The Impact of the Economic
Cycle on the Valuation of
Mergers and Acquisitions
A Thesis submitted for the Degree of
Master of Science Finance
Name: Thanh Ha Minh Nguyen
Student number: S2852055
Contact information: t.h.m.nguyen.1@student.rug.nl
Study Program: Master of Science Finance
Date: 11th of January 2018
Supervisor: Dr. Peter Smid
Abstract
This paper investigates whether market participants value Mergers & Acquisitions differently, depending on the phase of the economic cycle (recession or non-recession). Data on M&A announcements in the United States of America and Europe within the years 2000 – 2016 are analyzed. Using the event study approach, this research provides evidence that acquirers’ cumulative average abnormal returns from M&A announcements are 1.59% to 3.39% lower, when the M&A are announced during recessions compared to non-recessions. Contrarily, targets’ cumulative average abnormal returns during recessions are 3.95% to 4.99% higher than during non-recessions. Furthermore, using regression analyses, this research investigates other variables that potentially have an impact on firms’ abnormal returns experienced from M&A announcements. The relative size of the acquisition shows a significant negative impact on targets’ and acquirers’ abnormal returns. The magnitude of the impact between relative size and abnormal returns is dependent upon the phase of the economic cycle.
Keywords: Economic Cycle, Recession, Mergers and Acquisitions, Valuation JEL Codes: G34, E32
1 Introduction
Mergers and Acquisitions (M&A) are transactions that combine two firms with each other as a consolidation or transfer one firm to the other through a takeover. This paper defines M&A as acquisitions and the terms are used interchangeably.
The very first M&A date back to the late 1800s (Roberts, Wallace, and Moles, 2016). Since then, the world has faced five considerable merger waves. Every wave was unique concerning the activity level and length but there is one very striking similarity that all waves had in common: the relation to external economic forces. External forces like downswings and upswings of the economy, regulatory changes, industrial progress, and technological innovations triggered each wave. For example, at the end of the First World War, an M&A wave was driven by an improvement of the antitrust legislation. Another wave emerged in 1993 when capital markets started to recover from a downturn (Roberts et al., 2016). Another significant common feature is a downturn of the financial markets and a possible recession of the economy at the end of each merger wave (Lambrecht, 2004 and Dieudonné, Cretin, and Bouacha, 2014). The past shows that M&A have a cyclical nature and a certain connection to countries’ economic conditions.
During the last decades, global M&A activities have increased significantly. Since the 1980s, the number of worldwide M&A multiplied by 17 and the value of total M&A increased tenfold (Institute for Mergers, Acquisitions and Alliances, 2017). Furthermore, strong globalization drives a notable trend concerning cross-border M&A, which are transactions beyond home state borders. Cross-border M&A accounted for 47% of the total deal value in the first quarter of 2017 compared to 39% in 2015 (Baker & McKenzie, 2016, 2017).
According to Thomson, Dettmar, and Garay (2016), the number one obstacle to the success of M&A is economic uncertainty, which is mainly defined as issues of recessions. Thomson et al. (2016) survey M&A executives and 19% of all respondents considered economic certainty to be the most important factor in achieving a successful acquisition. Hence, the market’s opinion on the success of M&A is linked to the current phase of the economy. Wann and Lamb (2017, p.144) state that:
Traditional M&A studies, however, generally overlooked the link between M&A announcements and the underlying economic conditions that are present at the announcement.
The main research question of this paper is whether market participants value M&A differently during different phases of the economic cycle. In addition, the paper investigates the following research questions:
During which stage of the economic cycle is the M&A activity level higher?
Which other possible factors influence the valuation of M&A besides the economic cycle?
To answer the research questions, the abnormal returns resulting from M&A announcements are calculated for acquirers and targets by performing an event study. The event study uses stock price data of American and European companies involved in M&A from the year 2000 to 2016. The abnormal returns during recessions are compared to the ones during non-recessions, to investigate how valuations of M&A announcements vary due to the phases of the economic cycle. Thereafter, a regression analysis is performed using the empirical results from the event study to find other variables, which potentially influence the abnormal returns during recessions and non-recessions. Finally, the paper determines whether the economic cycle affects the potential influences of these variables on abnormal returns.
Section 2 presents the reviewed literature and the developed hypotheses for the research. Section 3 describes the data and research methodology. The empirical results and conclusion are presented in Section 4 and 5, respectively.
2 Literature review
2.1 M&A across the economic cycle
Traditional M&A studies do not take into account the economic conditions at the time of the M&A announcements. However, some theoretical studies show that the economic cycle is related to M&A activities and valuations. One of the theories about the influence of the economic cycle on the activity level of M&A is the Recession Cleansing Hypothesis by Schumpeter (1942). It states that recessions reveal unhealthy and unsustainable firms more easily and force them to exit the market leaving fewer firms in the market (Ding and Rahaman, 2010). As a consequence, the M&A activity level decreases during recessions due to the simple fact that there are fewer firms to acquire. The Economic Disturbance Theory, presented by Gort (1969), claims that economic shocks causes owners of firms and non-owner investors to value firms differently, which could lead to acquisitions. During economic shocks, investors cannot rely on past information to predict the future. Non-owner investors value firms higher than owners during positive economic shocks and lower during negative economic shocks. M&A are taking place if non-owners value firms higher than the owners. On the other hand, no M&A take place if the non-owners value the firms lower than the owners. The conclusion is that during a period of positive economic shocks, M&A activities are higher than during negative economic times. Another theory states that economic crises lead to liquidity and resources constraints, which consequently leads to low investment and less M&A activity (Aguiar and Gopinath, 2005).
(non-recession) due to the asymmetric steepness of the function. Based on this theory, the valuation of M&A during recessions is higher than the one during non-recessions.
2.2 Efficient market hypothesis
The Efficient Market Hypothesis (EMH) was first established in the 1960s by the pioneer Eugene Fama. The hypothesis states that markets are efficient which means that prices in the markets fully reflect all available information at all times (Fama, 1970). Based on this condition, he hypothesizes that stock prices underlie a random walk movement and cannot be predicted based on past information and price movements. In efficient capital markets, security prices change due to the arrival of new and relevant information. The EMH makes it possible to assess the market participants’ reactions to M&A announcements by determining the change of firms’ stock prices associated with the announcement of M&A. Fama (1970) presents three forms of efficient markets. The weak-form includes only historical price information, the semi-strong form takes publicly available information into account as well and the third form includes all previous information plus information that is only available to a monopolistic group. Since the development of the hypothesis, many studies criticize the existence of market efficiency due to its unrealistic underlying assumptions (Titan, 2015). That is why, Fama (1991) states that the strong form of efficiency is surely non-existent (Fama, 1991) but the less restrictive weak form persists if the assumptions apply to a sufficient number of investors (Fama, 1970).
2.3 Empirical results of M&A during recessions and non-recessions
Many empirical studies research the effect of M&A on firms’ stock prices. Among these studies, Bruner (2002), Goergen and Renneboog (2004), Mulherin and Boone (2000), and Servaes (1991) find that M&A announcements result in significantly positive abnormal returns for the target companies. Contrarily, the abnormal returns for acquiring companies are around zero and insignificant.
