deal value (Mn $) 354.6 68.5 61.6 118.8 142.5 156.0 Cross border 37.9% 22.6% 32.4% 27.7% 27.9% 30.0% Horizontal 43.8% 43.3% 47.5% 40.1% 44.6% 43.7% Vertical 4.8% 1.9% 3.7% 3.1% 3.4% 3.5% Conglomerate 51.4% 54.8% 48.8% 56.8% 52.0% 52.8%

Rest of the World Number of deals 371 553 831 1728 1779 5262 Average deal value (Mn $) 276.2 150.0 87.5 101.9 143.3 128.3 Cross border 49.6% 25.7% 32.8% 25.0% 26.5% 28.5% Horizontal 34.8% 36.2% 34.7% 36.7% 40.1% 37.3% Vertical 6.4% 4.3% 2.7% 3.2% 3.5% 3.6% Conglomerate 58.8% 59.5% 62.6% 60.1% 56.4% 59.1%

634 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 2A. Continued

Period: Until 1990 1991 / 92 1993 / 94 1995 / 96 1997 / 98 Whole period

All Mergers Number of deals 11 574 5657 7067 9924 10 378 44 600 Average deal value (Mn $) 256.5 129.3 114.7 181.9 313.4 220.0 Cross border 15.5% 21.2% 23.0% 24.2% 25.5% 21.7% Horizontal 38.6% 43.4% 42.1% 41.7% 44.2% 41.7% Vertical 5.5% 4.0% 3.5% 3.1% 3.2% 4.0% Conglomerate 55.9% 52.6% 54.4% 55.2% 52.6% 54.3%

The database is theGlobal Mergers and Acquisition database ofThompson Financial Securities. It covers all transactions with a value of at least US $1 million.Continental Europe includes Austria, Belgium, Germany, Denmark, Spain, Finland, France, Greece, Ireland, Italy, Luxembourg, The Netherlands, Norway, Sweden, Portugal, Switzerland and Island. TheRest of the World sample includes more than 100 other countries.Deal value is defined as the total consideration paid by the acquirer excluding fees and expenses. The dollar value (deflated by the US-CPI with base year 1995) includes the amount paid for all common stock, common stock equivalents, preferred stock, debt, options, assets, warrants and stake purchases made within 6 months of the announcement date of the transaction. Liabilities assumed are included in the value if they are publicly disclosed. If a portion of the consideration paid by the acquirer is common stock, the stock is valued using the closing price on the last full trading day prior to the announcement of the terms of the stock swap.Cross-border mergers are mergers where the acquiring and acquired companies stem from different nations. Horizontal mergers are defined as mergers between two companies with sales in the same primary four-digit SIC industry.Vertical mergers are mergers where at least 10% of the sales (purchases) of the primary four-digit industry to which one of the companies belongs must go to (come from) the industry to which the other belongs. We use the 1992 US input–output table.Conglomerate mergers consist of all mergers which are neither horizontal nor vertical.

Table 2B Characteristics of acquiring and target companies

Number Sales Profits Profit rate of Obs.

Acquirer Target Acquirer Target Acquirer Target Mn $ Mn $ Mn $ Mn $

United States of America 1967 1997.5 318.0 102.26 9.78 0.029 0.019 United Kingdom 379 2162.1 329.7 110.53 10.89 0.066 0.039 Continental Europe 172 4644.2 729.6 169.86 24.58 0.035 0.033 Japan 16 4349.1 876.1 165.10 26.47 0.011 0.030 Australia / N.Zealand / Canada 172 1940.8 391.9 93.45 15.53 0.024 0.027 Rest of the World 47 2132.4 443.0 157.64 22.88 0.052 0.013 All mergers 2753 2198.0 355.3 108.25 11.53 0.034 0.023

The sample includes those mergers where we have all the relevant data for yeart. Sales are average sales in million 1995 USD, profits are average profits before interest and taxes in million 1995 USD. The profit rate is profits before interest and taxes divided by total assets.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 635

mergers between two companies with sales in the same primary four-digit industry. Conglomerate mergers consist of all mergers which are neither horizontal nor vertical. It is interesting to note that a greater fraction of mergers in the United States appears to be horizontal than for any other area / country category. Despite the step-up in antitrust enforcement under the Clinton Administration, a greater fraction of mergers between 1993 and 1998 was horizontal in nature than from before 1990. In some years the proportion is nearly 50%. Roughly one-fifth of the mergers are cross-border transactions (22%). An upward trend in cross-border mergers is apparent in Table 2, and is confirmed by a regression of the fraction of mergers in a country which are cross-border on time. The upward trend is particularly pronounced for EU countries, which may reflect the greater integration

11of EU economies over the 1990s as a result of the Single Market Program. Japanese companies undertake relatively few mergers compared to other countries, with a much higher fraction of these (53%) being cross-border deals. To arrive at comparable real values, we first convert all variables to USD and deflate by the US-Consumer Price Index with base year 1995. Thus, the average deal value was

12220 million 1995-USD. The samples used for our analysis are much smaller than the numbers in Table

2A,B suggest due to missing data for relevant variables. Acquiring company balance sheet and market data for the yearst 2 1 to t 1 5 relative to the merger year t stem from theGlobal Vantage /Compustat database. Out of the 45 000 completed mergers of Table 2 we could match 17 863 to one of these databases. Some acquiring companies acquire more than one target in a given year, and since our balance sheet information for acquiring companies is on a yearly basis, we aggregate the relevant variables of these targets. This further reduces the merger

13sample to 14 269 merger years.

11 Call CB the fraction of mergers in a country which are cross-border mergers,Y the year in which the mergers took place, andY2EU an interaction betweenY and a 1 / 0 dummy for whether the country was an EU-member or not. A regression using data for the 1990s yielded the following results (t-statistics under the coefficients):

2CB5 0.251 0.008Y 1 0.007Y 2 EU, R 5 0.06 7.33 1.65 2.30

The upward time trend for EU countries was almost twice that for all other countries, a difference that was significant at the 1% level. (We did not include data for the 1980s in the regression, since our data from before 1990 for countries other than the USA contain a disproportionate fraction of large, and therefore also cross-border mergers.)

