- Email: [email protected]
Acquirer reference prices and acquisition performance
Accepted Manuscript
Acquirer Reference Prices and Acquisition Performance Qingzhong Ma , David A. Whidbee , Wei Zhang PII: DOI: Reference:
S0304-405X(18)30280-0 https://doi.org/10.1016/j.jfineco.2018.10.004 FINEC 2976
To appear in:
Journal of Financial Economics
Received date: Revised date: Accepted date:
27 February 2017 5 February 2018 5 March 2018
Please cite this article as: Qingzhong Ma , David A. Whidbee , Wei Zhang , Acquirer Reference Prices and Acquisition Performance, Journal of Financial Economics (2018), doi: https://doi.org/10.1016/j.jfineco.2018.10.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Acquirer Reference Prices and Acquisition Performance*
CR IP T
Qingzhong Ma David A. Whidbee Wei Zhang This version: 2018 03 18 Abstract In a comprehensive sample of mergers and acquisitions, we find a reference price effect: Acquirers earn higher (lower) announcement-period returns when their
AN US
pre-announcement stock prices are well below (near) their 52-week highs. This reference price effect is stronger in acquisitions of private targets, deals involving greater uncertainty, and acquirers with greater individual investor ownership, and it is reversed in the subsequent year. Further, acquirer reference prices affect bid
M
premia and target announcement-period returns in deals with greater uncertainty in acquirer valuation. The overall evidence is consistent with investors irrationally
ED
using 52-week high prices as a measure of acquirer valuation.
valuation.
PT
Keywords: Mergers; Acquisitions; Anchoring; Reference point; Perceived
*
CE
JEL classification: G31, G34
AC
Ma: 325 Tehama Hall, College of Business, California State University, Chico, CA. Tel: 530 898 6182; email: [email protected]. Whidbee (corresponding author): Todd Hall Addition 470, PO Box 644746, Carson College of Business, Washington State University, Pullman, WA. Tel: 509 335 3098; email: [email protected]; fax:509 335 3857. Zhang: 465 Tehama Hall, College of Business, California State University, Chico, CA. Tel: 530 898 6544; email: [email protected]. We are very grateful to Bill Schwert (the editor) and to Ming Dong (the referee) for thoughtful comments and suggestions that substantially improved the paper. We also appreciate discussions, comments, and suggestions from Mehmet Akbulut, Malcolm Baker (AFA discussant), Dalen Chiang, Jim Downing, Levon Goukasian, Jarrad Harford, Kateryna Holland (FMA discussant), Christine Hsu, Emily Jian Huang, Yoonsoo Nam, Micah Officer, Richard Ponarul, Sergey Smirnov, Hai Tran (CCFC discussant), Stephen Treanor, and seminar participants at California State University at Chico, Southwest University of Finance and Economics, Chengdu, China, Washington State University, California Corporate Finance Conference (CCFC) 2016 at Loyola Marymount University, the Financial Management Association (FMA) 2016 Annual Meeting, and the American Finance Association (AFA) 2017 Annual Meeting. All remaining errors are the authors’. Earlier versions of the paper were circulated under the title “Anchoring and Acquisitions.”