Wann and Lamb (2016, 2017) empirically investigate the direct effect of the phase of the economic cycle on the valuation of M&A. The goal of both studies is to find evidence for different reactions of market participants to M&A announcements during recessions and non-recessions. The first empirical evidence of an existing link between M&A and the economic cycle is the significantly higher M&A activity level during non-recessions compared to recessions (Wann and Lamb, 2016 and Cools, Gell, Kengelbach, and Roos, 2007).
how the market participants react to the news in the different phases of the business cycle. They deliver evidence that market participants react differently to M&A announcements depending on the economic conditions. Target firms experience a high positive cumulative average abnormal return (CAAR) during both phases of the economic cycle. However, the CAARs during recessions are significantly higher than during non-recessions (Wann and Lamb, 2017), which leads to the conclusion that “good news in bad times is stronger than good news in good times” (Wann and Lamb, 2016, p. 68). Market participants value M&A higher during recessions than otherwise from the perspective of the target firm. The results concerning acquiring firms are different. Their CAARs around announcements are very close to zero, significantly positive during recessions but insignificantly positive during non-recessions. Furthermore, the CAARs of M&A announcements during recessions are not significantly different from the ones during non-recessions. This result points out that the economic cycle does not influence the valuation of M&A for acquirers (Wann and Lamb, 2017).
2.4 Variables related to the abnormal returns of acquirers and targets
According to Wann and Lamb (2017), the economic cycle does not affect the CAARs of acquirers, however there are other variables that do. The ARs of targets are influenced by the economic cycle, as well as other variables. The following section presents four possible influential variables.
2.4.1 Relative size
Several studies state that the ARs of M&A announced are directly related to the relative size of the acquisition. The relative size of an acquisition is defined as the ratio of the market value (MV) of the target’s equity divided by the MV of the acquirer’s equity (Mulherin and Boone, 2000). In theory, several authors claim different consequences due to the size distribution of target and acquirer. Cohen and Levinthal (1990) and Krishanan, Hitt, and Park (2007) argue M&A are most likely to be successful if the target and acquirer are the same size. In that case, it is easier for the acquirer to value the target’s know-how and skills for an efficient business merger (Cohen and Levinthal 1990). Furthermore, it is easier to identify discrepancies and redundancies between the two companies (Krishanan et al., 2007). Other studies suggest that a difference in size drives positive abnormal returns and success of the acquisitions. If an acquirer is smaller than the target, the acquirer benefits from the gained market power and economies of scales (Seth, 1990). In contrast, Homberg, Rost, and Osterloh (2009) state that the acquirer can only experience synergies if it is larger than the target.
significantly negative impact on acquirers’ returns. Hence, acquirers and targets can only experience higher ARs if the relative size of the acquisition is small.
2.4.2 Industry effect
Wann and Lamb (2016) suggest that an industry effect influences the abnormal returns of firms. In essence, ARs of companies that operate in one industry are significantly different from the ARs of companies in another industry. Industry convergence has become a main subject concerning M&A. Certain industries and sectors are very popular and attractive to acquirers. According to Thomson et al. (2016), the technology sector is one of the top sectors because it promises the acquirer access to a fast growing and highly profitable market. The acquisition of technology firms is one of the most important strategic drivers for M&A nowadays. The top sectors in 2017, based on number and volume of deals, are automotive and transportation, consumer products and retail, mining and metals, oil and gas, power and utilities and telecommunications according to Ernst & Young (2017). They suggest that an announcement of an acquisition in these sectors results in a positive feedback from market participants. In this paper, the industries Mining, TCEGS, Wholesale Trade and Retail Trade (see Table 2 in Section 3.1) are expected to achieve significantly higher abnormal returns than the other industries.
Cooper, Dimitrov, and Rau (2001) show that during the time of the Dot-com bubble from 1997 to 2001, a mere association of a company with the Internet sector led to a permanent increase of the firm’s value. This even held for companies that did not operate in the Internet sector. This shows that market participant’s valuation of firms is influenced by their industry affiliation.
2.4.3 Sub-deal type
Empirical literature on abnormal returns of cross-border and domestic M&A shows mixed results. On the one hand, Kang (1993) finds that cross-border acquisitions generate higher returns than domestic acquisitions for both acquirers and targets. On the other hand, Moeller and Schlingemann (2005) find statistical evidence that cross-border M&A experience lower announcement abnormal returns than its counterpart for acquirers. According to Goergen and Renneboog (2004), there is no significant difference for neither acquirers nor targets.
2.4.4 Method of payment
Andrade, Mitchell, and Stafford (2001) explain that M&A paid with stocks represents two synchronous business transactions for acquirers. The first one is the acquisition itself and the second one is the equity issuance because the company must issue new stocks to be able to use them as a payment instrument. According to the Signal Theory by Leland and Pyle (1977), market participants view an equity issuance negatively because they expect a company to only issue stocks when the management believes that the stocks are overvalued. This initiates a decrease in security prices. Contrarily, a payment in cash or debt, signals good health of a company and hence leads to an increase in prices. For a target company, a payment with stocks also signals overvaluation of the acquirer’s stocks and is therefore valued more negatively than a non-stock payment (Andrade et al., 2001).
There is empirical evidence that the payment type has an impact on the abnormal returns of acquirers and targets. Moeller and Schlingeman (2005) and Travlos (1987) find that a stock-financed acquisition has a significantly negative impact on the acquirers’ abnormal returns compared to a non-stock financed acquisition. Servaes (1991) measures higher positive abnormal returns for targets if there is a non-stock financed payment, but finds no measurable significance.
2.5 Qualitative hypotheses
The following qualitative hypotheses are developed and tested primarily based on the gained insights from the existing literature presented in Sections 2.1 – 2.4.
Effect of M&A announcements on acquirers’ and targets’ abnormal returns:
H1: There are no significant CAARs for acquiring firms in the short-term due to the
announcement of M&A.
H2: There are significantly positive CAARs for target firms in the short-term due to the
announcement of M&A.
Differences of CAARs for acquirers and targets experienced during different phases of the economic cycle:
H3: The CAARs of acquiring firms, which are involved in M&A during recessions, are
H4: The CAARs of target firms, which are involved in M&A during recessions, are
significantly higher than during non-recessions.
The effect of the relative size, which is measured by the ratio of the market value of the target’s equity over the market value of the acquirer’s equity, on acquirers’ and targets’ abnormal returns:
H5: There is a significantly negative relation between the acquirers’ CARs and the
relative size of the acquisition.
H6: There is a significantly negative relation between the targets’ CARs and the relative
size of the acquisition.
The effect of the firm’s industry affiliation, which is defined according to the Standard Industrial Classification (SIC), on acquirers’ and targets’ abnormal returns:
H7: The CARs of acquiring firms depend on the firms’ industry affiliation.
H8: The CARs of target firms depend on the firms’ industry affiliation.
The effect of the sub-deal type of the acquisition, which is defined as either cross-border or domestic, on acquirers’ and targets’ abnormal returns:
H9: The CARs of acquiring firms depend on whether the acquirer is involved in a
cross-border or domestic M&A.
H10: The CARs of target firms depend on whether the target is involved in a cross-border
or domestic M&A.
The effect of the payment method of an acquisition on acquirers’ and targets’ abnormal returns:
H11: The CARs of acquiring firms are significantly negative if they pay with stocks
instead of non-stock means like cash, debts, and others.