12 A table summarizing the characteristics of divestitures is available upon request. In short, our database covers 9659 completed divestitures worldwide from 1981 to 1998, 31.4% of these were cross-border deals, 37.0% horizontal, 4.3% vertical, and 58.7% conglomerate. The average deal value was $181 million.

13 We could match 6616 divestitures to these databases aggregated to 4666 divestiture years.

636 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 3A Effects of mergers for full sample

Years after Number of Difference in Profits Difference in Sales the merger observations Mn $ Mn $

p-value % Positive p-value % Positive

t 1 1 2704 5.91 0.062 57.0 2214.16 0.000 51.5 t 1 2 2274 11.11 0.009 57.2 2382.81 0.000 49.5 t 1 3 1827 10.79 0.056 54.8 2549.59 0.000 46.4 t 1 4 1517 19.68 0.007 57.8 2633.46 0.000 46.3 t 1 5 1250 17.81 0.046 57.6 2714.04 0.000 44.6

On average, acquiring firms are present over a time period of 15 years in our databases. During this period acquiring firms make 2.25 acquisitions and divest 0.83 companies on average. Missing data and the elimination of outliers (we drop the left and right 1% tail of the distribution) reduces the sample to the numbers

14reported in Table 3A. We have attempted to make our samples as large as possible and thus do not limit ourselves to balanced panels, companies making only one merger or the like.

Table 2B presents means of the distributions of sales, profits and profit to assets ratios for the acquired and acquiring companies in our sample. Profits are measured before interest and taxes (COMPUSTAT item 18), net sales are item 12, and total assets are item 6. Again all variables are deflated by the Consumer Price Index with base year 1995.

On average the acquired companies are just 16% of the size of the companies which buy them and make only around a tenth of the profits. In the United States, the United Kingdom and Continental Europe the acquired firms are less profitable than their buyers, in Japan, Australia, Canada, and New Zealand they are more profitable. In the rest-of-the-world subsample, the acquired companies are much

15less profitable than their buyers.

5 . Overall results

5 .1. Full sample

In this section we present the main results for our full sample and for different subsets of mergers to see whether mergers on average have increased profits and sales or reduced them. In the following section we look more closely at the mergers that have increased profitability to see whether the changes appear to be due to increases in efficiency or market power.

14 In Tables 3–9 we drop the left and right 1% of the distribution in each (sub)sample. 15 Summary statistics on divestitures (available upon request) reveal that divested units are larger and

less profitable than acquired firms.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 637

Table 3A presents our findings for the full sample of companies. The size of the sample declines as we move away from the date of the merger because companies

16disappear from the data set. The profitability numbers consist of the difference in year t 1 n between the actual profits of the combined firm and its projected profits in this year. Thus a negative number implies a decline in profits. The mean difference between actual and projected profitability is positive in all 5 years after the mergers, and is significant in every year at the 10% level, or better. The $17.8 million constitutes a difference between actual and projected profits of 8.2 (0.20)% of the profits (assets) of the average acquirer in the sample in yeart 1 5.

The results for sales are again the difference between the actual and projected values for the combined companies. The mean difference in sales is negative in every year and continuously increases in absolute value through year 5. Five years after the mergers, the average acquiring firm had sales that were $714 million lower than their projected value. This constitutes a difference between actual and projected sales of 14.5% of the sales of the average acquirer in the sample in year t 1 5. The last column in each set of results gives the fraction of the sample for which the change was positive. While a majority of mergers led to higher actual profits than those predicted, the reverse was true for sales.

5 .2. Results by country

Table 3B reports the comparable figures by country or country group. The United States makes up a substantial fraction of the overall sample and so it is not surprising to find the pattern of results for it resembling that of the full sample as just discussed. Profits are higher than predicted in every post-merger year, although only three of the five differences are significant at the 10% level for the US. Actual sales are significantly less than predicted in every post-merger year. In percentage terms we predict that mergers increase profits by 8.1% (0.17%) of the profits (assets) of the average acquirer in the USA and decrease sales by 14.8% 5 years after the merger.

Essentially the same pattern can also be observed for the United Kingdom. Actual profits are greater than projected profits in all 5 years, although the difference is statistically significant in only the first post-merger year. Actual sales fall short of their projected values in all 5 years after the mergers, with all of the declines significant at conventional levels.

The pattern of results for Continental Europe is very similar to that for the USA and UK. The differences between actual and projected profitability are all positive, but the only significant difference is for the fourth post-merger year. Sales fall

16 Remember our last year is 1998, thus mergers having taken place in 1993 are the last mergers for which we have data until yeart 1 5, mergers having taken place in 1994 are in our sample only up to year t 1 4, mergers of year 1995 up to yeart 1 3 and so on.

638 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 3B Effects of mergers by country / country groupings

Years Number of Profits Sales after the observations

Difference p-value % Difference p-value % merger

in Mn $ Positive in Mn $ Positive

United States of America t 1 1 1950 3.735 0.307 57.0 2174.495 0.000 52.8 t 1 2 1641 12.457 0.013 58.1 2324.825 0.000 49.8 t 1 3 1272 10.490 0.133 55.6 2524.798 0.000 46.6 t 1 4 1067 16.654 0.054 57.9 2595.367 0.000 45.5 t 1 5 889 17.388 0.098 58.7 2730.236 0.000 44.3

United Kingdom t 1 1 362 15.440 0.061 65.7 2263.828 0.001 48.6 t 1 2 322 14.902 0.135 59.3 2445.977 0.000 48.9 t 1 3 297 12.545 0.287 52.2 2468.442 0.002 45.4 t 1 4 233 4.729 0.777 55.8 2380.410 0.050 47.0 t 1 5 181 24.149 0.201 53.6 2545.682 0.043 43.5