ACCEPTED MANUSCRIPT
1. Introduction The stock market’s reaction to announcements of mergers and acquisitions (M&A) has been extensively examined. 1 Most existing studies assume investors rationally process the information available at the time of the announcement and incorporate it into stock prices. In
CR IP T
recent years, however, a growing body of literature challenges the notion that stock prices rationally reflect public information in a timely manner. Several studies find evidence that a stock price’s proximity to past peak prices affects investor behavior. George and Hwang (2004), for example, find that a stock price’s proximity to its 52-week high predicts returns better than
AN US
traditional momentum strategies.2 Baker et al. (2012) find strong evidence that offer premia paid to publicly traded targets are significantly affected by the 52-week high prices of the targets, suggesting that reference prices play an important role in perceived valuation levels of targets in
M
acquisitions. In this paper, we investigate whether 52-week high prices play a role in the perceived valuation levels of acquirers at the M&A announcements.3
ED
In a comprehensive sample of 19,119 acquisitions over the 1981–2014 period involving both public and private targets, we find that bidders with pre-announcement stock prices well
PT
below their 52-week highs earn significantly higher announcement-period abnormal returns than
CE
those with prices at or near their 52-week highs. That is, the acquirer’s reference price ratio (or RPR, the ratio of the pre-announcement stock price to its 52-week high price) is negatively 1
AC
The literature is vast and is comprehensively surveyed in Jensen and Ruback (1983), Jarrell et al. (1988), Andrade et al. (2001), and Betton et al. (2008). 2 Additional evidence includes Heath et al. (1999) who find a strong tendency for company executives to exercise stock options when the underlying stock price moves past its 52-week high. Huddart et al. (2009) document a significant increase in trading volume when a stocks’ price crosses its 52-week high. Driessen et al. (2013) examine implied volatilities derived from option prices and find that implied volatilities decrease for stocks approaching their 52-week high, but then increase after prices cross this threshold. 3 Many of the studies examining the influence of the 52-week high on investor behavior also consider the role of the 52-week low and often find that the historic low plays a statistically significant role. However, the 52-week high seems to have a stronger influence. In addition, most acquisitions are made by firms with stock prices closer to their 52-week high than their low. In our sample, for example, 65% of acquirers have stock prices nearer to their 52-week high than their 52-week low. Therefore, we focus on the role of the 52-week high in the bulk of our analysis. Nevertheless, we also consider the role of the 52-week low in our robustness checks. Page 1
ACCEPTED MANUSCRIPT
related to the acquirer’s announcement-period abnormal returns. We call this result the reference price effect.4 Like other studies examining how the 52-week high affects investor behavior, we focus on the anchoring phenomenon, which is considered one of “the most reliable and robust” results of experimental psychology (Kahneman, 2011). We refer to the possibility that anchoring
CR IP T
influences investors’ responses to merger announcements and merger outcomes as the perceived valuation hypothesis.
According to the anchoring phenomenon, individuals commonly rely on salient, even if fundamentally irrelevant, anchors in forming beliefs or norms (Tversky and Kahneman, 1974). If
AN US
the 52-week high is a meaningful price reference point for investors, as suggested by the empirical evidence, it may influence investors’ perceptions of valuation levels in mergers in much the same way as other valuation levels. Typically, valuation levels refer to the relation
M
between current stock prices and their book values or estimated fundamental values. These conventional valuation levels influence market reactions to acquisition announcements and deal
ED
outcomes (e.g., Rhodes-Kropf et al., 2005; Dong et al., 2006). We hypothesize that the relation between the current stock price and recent peak values, such as the 52-week high, may also
PT
influence mergers and acquisitions. Specifically, if the acquirer's stock price is near its 52-week
CE
high, investors may perceive the stock as having a relatively high valuation level and therefore be reluctant to bid up the price and be more inclined to sell down the price in response to an
AC
acquisition announcement. By contrast, if the acquirer's pre-announcement stock price is well below the 52-week high, investors may perceive the acquirer's valuation to be low and therefore be more willing to bid up rather than sell down the stock price.
4
To be clear, our focus is on acquirer reference prices, as compared to the target reference price effect documented in Baker et al. (2012). For convenience, we use the term “reference price effect” to refer to the acquirer reference price effect. Whenever necessary to avoid confusion or to be more precise, we use the more explicit term “acquirer reference price effect.” Further, we use the terms acquirer and bidder interchangeably throughout. Page 2
ACCEPTED MANUSCRIPT
An additional implication of the anchoring phenomenon is that anchors are more likely to influence numeric estimates when subjects have less knowledge (Mussweiler and Strack, 2000) and/or there is greater uncertainty (Jacowitz and Kahneman, 1995). Therefore, the reference price effect should be more important in acquisitions with limited information about the target,
CR IP T
transactions that involve greater uncertainty, acquirers with an opaque information environment, and acquirers owned by less sophisticated investors. Consistent with this implication, we find a stronger reference price effect among acquisitions of private targets (including independent private firms and subsidiaries of other companies) than acquisitions of public targets. Further, the
AN US
reference price effect is stronger among acquirers with higher stock return volatility, acquirers that are followed by fewer analysts, acquisitions that are relatively large, and acquirers with higher individual investor ownership levels.