H12: There is no significant effect on the targets’ CARs if the acquirer pays with stocks
3 Data collection and methodology
3.1 Data collection
In order to test the developed hypotheses, this research utilizes a data sample that consists of acquisition deals that were executed in the United States of America (US), covering the time period from the 1st of January 2000 until the 31st of December 2016. Information regarding the deals is drawn from the M&A database Zephyr - Bureau van Dijk.
The following search strategy is applied to collect data for the sample:
1. Deal type: 100%-stake acquisitions. Other transactions like Management-Buy-in/Management-Buy-out, mergers, repurchases, several bids, and partial acquisitions are excluded due to their complexity.
2. Current deal status: only completed deals are included. Uncompleted or rumored deals are not taken into consideration because their uncertain status.
3. The following companies are considered: listed acquirers; listed and delisted (after an acquisition) targets. The acquirers need to be listed in order to find their security price data. Delisted targets are suitable for this research due to the large amount of available historical security data around the M&A announcement. Listed targets with their own exchange ticker symbol or ISIN number after the M&A are also appropriate.
4. Regions: deals include companies located in the US (acquirer or target) and companies located in Europe (acquirer or target) that are involved in M&A with a US company. Consequently, domestic deals are M&A within the US while cross-border deals are between US and Europe.
Cross-border transactions between the USA and Europe make up the largest amount of deals. The US is the major bidder in Europe and vice versa. In the first three quarters of 2015, Europe accounted for 45% of all inbound deals in the USA and 54% of all outbound deals (Baker & McKenzie, 2017). In this research, an inbound deal indicates that a European company acquires a US company and an outbound deal indicates an acquisition of a company in Europe by a US company.
in the event study [see Table 1, Column (B)]. Table 1, Column (C) shows the data sample for the regression analysis. Only deals with complete price data of acquirer and target observations can be used. In the end, 351 deals (out of 563) fulfill the data requirements and can be used in the regression analysis.
Table 1
Data sample of deals drawn from the database Zephyr.
(A) Original Sample (B) Event Study Sample (C) Regression Analysis Sample Observations of Acquirers 563 526 351 Observations of Targets 563 388 351 Total Observations 1,126 914 702 Deals 563 563 351
Source: Zephyr M&A Database by Bureau van Dijk.
This research makes use of two separate sub-samples (B) and (C) from the original sample (A). Table 2A and 2B show the observations of acquirers and targets, respectively, and their descriptive statistics. The %-columns in Table 2A and 2B are based on the total number of observations from the several samples presented in Table 1. There are no remarkable differences in the numbers and percentages of observations within the four aspects (economic condition, payment methods, industry, and sub-deal type) between the three samples. This means that sub-samples (B) and (C) are appropriate representations of the original sample (A).
Table 2A
Descriptive statistics of the observations of acquirers across the different samples.
(A) Original Sample (B) Event Study Sample (C) Regression Analysis Sample
Amount % Amount % Amount %
Economic Conditions
Obs. in Recessions 68 12.1% 65 12.4% 37 10.5%
Obs. in Non-Recessions 495 87.9% 461 87.6% 314 89.5%
Payment Method
Stock-financed payment 162 28.8% 150 28.5% 99 28.2%
Non-stock financed payment 401 71.2% 376 71.5% 252 71.8%
comprising of: Cash 307 76.6% 285 75.8% 203 80.6%
Debt 64 16.0% 63 16.8% 40 15.9%
Others 30 7.5% 28 7.4% 9 3.6%
Industry (according to SIC)1
AFF2 0 0.0% 0 0.0% 0 0.0% Mining 28 5.0% 26 4.9% 17 4.8% Construction 4 0.7% 4 0.8% 2 0.6% Manufacturing 201 35.7% 187 35.6% 128 36.5% TCEGS3 40 7.1% 37 7.0% 21 6.0% Wholesale Trade 13 2.3% 13 2.5% 7 2.0% Retail Trade 18 3.2% 17 3.2% 14 4.0% FIR4 165 29.3% 155 29.5% 106 30.2% Services 94 16.7% 87 16.5% 56 16.0% Sub-Deal Type Domestic M&A 498 88.5% 465 88.4% 316 90.0% Cross-border M&A5 65 11.5% 61 11.6% 35 10.0%
c. of: US acquirer/EU target 13 20.0% 13 21.3% 6 17.1%
US target/EU acquirer 52 80.0% 48 78.7% 29 82.9%
Source: Zephyr M&A Database by Bureau van Dijk (2017) and North American Industry Classification System (2017).
Note: Obs. = Observations. 1 The four-digit SIC codes are used; 2 AFF is Agriculture, Forestry and Fishing; 3 TCEGS is Transportation,
Communications, Electric, Gas and Sanitary Services; 4 FIR is Finance, Insurance and Real Estate; 5 EU countries that are included
Table 2B
Descriptive statistics of the observations of targets across the different samples.
(A) Original Sample (B) Event Study Sample (C) Regression Analysis Sample
Amount % Amount % Amount %
Economic Conditions
Obs. in Recessions 68 12.1% 40 10.3% 37 10.5%
Obs. in Non-Recessions 495 87.9% 348 89.7% 314 89.5%
Payment Method
Stock-financed payment 162 28.8% 111 28.6% 99 28.2%
Non-stock financed payment 401 71.2% 277 71.4% 252 71.8%
comprising of: Cash 307 76.6% 225 81.2% 203 80.6%
Debt 64 16.0% 41 14.8% 40 15.9%
Others 30 7.5% 11 4.0% 9 3.6%
Industry (according to SIC)1
AFF2 2 0.4% 1 0.3% 1 0.3% Mining 30 5.3% 19 4.9% 17 4.8% Construction 5 0.9% 3 0.8% 3 0.9% Manufacturing 170 30.2% 115 29.6% 107 30.5% TCEGS3 41 7.3% 25 6.4% 22 6.3% Wholesale Trade 16 2.8% 12 3.1% 12 3.4% Retail Trade 14 2.5% 12 3.1% 11 3.1% FIR4 152 27.0% 108 27.8% 98 27.9% Services 133 23.6% 93 24.0% 80 22.8% Sub-Deal Type Domestic M&A 498 88.5% 349 89.9% 316 90.0% Cross-border M&A5 65 11.5% 39 10.1% 35 10.0%
c. of: US acquirer/EU target 13 20.0% 6 15.4% 6 17.1%
US target/EU acquirer 52 80.0% 33 84.6% 29 82.9%
Source: Zephyr M&A Database by Bureau van Dijk (2017) and North American Industry Classification System (2017).
Note: Obs. = Observations. 1 The four-digit SIC codes are used; 2 AFF is Agriculture, Forestry and Fishing; 3 TCEGS is Transportation,
Communications, Electric, Gas and Sanitary Services; 4 FIR is Finance, Insurance and Real Estate; 5 EU countries that are included
(according to ISO 3166-1 alpha-2 code): BE, CH, DE, DK, ES, FI, GB, IT, LU, NL, NO, and SE.
3.2 Event study methodology
associated M&A. This research determines whether these reactions are significantly different during recessions compared to non-recessions. By implication, the business cycle has an effect on the value of the firm if a significant difference is measureable.