Continental Europe t 1 1 178 18.831 0.233 53.9 2568.403 0.001 47.3 t 1 2 140 16.015 0.462 55.7 21106.104 0.000 46.2 t 1 3 122 19.191 0.457 53.3 2972.056 0.006 47.9 t 1 4 108 81.284 0.016 60.2 21461.227 0.002 48.5 t 1 5 87 42.345 0.361 58.6 2666.390 0.272 54.2

Japan t 1 1 20 236.826 0.342 35.0 2238.893 0.652 61.1 t 1 2 19 263.507 0.276 21.1 378.774 0.474 56.3 t 1 3 19 18.149 0.660 42.1 396.802 0.284 52.9 t 1 4 16 4.031 0.934 43.8 270.744 0.900 56.3 t 1 5 15 241.621 0.740 73.3 22328.611 0.187 46.2

Australia / New Zealand / Canada t 1 1 165 23.275 0.801 45.5 2175.353 0.130 47.9 t 1 2 129 227.001 0.093 45.6 2357.068 0.087 51.2 t 1 3 101 29.984 0.640 55.4 2686.854 0.014 44.6 t 1 4 79 5.862 0.858 54.3 2962.244 0.016 48.1 t 1 5 66 233.577 0.308 47.0 2805.393 0.121 39.4

Rest of the world t 1 1 42 26.539 0.296 51.2 2346.740 0.106 45.2 t 1 2 35 71.808 0.086 61.8 2237.196 0.174 42.9 t 1 3 25 44.931 0.377 65.2 2880.127 0.018 40.0 t 1 4 22 93.866 0.153 89.5 2577.552 0.223 50.0 t 1 5 15 115.937 0.250 64.3 2281.547 0.390 46.7

‘Difference in Mn $’ is the difference between actual and projected profits or sales as obtained by Eqs. (3) and (6) in 1995 million USD. A positive number therefore implies that the merger increased profits or sales, a negative number implies that the merger decreased profits or sales. ‘p-value’ is the probability that the observed differences are zero (two-sided test). ‘% Positive’ is the percentage of positive differences between actual and projected values.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 639

short of their projected values in every year and four of the five differences are statistically significant at the 5% level.

The results for Japan are somewhat different than those already discussed. Three of the five profit comparisons are negative, while sales are greater than predicted for the first time in two of the five post-merger years. Our sample for Japan is quite small, however, and none of the differences is statistically significant.

The results for Australia, New Zealand and Canada resemble those for the US, UK and Continental Europe in so far as actual sales fall short of predicted sales in all five post-merger years with three of the short falls being significant at the 10% level or better. The post-merger profit differences are also generally insignificant, as was the case for the UK and Western Europe, although in the case of Australia, New Zealand and Canada the post-merger profits of the merging firms tend to be less than those predicted for them, and one of these differences is significant at the 10% level.

The pattern of results for the remaining countries also resembles that for the US, UK and Continental Europe. Profit differences are positive in all 5 years, but are usually insignificantly different from zero. Sales differences are again consistently negative, although only one of these is statistically significant

Thus, the results by country and country group tend by and large to resemble one another. Differences between actual and projected profits tend to be positive but often are not significantly different from zero. Differences between actual and projected sales tend to be negative and often significantly so.

The lack of significant differences in results across countries can be further illustrated through an analysis of variance. Table 4 reports the results from a regression of the differences between actual and projected profits and sales on country category dummies for yeart 1 5. An intercept has been included and the

Table 4 Analysis of variance in yeart 1 5 by country categories

Country / country group Profits Sales

Difference t-value Difference t-value in Mn $ in Mn $

Average 17.8 2.00 2714.0 6.63 USA 20.4 0.33 216.2 0.70 UK 6.3 0.38 168.3 1.13 Continental Europe 24.5 0.37 47.6 0.55 Japan 259.4 0.85 21615.0 1.83 Aus / NZ / Can 251.4 1.32 291.4 0.45 Rest of the world 98.1 1.26 432.5 0.63

2Adjusted R 20.0006 0.0003 Number of observations 1250 1250

‘Average’ denotes the overall average value of the difference of actual and projected profits or sales. All other coefficients are differences from this average.

640 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

country dummies constrained to sum to zero, so that the coefficient on a country dummy represents the difference between its mean and that for the full sample (Suits, 1984).

For the full sample, the mean difference between actual and predicted profits in year t 1 5 is positive and significant at the 5% level. No country category’s mean is significantly different from that of the full sample. The mean difference between actual and predicted sales is negative and significant at the 1% level. All country means are insignificantly different from the sample mean except for Japan, whose mean difference in sales is significantly less than the sample mean, although only at the 10% level.

5 .3. Results by sector and type of merger

In Table 5A,B we have separated mergers into the manufacturing and service sectors, and then within these divided them into horizontal, vertical and conglom- erate mergers. Mergers in the manufacturing sector tend to be less profitable than in the service sector. All 15 entries in the service sector are positive, while six of the 15 are negative in the manufacturing sector. The differences between actual and predicted sales are uniformly negative except for vertical mergers in the service sector, where two of the differences are positive.

Table 6 presents the results for an analysis of variance conducted in much the same way as in Table 4. Coefficients on the merger categories represent differences from the intercept. In yeart 1 5, horizontal mergers in manufacturing are significantly more profitable than the average merger in manufacturing, which had a near zero difference between its actual and projected values (Panel A). Vertical mergers in manufacturing are significantly (at the 10% level) less profitable, on the other hand. In contrast, all three categories of mergers are equally profitable in the service sector (Panel B). The difference between actual and projected profits for the average merger in the service sector is significantly

17higher than for the average merger in manufacturing. Although actual sales fall short of predicted sales in all three categories for the

manufacturing sector, the shortfall is significantly smaller for horizontal mergers. Thus, within the manufacturing sector, horizontal mergers appear to be con- siderably more successful than conglomerate and vertical mergers with respect to their effect on both profits and sales.