M
A possible alternative explanation for our findings is that the reference price effect is due to the impact of conventional valuation levels. When a bidder’s stock price is near its 52-week
ED
high, its stock price might also be high relative to its fundamental or intrinsic value. In the context of acquisitions, overvalued acquirers earn lower returns (e.g., Shleifer and Vishny, 2003;
PT
Rhodes-Kropf and Viswanathan, 2004). However, the reference price effect remains strong even
CE
after controlling for a host of firm and deal characteristics, including a variety of proxies for stock valuation level, suggesting the reference price effect is not explained by conventional
AC
valuation levels or other firm or deal characteristics. Further, in acquisitions of private targets, we find a significant reference price effect even among acquisitions paid entirely with cash, a sample least likely to be influenced by conventional valuation levels. Yet another possibility is that the reference price effect acts as a proxy for some fundamental but unidentified factor associated with acquisition decisions. For example, fueled by
Page 3
ACCEPTED MANUSCRIPT
high stock prices relative to their 52-week highs, managers might develop hubris or become overconfident (Roll, 1986; Malmendier and Tate, 2008) and make poor acquisition decisions. Higher stock prices might also give managers greater leverage over monitoring mechanisms, such as the board of directors and the market for corporate control, and protect them when they
CR IP T
engage in activities that benefit themselves at the expense of shareholder value (e.g., Amihud and Lev, 1981; Jensen, 1986; Morck et al., 1990; Fu et al., 2013; Akbulut, 2013; Duchin and Schmidt, 2013). As another possibility, when a company’s stock price is near its 52-week high, investors might have anticipated the acquisition and incorporated such an anticipation into the
AN US
price already. Given that acquirers on average earn positive returns, at least in our sample, such an anticipation may lead to lower announcement returns for anticipated acquisitions. Finally, bidders with stock prices near their 52-week highs may, indeed, be overvalued and the reference
M
price effect may be a rational response to acquisition announcements by overvalued acquirers. These indirect effects are plausible and may be reflected in the market’s reaction to the merger
ED
announcement.
If this is the case, however, we would expect the market’s reaction to these merger
PT
announcements to be permanent. By contrast, if the announcement-period return pattern is being
CE
driven by anchoring, an irrational behavioral bias, the return pattern is likely reversed over the long horizon when the bias is eventually corrected. To further distinguish the perceived valuation
AC
hypothesis from alternative explanations, we examine the long-horizon abnormal returns to merger announcements and find that the announcement-period return pattern is reversed over the subsequent one-year period. Specifically, the long-horizon abnormal returns for acquirers with pre-announcement stock prices well below their 52-week highs are lower than those for acquirers with pre-announcement stock prices near their 52-week highs. While statistical significance
Page 4
ACCEPTED MANUSCRIPT
varies, this conclusion holds for buy-and-hold abnormal returns in portfolio analysis and in regressions. The same pattern of return reversal is also observed in short-window earnings announcements over the one-year period following the M&A announcements. This finding is inconsistent with the reference price effect acting as a proxy for some unidentified fundamental
CR IP T
factor associated with merger decisions or as a proxy for conventional valuation levels. Rather, our results are more consistent with investors’ initial responses to acquisition announcements being irrationally influenced by perceived valuation levels relative to 52-week high prices.