To utilize this approach, the event must be carefully chosen and defined. M&A influence the financial markets long before their execution. Information leakages occur from rumors and announcements of M&A. According to Dodd (1980), the highest level of information spill-out occurs on the announcement day. Therefore, the announcement of the acquisition is defined as the event and set as day zero. Fig. 1 illustrates the time frame for the event study.
Fig. 1. Event study time frame. The different time windows of the event study are visualized and defined by the
start and end time t.
The event window lies around the event day to capture possible announcement effects. It is determined to be 11 days, starting five days before, ending five days after the event day and including the event day itself. It is common to use a one- to 11-day event window (Brown and Warner, 1985).
Information leakages (rumors) are likely to occur before the M&A announcement are made, which could cause price effects. Additionally, there are possible delayed effects that occur after the event. These are the reasons for the usage of an extended event window. A possible cause for delayed effects is an inefficient market. The price effects of the M&A announcements consequently occurs after the event day (Arnold, 2008). This paper defines sub-event windows to separately measure leakage, delayed and event effects. Table 3 presents the sub-event windows and their purposes.
The estimation window consists of 120 days as MacKinlay (1997) recommends, starting on day -129 and ending on day -10. The longer the estimation window, the more accurately the normal returns can be estimated (MacKinlay, 1997). However, there is evidence that the accuracy of normal returns is not sensitive to the length as long as the estimation window includes more than 100 days (Armitage, 1995 and Park, 2004). A buffer of four days lies between the two windows to prevent from overlapping.
-10 -5 0 +5
Event Window (11 days) Buffer (4
days) Estimation Window (120 days)
-129
Table 3
Sub-event windows used in the event study approach.
The sub-event windows are defined as (start time t1 / end time t2).
Sub-event windows Description
(-5/5) Entire event window
(-5/-3) Leakage effect window (short)
(-5/-1) Leakage effect window (long)
(-1/0) Early event-day window
(0/1) Late event-day window
(-1/1) Extended 3-day event window
(-2/2) Extended 4-day event window
(1/5) Delayed effect window (long)
(3/5) Delayed effect window (short)
3.2.1 Normal return model
Daily stock- and index returns are calculated using the linear return method due to its straightforwardness. There is evidence that there is no difference in results using the linear or continuously compounded return method (Henderson, 1990). The linear return method is calculated as
where represents the adjusted closing price (if available, otherwise it is the closing price) at time t and is the adjusted closing price (if available, otherwise it is the closing price) one day before t.
Several models are available to estimate normal returns for the event study. Normal returns (also called expected returns) are returns that could have been earned given no event (Cable and Holland, 1991). The one-factor Market Model (MM) that MacKinlay (1997) recommends is chosen in this research. Studies like Henderson (1990), MacKinlay (1997), and Ahern (2009) find that the use of multifactor models (among them e.g. the Fama-French 3-factor model) over the simpler one-factor MM does not necessarily lead to improved forecasts of normal returns because the marginal explanatory power is low. The Capital Asset Pricing Model (CAPM) is inferior to the MM because it imposes too many restrictions with low validity (MacKinlay, 1997). The MM relates the return of a security to the return of a market portfolio:
where is the return of security i in time t , t = -129 to -10, is the constant term, is the
beta and is the error term of security i, is the return of market portfolio m in t.
The market portfolio is represented by the market index S&P 500 for US stocks and Stoxx Europe 600 for European stocks. Both indices are suitable due to the broadness and representativeness of their respective regions.
To measure whether the M&A announcements have a price effect on the firms’ stocks within the event window, the AR is calculated for observation i in time t which is defined as the
residual of the actual return subtracted by the normal return:
where t = -5 to +5, is the estimated constant term, is the estimated beta.
The cumulative abnormal return (CAR) is the aggregation of ARs over time. Eq. (4) expresses the CAR:
where t1 is the start and t2 is the end day of the investigated time window.
The AR cannot be used to make overall statistical inferences about the M&A announcement effect because it just takes one observation into account. Therefore, it is essential to aggregate the abnormal returns cross-sectional and through time. The average abnormal return (AAR) is the cross-sectional aggregation of ARs in time t, and the cumulative average abnormal return
(CAAR) is the aggregation of AARs through time t1 to t2. The following equations express
AAR and CAAR:
3.2.2 Significance test
Parametric tests make the stringent assumption of normally-distributed data. They are inappropriate if the sample does not fulfill this assumption (Robson, 2006). Fama (1976) and Brown and Warner (1985) find that the distribution of daily stock returns is not normal. However, the Central Limit Theorem enables the usage of parametric tests if the sample number N is large enough, even in case of non-normality (DasGupta, 2010). The Central Limit Theorem states that the distribution of the sample means converge approximately to normality, if the sample includes a large amount of observations (DasGupta, 2010). A sample is considered large if it contains more than 30 observations (Cohen and Levinthal, 1990). Since the Event Study Sample (B) in Table 1 includes more than the suggested number of observations, a parametric test can be used. Furthermore, a non-parametric test is used as a robustness test. Non-parametric tests do not have strict assumptions about the distribution of the data set (Robson, 2006) and therefore are appropriate for research on daily stock returns. The chosen parametric test is the Cross-Sectional T-Test (CST) by Brown and Warner (1985). It is appropriate for hypothesis testing that involves several cross-sectional observations. Under the null hypothesis, the CAAR is equal to zero. The estimator of the variance is based on the ARs of all cross-sections. Brown and Warner (1980) show that the CST is robust to an event induced volatility increase, which is likely to appear in events like M&A. Eq. (7) expresses the test statistic for the CST on the CAAR and (8) is presenting the variance estimator:
where and are the estimator of standard deviation and variance, respectively.
Rank. Secondly, the Rank loses its power very quickly as the length of the event window increases. It is only powerful for one-day event windows. Thirdly, the Gsign is more robust against outliers in ARs relative to the Rank. The Gsign test statistic is calculated as
where w is the number of companies with positive CARs in the event window and is proportion of positive ARs in the estimation window.
Eq. (10) shows that is required to calculate the Gsign test statistic in Eq. (9):
where t = -129 to -10, L = 120 (days in estimation window) and .
This research compares the abnormal returns due to M&A announcements during recessions and non-recessions of acquirers and targets, to find out whether market participants’ reactions are different during the two phases. An Independent T-Test with unequal variances (TT) is used to test the significance of that difference. It tests whether the means (CAARs) of two independent samples are significantly different from each other, taking into account different samples sizes and unequal variances (Keller, 2012). Under the null hypothesis, the means are not significantly different from each other. The Independent T-Test is calculated as
where and are the variances of the abnormal returns of observations involved in M&A during recessions and non-recessions, respectively, NR and NN are the observations in
recessions and non-recessions, respectively.
3.3 Regression analysis
14 days before the announcement, as Servaes (1991) recommends. MVs of European companies that do not denote stock prices in American Dollars (USD) are exchanged to USD using daily historical exchange rates1. According to the existing literature, the acquirers’ and targets’ ARs decrease when the relative size increases.
Secondly, the regression includes industry dummy variables. The SIC defines ten industries (see Table 2 in Section 3.1). The three industries (Agriculture, Forestry and Fishing, Mining, and Construction) are involved in the production and handling of natural resources and the production of establishments with mostly natural resources (Executive Office of the President, 2017). Therefore, they are aggregated into one classification (called D1) in this research. Furthermore, the industries Wholesale and Retail Trade include companies that sell merchandise and goods (Executive Office of the President, 2017). These two industry classifications are also combined into one common classification (called D2). These aggregations also have a statistical reason: the above-mentioned industries do not have many observations, which is why an aggregation ensures more precise statistical results. The industry Services is omitted in the regressions.