Within the service sector, vertical mergers exhibit the best performance in terms of sales, although the small number of vertical mergers makes the difference statistically insignificant. Horizontal mergers still produce smaller shortfalls between actual and projected sales than do conglomerate mergers.

Thus, we conclude that mergers in the service sector are generally more successful than those in manufacturing, at least as far as their effects on

17 Test results are available from the authors upon request.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 641

Table 5 Effects of mergers by sector

Years after Number of Profits Sales after the observations

Difference p-value % Difference p-value % merger

in Mn $ Positive in Mn $ Positive

Panel A. Effects of mergers in the manufacturing sector by category

Horizontal mergers t 1 1 411 28.006 0.370 51.3 2180.323 0.002 47.1 t 1 2 352 3.130 0.761 57.1 2288.936 0.000 44.9 t 1 3 274 15.924 0.252 56.9 2466.510 0.001 48.2 t 1 4 233 41.933 0.007 60.5 2467.476 0.002 46.6 t 1 5 193 41.751 0.017 56.5 2195.891 0.268 43.2

Vertical mergers t 1 1 66 31.234 0.270 55.4 284.619 0.637 53.0 t 1 2 53 211.697 0.702 42.6 242.079 0.897 49.1 t 1 3 47 252.549 0.112 38.3 2397.957 0.343 46.8 t 1 4 43 271.252 0.231 43.2 2773.660 0.152 55.8 t 1 5 34 288.254 0.340 51.4 2989.052 0.188 50.0

Conglomerate mergers t 1 1 877 8.133 0.175 55.8 2411.540 0.000 45.5 t 1 2 761 12.253 0.115 54.5 2605.256 0.000 44.3 t 1 3 641 7.833 0.409 52.3 2768.647 0.000 42.0 t 1 4 541 8.567 0.494 52.7 2735.062 0.000 42.6 t 1 5 475 25.879 0.674 52.4 2824.688 0.000 42.9

Panel B. Effects of mergers in services by category

Horizontal mergers t 1 1 775 12.177 0.017 60.0 244.617 0.369 61.7 t 1 2 624 14.211 0.093 59.5 2189.847 0.009 59.8 t 1 3 470 5.772 0.627 55.5 2316.710 0.004 52.1 t 1 4 368 22.877 0.088 63.1 2492.849 0.001 50.0 t 1 5 287 39.167 0.038 65.7 2545.498 0.007 52.3

Vertical mergers t 1 1 22 23.377 0.248 50.0 2234.462 0.399 45.5 t 1 2 19 9.967 0.543 52.6 211.693 0.919 42.1 t 1 3 17 38.608 0.031 64.7 48.534 0.929 41.2 t 1 4 15 11.566 0.781 73.3 2376.665 0.338 50.0 t 1 5 8 104.254 0.013 100.0 933.507 0.588 50.0

Conglomerate mergers t 1 1 550 0.716 0.914 59.8 2178.648 0.016 50.7 t 1 2 465 14.446 0.095 60.6 2406.578 0.001 48.4 t 1 3 374 26.555 0.034 59.1 2584.358 0.000 46.0 t 1 4 309 33.924 0.059 59.9 2735.722 0.000 47.2 t 1 5 247 36.059 0.100 59.5 21112.637 0.000 40.1

The manufacturing sector includes all firms with SIC codes smaller than 4000, theservice sector includes those firms with SIC codes greater than or equal to 4000. See also the note to Table 3.

642 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 6

Category Profits Sales

Difference t-value Difference t-value in Mn $ in Mn $

Panel A. Analysis of variance in year t 1 5 in the manufacturing sector by merger categories Average 3.1 0.27 2660.0 5.19 Horizontal 38.7 2.07 464.1 2.25 Vertical 291.4 1.82 2329.1 0.59 Conglomerate 29.0 1.13 2164.7 1.87

2Adjusted R 0.0066 0.0045 Number of observations 702 702

Panel B. Analysis of variance in year t 1 5 in services by merger categories Average 38.7 2.75 2782.1 4.87 Horizontal 0.5 0.03 236.6 1.56 Vertical 65.5 0.57 1715.6 1.31 Conglomerate 2.6 0.17 2330.5 1.88

2Adjusted R 20.0031 0.0051 Number of observations 542 542

‘Average’ denotes the overall average value of the difference of actual and projected profits or sales. All other coefficients are differences from this average. See also the note to Table 5.

profitability are concerned, and that horizontal mergers have more favorable effects on sales than do conglomerate mergers in both sectors, and on profits in

18manufacturing.

5 .4. The effects of cross-border mergers

Table 7 breaks the sample into cross-border and domestic mergers. We have at most 429 observations on cross-border mergers, and so the results for domestic mergers look a lot like those for the full sample. The same can more or less be said for the cross-border mergers. Mean differences between actual and projected profits are positive in all five post-merger years, but are significantly different from zero in only one of them. Mean differences between actual and projected sales are negative and significant in all five post-merger years. We tested for differences in the effects of cross-border mergers that were related to the origin of either the acquiring or target company, but did not find any significant differences. Cross- border acquisitions by (of) UK companies did not generate significantly larger

18 We also tested for significant differences in the effects of mergers depending on the industry of the acquiring companies. Almost no significant differences were found. The most interesting exceptions were for the chemicals and insurance industries. Mergers in these industries were followed by profit increases significantly above the sample mean, and sales declines below the mean. The patterns of profits and sales changes following mergers in the chemicals and insurance industries strongly resemble those that we associate with market power increases.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 643

Table 7 Effects of domestic and cross-border mergers

Years Number of Profits Sales after the observations

Difference p-value % Difference p-value % merger

in Mn. $ Positive in Mn. $ Positive

Cross-border mergers t 1 1 429 16.136 0.121 58.3 2385.824 0.000 48.1 t 1 2 336 15.727 0.170 58.3 2555.023 0.000 47.3 t 1 3 286 3.886 0.803 53.8 2871.451 0.000 44.9 t 1 4 236 37.202 0.050 66.1 2785.575 0.002 47.0 t 1 5 183 41.826 0.132 62.8 2867.729 0.022 46.2

Domestic mergers t 1 1 2288 3.986 0.214 56.8 2182.953 0.000 52.1 t 1 2 1940 10.305 0.025 57.0 2353.158 0.000 49.9 t 1 3 1544 12.067 0.046 55.0 2490.591 0.000 46.6 t 1 4 1281 16.454 0.036 56.3 2605.429 0.000 46.1 t 1 5 1064 13.689 0.141 56.7 2687.170 0.000 44.4

See the note to Table 3.

changes in sales and profits than was true for other cross-border acquisitions, and the same was true for all other countries.