Thirdly, the deal-sub type variable shows whether a firm is involved in cross-border or domestic acquisitions. According to the literature, the sub-deal type has an impact on the ARs of targets and acquirers.
Fourthly, the payment method of a deal influences the abnormal returns of firms according to the Signal Theory (Leland and Pyle, 1997). It is separated into stock-financed payments and non-stock financed payments, which includes cash, debt, and others.
Eq. (12) expresses the regression. The regression equation is developed for first, acquirers in non-recession, second, acquirers in recession, third, targets in non-recession and fourth, target in recession:
where
- is the cumulative abnormal returns of a firm i in time t
- is the unknown constant term
- is the estimated coefficient of the associated independent variable - is the ratio
- D1 is the dummy variable for the industries Agriculture, Forestry and Fishing, Mining, and Construction. It has the value of 1 if the entity has a SIC code between 0100 -1799, otherwise 0.
- MA is dummy variable for the industry Manufacturing. It has the value of 1 if the entity has a SIC code between 2000-3999, otherwise 0.
- TCEGS is dummy variable for the industry Transportation, Communications, Electric, Gas and Sanitary Services. It has the value of 1 if the entity has a SIC code between 4000-4999, otherwise 0.
- D2 is dummy variable for the industries Wholesale- and Retail Trade. It has the value of 1 if the entity has a SIC code between 5000 -5999, otherwise 0.
- FIR is dummy variable for the industry Finance, Insurance and Real Estate. It has the value of 1 if the entity has a SIC code between 6000-6799, otherwise 0.
- If all industry dummy variables have the value of 0, then the industry is Services by default which has the SIC code between 7000-8999.
- SUB is dummy variable for the subtype of the deal. It has the value of 1 if the entity is involved in a domestic M&A and 0 in a cross-border M&A.
4 Empirical results
4.1 Analysis of the abnormal returns of the entire Event Study Sample (B)
Sample (B) (Table 1, Section 3.1) is used for the event study. Abnormal returns are calculated for all 914 observations. Afterwards, significance tests are conducted to analyze whether the abnormal returns from M&A announcements are significant. The empirical results for acquirers are presented, followed by the results for targets.
4.1.1 All observations of acquirers
The AARs of the acquirers in the entire event window change frequently from positive to negative (see Table 4). Six out of 11 days show negative values. Fig. A1 in Appendix A displays no visually noteworthy peaks of the AAR line.
Table 4
(Cumulative) average abnormal returns of all observations of acquirers (N = 526).
This table reports the AARs and CAARs, standard deviations and the test statistics of the CST and Gsign, which are calculated for 526 observations of acquirers in sample (B) on each day t in the event window.
t AAR (%) Std. Dev. (%) CST (AAR) Gsign (AAR) CAAR (%) Std. Dev. (%) CST (CAAR) Gsign (CAAR) -5 0.127 2.585 1.124 -0.701 0.127 2.585 1.124 -0.701 -4 0.086 2.214 0.892 0.171 0.213 3.371 1.448 -0.178 -3 -0.112 2.063 -1.249 -1.922** 0.101 4.246 0.543 -1.137 -2 0.092 3.092 0.683 0.432 0.193 5.670 0.779 -0.352 -1 -0.124 3.457 -0.821 -1.661** 0.069 6.569 0.240 -1.573 0 -0.292 5.454 -1.226 -2.010** -0.223 8.491 -0.602 -2.446*** 1 0.017 3.748 0.104 0.694 -0.206 8.521 -0.554 -1.835** 2 -0.015 2.705 -0.130 0.956 -0.221 8.714 -0.582 -1.399 3 0.133 2.407 1.268 0.869 -0.088 8.904 -0.227 -2.010** 4 -0.283 2.446 -2.658*** -2.010** -0.371 9.454 -0.901 -1.573 5 -0.030 2.727 -0.251 0.171 -0.401 10.485 -0.878 -2.097**
Note: *** and ** denote 1% and 5% significance level, respectively.
The CAARs in Table 4 are calculated to make inferences about the overall abnormal performance across observations and time. Following the announcement day, the CAARs of the acquirers become negative. The Gsign show significantly negative CAARs on the announcement day and at days 1, 3, and 5. This indicates that the M&A announcements trigger a negative reaction from the market participants. In contrast, none of the CAARs are significant according to the CST (CAAR). Due to the fact that the cross-sectional abnormal returns of each day t within the event window are not normally-distributed, the results of the
non-parametric Gsign are more reliable in comparison to the results of the parametric CST (see Table B1, in Appendix B for the result of the Jarque-Bera normality test).
Table 5
Cumulative average abnormal returns of all observations of acquirers across various event windows (N = 526). This table reports the CAARs, standard deviations and the test statistics of the CST and Gsign, which are calculated for 526 observations of acquirers in sample (B) across various sub-event windows.
Window CAAR (%) Std. Dev. (%) CST (CAAR) Gsign (CAAR) (-5/5) -0.401 10.485 -0.878 -2.097** (-5/-3) 0.100 4.246 0.543 -1.137 (-5/-1) 0.069 6.569 0.240 -1.573 (-1/0) -0.415 5.521 -1.726 -2.358*** (0/1) -0.275 6.075 -1.037 -1.661** (-1/1) -0.398 6.215 -1.470 -1.922** (-2/2) -0.322 7.196 -1.025 -0.701 (1/5) -0.178 7.577 -0.540 0.345 (3/5) -0.180 4.796 -0.862 0.345
Note: *** and ** denote 1% and 5% significance level, respectively.
In addition, the CAARs are calculated for the sub-event windows to analyze the important periods within the event window (see Table 5). The CST does not find any event windows with significant CAARs for acquirers. In contrast, the Gsign finds significance at the 1% level for the early event-day window (-1/0). It also finds evidence at the 5% significance level for the entire event window (-5/5), late event-day window (0/1) and extended 3-day event window (-1/1). These event windows exhibit significantly negative CAARs.
The qualitative hypothesis H1 (presented in Section 2.5) states that acquirers do not
CST finds enough evidence to reject the null hypothesis of zero AAR at 1% significance level on day -4 (see Table 4). Furthermore, the Gsign finds enough evidence to reject the null hypothesis of zero CAARs in favor of the alternative hypothesis that the CAARs are significantly different from zero for the days 0, 1, 3, and 5 as well as for the sub-event windows (-5/5), (-1/0), (0/1), and (-1/1). These results are in line with Servaes (1991), who finds significantly negative abnormal returns for acquirers. All in all, the announcements of M&A seem to have a negative impact on acquirers’ (C)AARs. Significantly negative CAARs are found around the event day, indicating that market participants value announcements of M&A as bad news for the acquirers.
4.1.2 All observations of targets
The averaged abnormal return of targets peaks at 21.001% on the announcement day and is highly significant according to both CST and Gsign (see Table 6). The following day also exhibits a significantly positive AAR of 7.385%. The pre-announcement period is positive for all five days. The Gsign finds significant returns on days -4, -2, and -1, indicating positive leakage effects. Except for day 3, the returns in the post-event period from day 2 are negative.
Table 6
(Cumulative) average abnormal returns for all observations of targets (N = 388).
This table reports the AARs and CAARs, standard deviations and the test statistics of the CST and Gsign, which are calculated for 388 observations of targets in sample (B) on each day t in the event window.
t AAR (%) Std. Dev. (%) CST (AAR) Gsign (AAR) CAAR (%) Std. Dev. (%) CST (CAAR) Gsign (CAAR) -5 0.431 7.100 1.196 -0.029 0.431 7.100 1.196 -0.029 -4 0.124 4.073 0.598 1.904** 0.555 8.112 1.347 1.599 -3 0.214 6.219 0.677 0.378 0.769 10.533 1.437 2.210** -2 0.091 5.416 0.333 1.904** 0.860 9.856 1.719** 2.006** -1 0.367 4.581 1.577 1.803** 1.227 10.244 2.359*** 2.820*** 0 21.001 30.343 13.633*** 12.385*** 22.228 33.236 13.174*** 13.301*** 1 7.385 27.161 5.356*** 4.143*** 29.613 38.440 15.175*** 15.641*** 2 -0.067 3.656 -0.361 -1.758** 29.546 38.654 15.056*** 15.641*** 3 0.004 2.612 0.034 -0.639 29.550 38.781 15.009*** 15.234*** 4 -0.011 4.565 -0.048 0.683 29.539 39.291 14.809*** 15.132*** 5 -0.122 3.978 -0.604 -0.131 29.417 38.770 14.946*** 15.438***
On day 2, the Gsign finds a significantly negative AAR at the 5% level, indicating negative delayed effects. During the post-event period, the change of AAR from positive to slightly below zero shows that stock prices remain at a high level after the event. Figure A2 in Appendix A illustrates graphically that the AAR exhibits a temporary striking positive increase on day 0 and 1. Afterwards, the AAR decreases to a low level similar to the pre-event window. This suggests that the stock price increases on days 0 and 1 are caused by M&A announcements. Consequently, the market participants consider M&A announcements as good news for targets even shortly after the announcement, until day 2, where the positive price effect vanishes.
The CAARs, presented in Table 6, are positive for every day within the event window. The values of the CAARs are around 0.431% to 1.227% in the pre-event window and experience an extreme increase to the maximum of 29.613% on the day after the event. Afterwards, the CAAR settles down to 29.417% on the last day of the event window. The constant high CAARs suggest no drastic downward price corrections after the event. The CAARs show high significance for eight and nine subsequent days according to the CST and Gsign, respectively.
In Table 7, the CST and Gsign show significantly positive CAARs in seven and eight sub-event windows, respectively. The CAARs of the entire sub-event window (-5/5) and the windows including the event day [(-1/0), (0/1), (-1/1) and (-2/2)] show highly significant double-digit returns. Notably, the entire event window has the highest CAAR of 29.417%. The window (3/5) is the only insignificant one according to the CST and Gsign.
Table 7
Cumulative average abnormal returns of all observations of targets across various event windows (N = 388). This table reports the CAARs, standard deviations and the test statistics of the CST and Gsign, which are calculated for 388 observations of targets in sample (B) across various sub-event windows.
Window CAAR (%) Std. Dev. (%) CST (CAAR) Gsign (CAAR) (-5/5) 29.417 38.770 14.946*** 15.438*** (-5/-3) 0.769 10.533 1.437 2.210** (-5/-1) 1.227 10.244 2.359*** 2.820*** (-1/0) 21.368 30.573 13.767*** 12.894*** (0/1) 28.386 36.586 15.283*** 15.641*** (-1/1) 28.753 36.281 15.610*** 15.641*** (-2/2) 28.777 35.728 15.866*** 15.539*** (1/5) 7.189 27.116 5.223*** 3.329*** (3/5) -0.129 5.514 -0.460 1.192
The results for the targets agree with the existing literature (see Bruner, 2002, Goergen and Renneboog, 2004, Mulherin and Boone, 2002, and Servaes, 1991). This research finds highly significant and positive (cumulative) averaged abnormal returns on and around the announcement day. The CST and Gsign obtain predominantly consistent results for the returns of single days and sub-event windows. Therefore, there is enough evidence to reject the null hypothesis of zero (C)AARs for targets due to announcements of M&A. Furthermore, the qualitative hypothesis H2 (in Section 2.5), which states that the CAARs of
targets are significantly positive due to M&A announcements, can be accepted. In general, the announcements of M&A are good news for targets because they experience a positive and highly significant increase in their abnormal returns.
4.2 Analysis of the economic cycle’s impact on the abnormal returns due to M&A announcements
This section presents the results of the abnormal returns obtained by acquirers and targets involved in M&A during recessions and non-recessions. Detailed results for acquirers and targets, during recessions and non-recessions, can be found in Appendix C, Tables C1 – C8. The difference between ARs during recessions and during non-recessions are calculated and tested. A significant difference implies that market participants value M&A differently depending on the phase of the economic cycle.
4.2.1 M&A activity level and the economic cycle
The findings of Wann and Lamb (2016) and Cools et al. (2007) about the existing relation between the economic cycle and M&A activity level are confirmed by this research. Table 2A and 2B in Section 3.1 show that all three samples (A), (B) and (C) include more observations of acquirers and targets involved in M&A during non-recessions than during recessions. An additional analysis of sample (B) shows that on average per month there are more M&A announced during non-recessions than during recessions. These findings confirm the relation between the economic cycle and M&A activity level. Recessions have a negative effect on the level of M&A activities and non-recessions have a positive effect. This result complies with the Recession Cleansing Hypothesis by Schumpeter (1942), Economic Disturbance Theory by Gort (1969) and Aguiar and Gopinath (2005), which explain the higher M&A activity level during non-recessions.
4.2.2 M&A valuation for acquirers and the economic cycle
5% significance level during recessions. This can hardly be associated with the event since it occurs much later and is the only significant return within the entire event window. During non-recessions, acquirers observe significant CAARs around the event day. On days -1 and 0 the CAARs are negative, but on day 3 it is positive. This indicates that the market participants’ negative attitude towards the acquisitions changes positively in the post-event window. These results are in line with Wann and Lamb (2017), who do not find significant CAARs for acquirers during recessions but find significant CAARs during non-recessions.
Table 8
Differences in CAARs during recessions and non-recessions for acquirers and targets (single days).
This table reports the CAARs for acquirers and targets during recessions and non-recessions on each day t in the event window. The difference in CAARs during recessions and non-recessions is calculated and tested on significance. Results of the columns (A1), (A2), (B1) and (B2) are based on the Gsign. Results of the columns (A1-A2) and (B1-B2) are based on the Independent T-test with unequal variances.