6 . Results: market power and efficiency

Mergers that increase the efficiency of the merging firms should increase both their profits and their sales. Mergers that increase market power should increase profits and reduce sales. A merger which reduces efficiency should reduce both profitability and sales. In this section we attempt to increase our understanding of the causes and effects of mergers, by dividing our sample into subsets of mergers that either increase or reduce profitability.

Panel A1 of Table 8 reports the results for all mergers for which post-merger profitability changes are greater than those of the matching industries, while Panel A2 reports the figures for the mergers that reduced profitability relative to the control group. The mean difference between actual and projected sales is negative and significant in every post-merger year. The difference between actual and projected profits in yeart 1 5 is more than $150 million for profitable mergers, a difference of 70.0% (1.70%) of the actual profits (assets) of the average acquirer in the sample in yeart 1 5. The difference between actual and projected sales int 1 5 is $2475 million, 29.6% of the sales of the average acquirer int 1 5. This is the pattern we expect for mergers that increase market power, and thus we conclude that theaverage profitable merger in our sample would appear to have increased market power.

644 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 8 Tests for efficiency and market power effects

Years Number of Profits Sales p-value after the observations

Difference Difference merger

in Mn. $ in Mn. $

Panel A1. Mergers with profits changes above zero t 1 1 1512 76.129 292.148 0.013 t 1 2 1276 97.129 2247.630 0.000 t 1 3 981 117.517 2328.543 0.000 t 1 4 857 140.957 2399.243 0.000 t 1 5 706 152.181 2475.338 0.000

Panel A2. Mergers with profits changes below zero t 1 1 1192 283.419 2368.936 0.000 t 1 2 998 299.076 2555.640 0.000 t 1 3 846 2113.522 2805.902 0.000 t 1 4 660 2139.492 2937.575 0.000 t 1 5 544 2157.147 21023.821 0.000

Panel B1. Mergers with profits changes in top quartile t 1 1 661 160.825 2191.652 0.017 t 1 2 557 205.372 2554.697 0.000 t 1 3 447 240.393 2600.628 0.000 t 1 4 368 299.393 2821.384 0.000 t 1 5 305 323.198 2817.953 0.004

Panel B2. Mergers with profits changes in lower quartile t 1 1 664 2144.851 2666.086 0.000 t 1 2 558 2171.933 2903.308 0.000 t 1 3 450 2205.304 21424.606 0.000 t 1 4 377 2239.140 21501.687 0.000 t 1 5 308 2269.075 21631.660 0.000

Panel C. Horizontal mergers with profits changes above zero t 1 1 664 70.810 220.014 0.684 t 1 2 558 91.082 2157.492 0.048 t 1 3 410 113.251 2249.082 0.030 t 1 4 367 125.995 2252.037 0.058 t 1 5 294 148.933 2238.859 0.183

Panel D. Vertical mergers with profits changes above zero t 1 1 47 132.926 2192.295 0.451 t 1 2 33 115.225 174.208 0.641 t 1 3 29 84.576 287.980 0.499 t 1 4 28 137.429 153.551 0.705 t 1 5 25 161.787 710.515 0.240

Panel E. Conglomerate mergers with profits changes above zero t 1 1 796 77.713 2146.062 0.008 t 1 2 680 101.754 2344.146 0.000 t 1 3 539 122.894 2424.531 0.000 t 1 4 457 153.935 2551.364 0.001 t 1 5 384 154.530 2735.240 0.000

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 645

Table 8. Continued

Years Number of Profits Sales p-value after the observations

Difference Difference merger

in Mn. $ in Mn. $

Panel F. Horizontal mergers with profits changes below zero t 1 1 519 279.689 2182.776 0.002 t 1 2 416 297.804 2316.613 0.000 t 1 3 334 2116.183 2522.617 0.000 t 1 4 235 2117.672 2843.660 0.000 t 1 5 185 2130.796 2669.969 0.003

Panel G. Vertical mergers with profits changes below zero t 1 1 41 287.076 241.590 0.770 t 1 2 39 2106.117 2210.288 0.504 t 1 3 35 2121.891 2749.439 0.140 t 1 4 29 2244.293 21477.246 0.038 t 1 5 17 2380.079 22583.682 0.065

Panel H. Conglomerate mergers with profits changes below zero t 1 1 627 286.521 2544.284 0.000 t 1 2 539 299.820 2763.271 0.000 t 1 3 471 2111.278 21016.108 0.000 t 1 4 392 2144.490 2949.741 0.000 t 1 5 336 2160.265 21138.591 0.000

Not surprisingly, actual sales for companies undertaking unprofitable mergers (Panel A2) fall way below their projected values. We predict that had the acquiring firms not undertaken these mergers they would have had 72.3% more profits and 20.8% more sales than they actually had in yeart 1 5. These mergers are unsuccessful in both dimensions and imply that they lowered efficiency.

In Panel B1 of Table 8 the results are reported for the highest quartile of mergers ranked by the difference between actual and projected profits. The average profit changes are roughly three times as large as those in Panel A1. Mean actual sales continue to fall short of their projected values in every year after the mergers. All sales comparisons are highly significant. In Panel B2 of Table 8 the results are reported for the lowest quartile of mergers ranked by changes in profits. These mergers appear as unmitigated disasters.