A. Acquirers B. Targets (A1) Recession (A2) Non-Recession (A1-A2) Difference (B1) Recession (B2) Non-Recession (B1-B2) Difference
t CAAR (%) CAAR (%) ΔCAAR (%) CAAR (%) CAAR (%) ΔCAAR (%)
-5 0.513 0.057 0.456 -0.346 0.476 -0.822 -4 0.539 0.125 0.414 1.455 0.425 1.030 -3 0.711 -0.019 0.730 5.251 0.264** 4.987* -2 1.145 0.024 1.121 4.391 0.445** 3.946* -1 0.629 -0.022** 0.651 5.472 0.731*** 4.741** 0 -0.631 -0.161*** -0.470 27.027*** 21.757*** 5.270 1 -2.031 0.050 -2.081 35.764*** 29.071*** 6.693 2 -2.561 0.098 -2.659 35.970*** 28.998*** 6.972 3 -2.027 0.183** -2.210 36.478*** 28.926*** 7.552 4 -2.872 0.000 -2.872 35.182*** 29.039*** 6.143 5 -3.546** 0.075 -3.621 34.831*** 28.914*** 5.917 Obs. 65 461 40 348
Note: ***, ** and * denote 1%, 5% and 10% significance level, respectively.
that the reactions of market participants to M&A announced during recessions and non-recessions are not significantly different from each other.
Table 9
Differences in CAARs during recessions and non-recessions for acquirers and targets (sub-event windows). This table reports the CAARs for acquirers and targets during recessions and non-recessions across various sub-event windows. The difference in CAARs during recessions and non-recessions is calculated and tested on significance. Results of the columns (A1), (A2), (B1) and (B2) are based on the Gsign. Results of the columns (A1-A2) and (B1-B2) are based on the Independent T-test with unequal variances.
C. Acquirers D. Targets (C1) Recession (C2) Non-Recession (C1-C2) Difference CAAR (D1) Recession (D2) Non-Recession (D1-D2) Difference CAAR
Window CAAR (%) CAAR (%) ΔCAAR (%) CAAR (%) CAAR (%) ΔCAAR (%)
(-5/5) -3.546** 0.075 -3.621 34.831*** 28.914*** 5.917 (-5/-3) 0.711 -0.019 0.730 5.251 0.264** 4.987* (-5/-1) 0.629 -0.022** 0.651 5.472 0.731*** 4.741** (-1/0) -1.776** -0.186** -1.590* 22.636*** 21.311*** 1.325 (0/1) -2.661** 0.072 -2.733** 30.293*** 28.341*** 1.952 (-1/1) -3.176** 0.026 -3.202*** 31.374*** 28.626*** 2.748 (-2/2) -3.272 0.117 -3.389** 30.719*** 28.733*** 1.986 (1/5) -2.915 0.236 -3.151** 7.805 7.158*** 0.647 (3/5) -0.985 -0.023 -0.962 -1.138 -0.083 -1.055 Obs. 65 461 40 348
Note: ***, ** and * denote 1%, 5% and 10% significance level, respectively.
Table 9 presents the results for the same analysis, but on various sub-event windows within the entire event window. There are four event windows [(-5/5), (-1/0), (0/1) and (-1/1)] with significantly negative CAARs for acquirers that are involved in M&A during recessions [see column (C1)]. For acquirers during non-recessions, the event windows (-5/-1) and (-1/0) are significant [see column (C2)]. All aforementioned windows are significant at 5%. The TT finds evidence that the CAARs of acquirers during recessions are significantly different from the CAARs during non-recessions in the following windows: (-1/0), (0/1), (-1/1), (-2/2) and
(1/5). All differences are negative and the most negative occurs during the event window (-2/2), where the difference between recession- and non-recession CAARs is -3.389%.
Contrarily to the results presented in Table 8 (using single days), it seems that the economic cycle has an effect on the valuation of M&A for acquirers, when investigating sub-event windows.
findings of Wann and Lamb (2017). However, there are significantly negative differences when using sub-event windows (Table 9). Due to the results in Table 9, the null hypothesis of no significant differences in CAARs experienced by acquirers during recessions and non-recessions is rejected, as well as the qualitative hypothesis H3 in Section 2.5. The CAARs are
significantly lower during recessions than non-recessions. Hence, the economic cycle causes market participants to value M&A, which are announced during recessions, lower than during non-recessions.
4.2.3 M&A valuation for targets and the economic cycle
Columns (B1) and (B2) in Table 8 show the CAARs of target firms during recessions and non-recessions, respectively. The CAARs for targets during recessions are all positive (except for day -5) and significantly positive at 1% from the event day onwards. The CAARs during non-recessions in (B2) are all positive without exception. Days -3 and -2 exhibit significant CAARs at the 5% significance level and days from -1 to 5 show highly significant values at the 1% level. The CAARs from the event day onwards are remarkably high during both phases of the economic cycle [see (B1) and (B2)] and can be associated with the M&A announcements. Wann and Lamb (2017) also find significant CAARs for targets due to announcements of M&A in both recessions and non-recessions. Although the CAARs of targets for both economic phases are high and significant, the values during recessions are greater than the ones during non-recessions (except for day -5). This leads to the result, that all differences of CAARs in column (B1-B2) are positive (except for day -5). Three out of 11 CAAR differences are significant according to the TT. Day -3 and -2 show weak evidence (at 10% significance level) and day -1 shows strong evidence (at 5% significance level) that the economic cycle influences the valuation of M&A announcements for targets.
The CAARs for targets in (D1) (Table 9) are greater than the ones in (D2), except for the event window (3/5). Five out of nine windows exhibit significant CAARs during recessions and eight out of nine windows during non-recessions. Column (D1-D2) shows the differences in CAARs experienced by targets during recessions and non-recessions due to M&A announcements. They are significant at the 10% and 5% significance level for the pre-event windows (-5/-3) and (-5/-1), respectively. For the windows close to the event day and in the post-event period, the differences in CAARs are not significant.
In summary, the impact of the business cycle on the valuation of M&A for targets is confirmed given the significant differences between CAARs during recessions and non-recessions. Therefore, the qualitative hypothesis H4 in Section 2.5 is accepted. CAARs of
careful about the economic environment. Their early and sensitive reactions to news and events like M&A leads to leakage effects. In contrast, the returns on the event day and in the post-event window are not significantly different during recessions compared to non-recessions. Wann and Lamb (2017) find significant differences between the CAARs during recessions and non-recessions as well, but their findings are more pronounced. According to them, significant CAAR differences additionally exist during the announcement day and post-event window, which cannot be confirmed by this research. Therefore, this paper can only confirm the impact of the economic cycle on the valuation of M&A announcements during the pre-event window.
4.3 The impact of variables on targets’ and acquirers’ abnormal returns
This section investigates the potential impact of the four variables from Section 2.4 on the abnormal returns of acquirers and targets during recessions and non-recessions. Thereafter, the relation between the potential impacts of the variables and the economic cycle is investigated by comparing the impacts of the variables during recessions to the impacts during non-recession using the Z-test by Paternoster, Brame, Mazerolle and Piquero (1998) (see Appendix D for detailed information on the Z-test).
The abnormal returns of acquirers and targets calculated in the event study in Section 4.1 are used as the dependent variables of the regressions. The abnormal returns of three different time windows are chosen. Due to the Efficient Market Hypothesis by Fama (1970), which states that ARs appear only on the event day, the abnormal returns of the event day [AR(0)] are used. In contrast to the EMH, the event study shows that there are leakage- and delayed effects around the event day. Therefore, the CARs of the 3-day event window (-1/1) are also used. The entire 11-day window consistently shows significant abnormal returns in the event study. For this reason, the regression utilizes the CARs of the window (-5/5) as well. Section 4.3.1 presents the results of the regression using the abnormal returns of acquirers and Section 4.3.2 shows the results using the targets’ abnormal returns.