Panels C, D and E in Table 8 divide mergers with changes in profitability above the matching industries into the horizontal, vertical and conglomerate categories. The first thing to note is that all three categories of successful mergers exhibit roughly similar increases in profitability. The mean differences between actual and projected profits tend to get larger as one moves away from the mergers, and fall roughly in a range from $150 to $160 million in yeart 1 5.

The mean differences between actual and projected sales for companies undertaking profitable horizontal and conglomerate mergers are negative in all 5

646 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

years following the mergers. Thus, the average merger falling in both categories appears to result in an increase in market power. In contrast the mean difference between actual and projected sales for firms undertaking vertical mergers is negative only in year one. Although none of the other four entries is statistically significant, the results for profitable vertical mergers areweakly consistent with their increasing efficiency.

Panels F, G and H in Table 8 parallel C, D and E for mergers that lowered profitability. All 15 post-merger sales comparisons are negative, with all differ- ences for horizontal and conglomerate mergers being statistically significant, as were two for vertical mergers. The average unprofitable merger fits the pattern we anticipate for efficiency reducing mergers regardless of what type of merger it is.

One might expect mergers between small firms to be more likely to increase efficiency by creating economies of scale and scope, while mergers between large firms would be more likely to increase market power. These conjectures would lead us to expect sales increases following profitable mergers between small companies, and sales decreases following profitable mergers between large companies. Our final test for the effects of mergers splits our sample into small

19and large acquirers, and profitable and unprofitable mergers. The results of these tests are reported in Table 9. The mean differences between

actual and projected sales are positive and significant in all five post-merger years for the small firms making profitable mergers (Panel A). These differences suggest that profitable mergers of small firms increase sales by around $150 million or 25.0% relative to the average small acquirer’s size in yeart 1 5, while profits nearly double. This pattern accords with our prediction for efficiency enhancing mergers and is the first time that actual sales have exceeded their projected values on average in each of the five post-merger years. These results strongly suggest that these mergers increased the efficiency of the merging firms.

In contrast mean differences between actual and projected sales are negative and significant in all five post-merger years for the large firms making profitable mergers. These differences suggest that profitable mergers of large firms decrease sales by around $1 billion or 10.7% relative to the average large acquirer’s size in year t 1 5, while the change in profits is 60.7% of the profits of the average large acquirer int 1 5. These differences accord with our prediction for market power enhancing mergers. The average profitable merger among small firms appears to increase their efficiency, the average profitable merger by a large firm appears to increase its market power.

The results of Panel B of Table 9 are for the firms, which undertook unprofitable

19 The full sample was first divided into ‘small’ and ‘large’ companies using the median sales of acquiring firms in yeart 2 1 as the dividing line. These two samples were then subdivided on the basis of whether profits were greater or less than their projected values. ‘Large’ firms have average sales (profits) of $5713 (264) million and ‘small’ firms have average sales (profits) of $341 (18.1) million in yeart 2 1. The average deal value of transactions involving ‘large’ acquirers is $667 million, while the average deal value involving ‘small’ acquirers is $103 million.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 647

Table 9

Years Number of Profits Sales p-value after the observations

Difference Difference merger

in Mn. $ in Mn. $

Panel A. Mergers with profits changes above zero by size

I. Small Firms t 1 1 766 20.440 54.953 0.000 t 1 2 642 27.947 72.190 0.000 t 1 3 476 36.465 83.328 0.001 t 1 4 418 40.155 129.245 0.000 t 1 5 349 47.001 148.724 0.002

II. Large Firms t 1 1 746 133.310 2243.194 0.001 t 1 2 634 167.294 2571.486 0.000 t 1 3 505 193.925 2716.762 0.000 t 1 4 439 236.070 2902.450 0.000 t 1 5 357 255.298 21085.415 0.000

Panel B. Mergers with profits changes below zero by size

I. Small Firms t 1 1 610 228.854 259.829 0.001 t 1 2 514 230.951 288.808 0.000 t 1 3 453 243.943 2128.520 0.009 t 1 4 356 255.022 2105.590 0.004 t 1 5 288 253.384 265.567 0.215

II. Large Firms t 1 1 582 2140.806 2692.914 0.000 t 1 2 484 2171.866 21051.408 0.000 t 1 3 393 2194.135 21586.702 0.000 t 1 4 304 2239.117 21911.873 0.000 t 1 5 256 2274.339 22101.856 0.000

The full sample was first divided into ‘small’ and ‘large’ companies using the sales median of acquiring firms in yeart 2 1 as the dividing line. These two samples were then subdivided on the basis of whether profits were greater or less than their projected values.

mergers. Here we see consistent declines in post-merger sales for both size classes. Unprofitable mergers by both small and large companies tend to be the result of reduced economic efficiency.

7 . Comparisons with previous results in the literature

The results reported above with respect to the effects of mergers on profitability and sales are broadly consistent with those obtained by others. In a recent survey of the literature Mueller (1997) summarized the results from 20 studies drawn

648 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

from 10 countries over the post-World War II period that generally followed the methodology that we have employed here to determine the effects of mergers on profitability, namely compared actual post-merger profits with those predicted using a control group.

The most ambitious of all of the studies in terms of sample size, time span, and care in handling the data was that of Ravenscraft and Scherer (1987) for the United States. They concluded that the profitability of acquired firms declined after they were acquired. On the other hand, Healy et al. (1992) found a significant increase in the pre-tax cash flows of the companies involved in the 50 largest mergers between 1979 and 1984 implying that the largest mergers in the US during the early 1980s did increase either the market power or the efficiency of the

20merging firms. Our results suggest that the profit increases that Healy et al. observed were mostly due to increases in market power.

The largest study of mergers in the UK (Meeks, 1977) concluded as did Ravenscraft and Scherer that mergers reduced the profitability of the merging companies. Other studies for the UK have, however, reached the opposite conclusion (Cosh et al., 1980). Although the preponderance of evidence for the UK suggests that mergers tend to reduce profitability (Hughes, 1989), not all studies have reached this conclusion.