4.3.1 Regression on acquirers’ abnormal returns
Relative Size
The coefficients of the SIZE variable show negative values in all regressions [(A1) to (A6)] but only the ones exhibited during recessions [in (A1) and (A3)] are significant at the 5% level. Acquirers involved in M&A during recessions experience a decrease in their abnormal returns of 0.54%- and 1.16%-points if the relative size of the acquisition increases by 10%-points [see the SIZE coefficients in (A1) and (A3)]. This reveals that the larger the MV of the target’s equity, the larger the relative size ratio, which consequently leads to lower ARs for acquirers when conducting an acquisition during recessions. On the other hand, there is no significant relative size effect on the acquirers’ abnormal returns during non-recessions [see (A2), (A4) and (A6)]. H5 in Section 2.5 states that there is a significant negative relation
between the acquirers’ abnormal returns and the relative size of the acquisition. This hypothesis is accepted for the recessions but rejected for the non-recessions.
Table 10
Regression results on acquirers’ abnormal returns during non-recession and recession.
The table presents the results of the regression analysis (OLS) of the ARs and CARs of acquirers. Size is the relative size of the acquisition. D1, MA, TCEGS, D2, and FIR are dummy variables presenting the different industries of the firms. SUB is a dummy variable taking the value of 1 if the M&A are domestic. PAY is a dummy variable taking the value of 1 if the M&A are paid with stocks.
AR(0) CAR(-1/1) CAR(-5/5)
Variable (A1) Recession (A2) Non-Recession (A3) Recession (A4) Non-Recession (A5) Recession (A6) Non-Recession INTERCEPT 0.051 -0.014 0.075 -0.006 0.029 -0.013 SIZE -0.054** -0.004 -0.116** -0.002 -0.105 -0.001 D1 0.010 -0.026 -0.067 -0.007 -0.367* 0.005 MA -0.030 0.010 -0.066 0.009 0.011 0.009 TCEGS 0.012 0.005 -0.014 0.000 0.009 0.009 D2 0.002 0.022 -0.052 0.058*** -0.046 0.077*** FIR -0.015 0.004 -0.023 0.011 0.024 0.015 SUB -0.032 0.008 -0.053 -0.004 -0.056 0.001 PAY 0.014 -0.006 0.025 -0.007 0.026 -0.017* Adjusted R² 0.094 0.002 0.076 0.043 -0.034 0.026 S.E. of Regression 0.049 0.058 0.096 0.052 0.243 0.080 Observations 37 314 37 314 37 314
Note: ***, ** and * denote 1%, 5% and 10% significance level, respectively.
level, respectively. Acquirers involved in M&A during recessions experience a significantly larger decrease in abnormal returns when the relative size increases, compared to acquirers during non-recessions. A possible explanation, according to Hansen (1987), is a revaluation loss for acquirers, which increases when the size of the acquired target increases. Filipovic (2015) states that agency problems increase if the targets’ size is larger than- or similar to the acquirers’ size. These situations are more problematic for firms operating during uncertain economic conditions (recessions) than during non-recessions, which is mirrored in the results. The presented result provide enough evidence to conclude that the economic cycle affects the impact of the relative size variable.
Industry effect
The coefficients of all industry dummies (D1, MA, TCEGS, D2 and FIR) show the difference in abnormal returns between acquirers that operate in the associated industry and acquirers that operate in the Services industry. Significant industry coefficients mean that the ARs of acquirers are significantly different from each other based on their industry affiliation. This is the case for the industry D2 during non-recessions because it shows significantly positive coefficients (at the 1% significance level) in (A4) and (A6). In economic terms, acquirers that operate in the D2 industry (Wholesale- or Retail Trade) when M&A are announced during non-recessions have significantly higher abnormal returns by 5.8% and 7.7%, compared to firms in the Services industry. The coefficient of D1 in (A5) is 36.7% and significant at 10%, but this occurs due to the small amount of observations (2 obs.) and is therefore not meaningful. Due to the fact that at least one industry variable is significant, H7 (in Section
2.5) can be accepted. Contrarily, one significant variable out of five is economically-speaking not enough evidence for an existing industry effect. Therefore, it is concluded that acquirers’ industry affiliation has no significant impact on the ARs.
The differences between the coefficients of the industry variables due to the economic cycle show occasional significance. Table D1 in Appendix D shows weak significant differences in coefficients at the 10% significance level. Therefore, there is not enough evidence to conclude that the economic cycle causes coefficients to be significantly different from each other.
Sub-deal type
The coefficients of the sub-deal type variable do not show any significance at any window. Consequently, the sub-deal type of an acquisition does not influence the abnormal returns of acquirers. Similar to Goergen and Renneboog (2004), significant differences in ARs between domestic and cross-border M&A are not found. Therefore, the hypothesis H9 in Section 2.5,
which hypothesizes that the ARs of acquirers depend on whether the firm is involved in a cross-border or domestic acquisition, is rejected.
conclude that the economic cycle influences the impact of the sub-deal type variable without risking a Type-1 error.
Payment method
The dummy variable PAY shows overall insignificant coefficients. Only the coefficient in (A6) is weakly significant at 10% and negative. However, the significance level is not low enough to accept the qualitative hypothesis H11 without risking a high probability of a Type-1
error. Consequently, no statistically significant linear dependence of the payment method on the abnormal returns of acquirers is detected in this research.
Furthermore, the Z-test executed on the difference of the PAY coefficients does not show any significance. Hence, there is no proof that the impacts of the payment method on the ARs of acquirers are significantly different from each other during different phases of the economic cycle.
4.3.2 Regression on targets’ abnormal returns
In this section, the same independent variables as in Section 4.3.1 are regressed on the abnormal returns of target firms. Table 11 shows the results. Similar to the results for acquirers, the Adjusted R values are low.
Table 11
Regression results on targets’ abnormal returns during non-recession and recession.
The table presents the results of the regression analysis (OLS) of the ARs and CARs of targets. Size is the relative size of the acquisition. D1, MA, TCEGS, D2, and FIR are dummy variables presenting the different industries of the firms. SUB is a dummy variable taking the value of 1 if the M&A are domestic. PAY is a dummy variable taking the value of 1 if the M&A are paid with stocks.
AR(0) CAR(-1/1) CAR(-5/5)
Variable (B1) Recession (B2) Non-Recession (B3) Recession (B4) Non-Recession (B5) Recession (B6) Non-Recession INTERCEPT 0.275 0.323*** 0.444** 0.437*** 0.381 0.463*** SIZE -0.307* -0.063** -0.446** -0.085*** -0.417* -0.079** D1 0.302 -0.117 0.224 -0.149 0.303 -0.150 MA 0.006 -0.078 0.052 -0.078 0.208 -0.080 TCEGS -0.002 -0.090 -0.112 -0.168* -0.008 -0.162* D2 0.008 -0.015 0.160 -0.069 0.255 -0.069 FIR 0.094 -0.137*** 0.034 -0.138** -0.036 -0.122** SUB -0.004 -0.013 -0.092 -0.032 -0.069 -0.067 PAY -0.028 0.003 0.101 0.007 0.156 0.020 Adjusted R² -0.103 0.027 0.052 0.028 -0.015 0.022 S.E. of Regression 0.374 0.299 0.393 0.361 0.509 0.377 Observations 37 314 37 314 37 314