No distinct pattern emerges in the studies from other countries. Profit increases were observed in Canada (Baldwin, 1995, Chapter 10) and Japan (Ikeda and Doi, 1983), profit decreases in Holland (Peer, 1980) and Sweden (Ryden and Edberg, 1980). In all other countries the differences were statistically insignificant. Where mergers seem to result in profit increases in one country (e.g., Germany), they result in declines in another (e.g., France). Thus, our overall finding that the actual post-merger profits of merging companies are in many cases insignificantly different from their predicted values is in general accordance with the findings of previous studies. Where we perhaps differ from them is that we have observed a greater preponderance of positive and significant profit changes following mergers.

Our findings with respect to post-merger changes in sales for the surviving firms also accord with the main results reported in the literature. Since we project a merging company’s sales using the median sales of a non-merging company in the same industry, one might expect that relative declines in sales will translate into declines in market shares. Three studies of the effects of mergers on market shares exist. Goldberg (1973) observed insignificant changes in market shares for a sample of 44 advertising intensive firms over an average of 3.5 years following their undertaking a merger. Mueller (1985) observed significantdeclines in market shares for a sample of 209 manufacturing companies over an average of 11 years following the mergers. Baldwin and Gorecki (1990) found significant declines in market shares for plants acquired in horizontal mergers, but no significant changes

20 Ravenscraft and Scherer also reported that ‘mergers among equals’—which is to say between two large firms—were more profitable than the average merger in their sample.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 649

for plants acquired in other sorts of mergers. They concluded that their results were consistent with the mergers having increased market power.

Nine studies that measured changes in the growth rates of merging firms following the mergers using either their industries or matched non-merging firms as control groups found either that the mergers produced no significant change in growth rates [see McDougall and Round (1986) for Australia; Kumps and Wtterwulghe (1980) for Belgium; Jenny and Weber (1980) for France; Cable et al. (1980) for Germany; Ryden and Edberg (1980) for Sweden; Cosh et al. (1980) for the United Kingdom; and Amel and Rhoades (1989) for acquired US banks], or significant declines [Peer (1980) for Holland and Mueller (1980b) for the United States]. Thus, no study of which we are aware has found significant increases in either the internal growth rates of merging companies or their market shares following their acquisitions, and several have reported significant declines. Our general finding of smaller sales for merging companies than are projected using the sales changes of the median nonmerging firm in the merging companies’ industries is consistent with these results from the literature.

8 . A categorization of mergers according to their effects on market power and efficiency

We begin this paper by stating that mergers can be divided into three broad categories: those that increase profits by increasing market power, those that increase profits by increasing efficiency and those that reduce profits and efficiency. In Table 1 we categorized these and the other logically possible consequence of mergers. Table 10 summarizes the results of our study by reporting the fractions of mergers that fall into each of the four categories. The first entry in each sell gives the percentage of all acquisitions by small companies falling into this cell, the second entry is for large acquirers, and the third is for all mergers regardless of size. Cell 1 reveals that 29.1% of the mergers in our sample resulted in increases in both sales and profits, and thus met our criterion for efficiency increasing mergers. A larger fraction of mergers by small firms (34.7%) satisfied our criterion for an efficiency-increasing merger than was true for large firms (23.4%) (difference significant at the 1% level).

Roughly the same fraction of mergers reduced efficiency (cell 4) as increased it. Here, however, there was no difference related to size. Small firms were just as likely to undertake a merger that reduced both profits and sales as were large firms.

A slightly smaller fraction of mergers met our criteria for a market power increase than did so for an efficiency increase. As one expects, large firms accounted for a significantly larger fraction of market power increasing mergers (34.8%) than did small companies (20.4%). Thus, some 85% of the mergers in our sample fall into the three main categories ‘efficiency increasing’, ‘efficiency

650 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

Table 10 Classification of mergers by firm size in yeart 1 5 (Percent of mergers)

DP .0 DP ,0

1 3 Small 34.7 17.5

DS.0 Large 23.4* 12.7* All 29.1 15.1

2 4 Small 20.4 27.4

DS,0 Large 34.8* 29.1 All 27.6 28.2

DP .0 (DP ,0) denotes that the mergers resulted in a profit increase (decrease) relative to yeart and relative to industry and country peers.DS.0 (DS,0) denotes that the mergers resulted in a sales increase (decrease) relative to yeart and relative to industry and country peers. The first number in each cell is for small firms (total sales less than the median in yeart21), the second number in each cell is for large firms (total sales more than the median in yeart21), and the third number in each cell is the overall proportion. A * denotes that the proportion of small firms is significantly different from the proportion of large firms at the 1% level, two-sided test.

reducing’ or ‘market power increasing’, and they are divided roughly equally across them.

These comparisons leave out the somewhat puzzling cell 3. As we noted in Section 2, this pattern of effects—sales rise and profits fall—is what one might expect of firms whose managers were size or growth maximizers. It is also what one would expect if the mergers led to a decrease in market power using the same logic that we employ to determine increases in market power. Even if one uses this logic to classify mergers in cell 3 as socially beneficial, however, the fraction of beneficial mergers in our study (44.2) falls short of the fraction that is harmful

21because they either increase market power or reduce efficiency.

21 Of course, some of the differences between actual and projected profits and sales that we record are small andeconomically insignificant. Thus, some of the mergers falling into each category might be judged to have resulted in small and insignificant increases in market power, etc. An alternative way to proceed would be to define an additional category—no significant difference—where significant difference is interpreted as an economically meaningful difference between the actual and projected values. We made such a classification using a one percentage point difference in profits relative to assets and a 10% difference in sales as our criteria for significant difference. Using these criteria, 3% of the mergers fell into the no difference category for both profits and sales, and 60% of all remaining mergers fell into the three main categories identified in Table 1, with the division among them remaining roughly equal—20% in each cell. Thus, one’s judgement as to therelative proportions of socially beneficial and harmful mergers is not affected by introducing an additional, no-difference category.

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 651

9 . Conclusions

We believe that the present study is the first to determine the effects of mergers on efficiency and market power by first separating mergers into those that increase profits and those that reduce them, and then examining the patterns of sales changes following the mergers. Most previous studies have judged the conse- quences of mergers by examining their average effects on either the profits or sales of the merging companies. As the previous section suggests, our results using these tests are broadly consistent with what others have found. We find that 56.7% of all mergers result in higher than projected profits, but almost the same fraction of mergers results in lower than projected sales after 5 years. Both mean differences are significantly different from zero. Thus, using profits as the measure of success would lead one to conclude that the average merger was a success, using sales one would reach the opposite conclusion. By basing our judgement of the welfare effects of mergers upon criteria that look at both the sales and profits changes following mergers, we have been able to resolve this ambiguity. We predict profit increases and sales declines for mergers that increase market power. More than a fourth of all mergers exhibit this pattern, and this helps to explain why mergers look more successful, when one examines post-merger profits than for post-merger sales. If one categorizes mergers that increase market power or that reduce efficiency as welfare reducing, then a majority of the mergers taking place around the world over the last 15 years appear to be welfare reducing.

Our study is the largest cross-national comparison of the effects of mergers to date. In this respect one of our most interesting findings is how similar the post-merger patterns of profit and sales changes look across the different countries. We also did not find significant differences between domestic and cross-border mergers. Although individual mergers can have quite different consequences in terms of efficiency and market power, their effects do not appear to depend on the country origins of the merging companies.

A cknowledgements

We would like to thank A. D. Cosh, M. Erlei, O. Fabel, P. A. Geroski, P. Guest, A. Hughes, D. Marin, S. Martin, H. Odagiri, A. Singh, K. Stahl and P. Zweifel for comments. Thanks are also due to participants in presentations at the Corporate Governance conference organized by the DIW, Berlin, November, 2000; at the 9th Annual WZB conference on Industrial Organization, Berlin, December 2000; and

¨ ¨ ¨at the ‘Ausschuß fur Industrieokonomik’ of the Verein fur Socialpolitik, Passau, March 2001. We also thank J. Jung for excellent research assistance. Financial

¨support of the ‘Jubilaumsfonds der Oesterreichischen Nationalbank’, Project 8861, is gratefully acknowledged.

652 K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653

R eferences

Amel, D.F., Rhoades, S.A., 1989. Empirical evidence on the motives for bank mergers. Eastern Economic Journal 15, 17–27.

Baldwin, J., 1995. In: The Dynamics of the Competitive Process. Cambridge University Press, Cambridge.

Baldwin, J., Gorecki, P., 1990. Mergers placed in the context of firm turnover. In: Bureau of the Census, Annual Research Conference, Proceedings, Washington, DC: US Department of Commerce, pp. 53–73.

Cable, J.R., Palfrey, J.P.R., Runge, J.W., 1980. Federal Republic of Germany, 1964–1974. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 99–132.

Cosh, A., Hughes, A., Singh, A., 1980. The causes and effects of takeovers in the United Kingdom: an empirical investigation for the late 1960s at the microeconomic level. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 227–270.

Goldberg, L.G., 1973. The effect of conglomerate mergers on competition. Journal of Law and Economics 16, 137–158.

Healy, P.M., Palepu, K.G., Ruback, R.S., 1992. Does corporate performance improve after mergers? Journal of Financial Economics 31, 135–175.

Hughes, A., 1989. The impact of merger: a survey of empirical evidence for the UK. In: Fairburn, J., John, K. (Eds.), Mergers and Merger Policy. Oxford University Press, Oxford, pp. 30–98.

Ikeda, K., Doi, N., 1983. The performance of merging firms in Japanese manufacturing industry: 1964–75. Journal of Industrial Economics 31, 257–266.

Jenny, F., Weber, A.P., 1980. France, 1962–72. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 133–162.

Kumps, A., Wtterwulghe, R., 1980. Belgium, 1962–74. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 67–97.

McDougall, F.M., Round, D.K., 1986. The Determinants and Effects of Corporate Takeovers in Australia, 1970–1981. Australian Institute of Management, Victoria.

Meeks, G., 1977. In: Disappointing Marriage: A Study of the Gains From Merger. Cambridge University Press, Cambridge.

Mueller, D.C., 1980a. The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA.

Mueller, D.C., 1980b. The United States, 1962–1972. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 271–298.

Mueller, D.C., 1985. Mergers and market share. Review of Economics and Statistics 67, 259–267. Mueller, D.C., 1997. Merger policy in the United States: a reconsideration. Review of Industrial

Organization 12, 655–685. Peer, H., 1980. The Netherlands, 1962–1973. In: Mueller, D.C. (Ed.), The Determinants and Effects of

Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 163– 191.

Ravenscraft, D.J., Scherer, F.M., 1987. Mergers Sell-offs and Economic Efficiency. The Brookings Institution, Washington, DC.

Ryden, B., Edberg, J.O., 1980. Large mergers in Sweden, 1962–1976. In: Mueller, D.C. (Ed.), The Determinants and Effects of Mergers: An International Comparison. Oelgeschlager, Gunn & Hain, Cambridge, MA, pp. 193–226.

Salant, S.W., Switzer, S., Reynolds, R.J., 1983. Losses from horizontal merger: the effects of an

K. Gugler et al. / Int. J. Ind. Organ. 21 (2003) 625–653 653

exogenous change in industry structure on Cournot-Nash equilibrium. Quarterly Journal of Economics 98, 185–199.

Scott, J.T., 1993. Purposive Diversification and Economic Performance. Cambridge University Press, Cambridge.

Suits, D.B., 1984. Dummy variables: mechanics v. interpretation. Review of Economics and Statistics 66, 177–178.