The reference price effect survives a battery of robustness checks and falsification tests. It
AN US
is robust to alternative models of estimating abnormal returns and alternative reference price ratio measures. The effect is strong in the 1980s, 1990s, and the new millennium. We also examine the effects of bidder and target reference prices on bid premium and
M
target announcement return in the subsample of deals involving targets that are covered in the Center for Research in Security Prices (CRSP). We find that bidders with pre-announcement
ED
prices closer to their 52-week highs pay higher bid premia, although the effect is significant only among non-cash deals and deals involving more uncertainty. This is consistent with the
PT
interpretation that in deal negotiations, acquirers perceived to be overvalued are compelled to
CE
offer high premia to their targets. Also, as discussed in Dong et al. (2006), perceived acquirer overvaluation can influence target announcement returns for two reasons: Acquirer overvaluation
AC
should give rise to a higher bid premium, and bid announcement should alert the market to reassess the pre-existing (acquirer and target) misvaluation. Indeed, we observe that acquirers with stock prices closer to their 52-week highs are associated with higher target announcement returns in the whole CRSP sample and among the subsamples of non-cash deals and deals involving more uncertainty.
Page 5
ACCEPTED MANUSCRIPT
The 52-week high price of the target also has a significant impact on bid premium and target announcement return. Consistent with Baker et al. (2012), we find that targets with stock prices well below their 52-week highs are associated with higher bid premia in the whole CRSP sample, suggesting that targets perceived to be undervalued are able to negotiate high premia.
CR IP T
However, we find that the target reference price effect on bid premium is significant only in the subsample of cash deals or when acquirer uncertainty is low. A possible interpretation in the context of reference prices is that, when the uncertainty of bidder valuation is high (in deals with non-cash payment or high acquirer stock volatility), the target firm mainly uses the bidder
AN US
reference price to negotiate a high premium. When acquirer valuation uncertainty is low or in cash deals, on the other hand, the target primarily relies on the target price distance to its 52week high in the bid premium negotiation. Similarly, we find the ratio of target stock price to
M
target 52-week high exerting a negative impact on target announcement return in the CRSP sample, with a stronger effect in subsamples of low acquirer and deal uncertainty.
ED
In contrast to the reference price influence on target and bidder announcement-period returns and bid premia, we find little consistent evidence that reference prices (for either the
PT
bidder or target) affect the method of payment or deal combativeness. A possible interpretation is
CE
that stock market reactions depend on investor perceptions of acquirer and target valuations, and investors have less information than managers. Therefore, investors are more likely to rely on
AC
heuristics when assessing the valuation implications of an acquisition announcement. Similarly, even though an offer premium is a managerial decision, it is commonly the result of negotiations between acquirer and target managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and deal combativeness, on the other hand, are primarily decisions by the acquirer and target managers and, if acquirer and target managers are
Page 6
ACCEPTED MANUSCRIPT
more informed of their own firms’ true valuations, these decisions will be less affected by their stocks’ reference prices. The paper contributes to two important areas in the finance literature. First, we contribute to the M&A literature by documenting a new, non-fundamental factor in affecting the acquirer
CR IP T
and target investor reactions to merger announcements and bid premia. Our paper adds to the evidence that behavioral factors influence the takeover market. The literature already documents such behavioral factors as investor misvaluation (Dong et al., 2006) and managerial overconfidence (Malmendier and Tate, 2008). More recently, Danbolt et al. (2015) find a
AN US
positive relation between a measure of investor sentiment and acquirer announcement-period abnormal returns. Otherwise, most existing studies interpret market reactions at the announcements as reflecting expected changes in firm fundamentals. Part of the relatively recent
M
literature in this vein focuses on management performance (Lang et al., 1989), asset relatedness (Matsusaka, 1993), corporate governance (Masulis et al., 2007), free cash flows (Lang et al.,
ED
1991; Harford, 1999), firm size (Moeller et al., 2004), chief executive officer (CEO) network centrality (El-Khatib et al., 2015), target asset uncertainty (Officer et al., 2009), dormant period
PT
(Cai et al., 2011), management entrenchment (Harford et al., 2012), and cultural similarity
CE
(Bereskin et al., 2017). Studies of the long-term performance of mergers emphasize the role of over-extrapolation (Rau and Vermaelen, 1998), market-wide valuation (Bouwman et al., 2009),
AC
firm-level valuation (Akbulut, 2013; Fu et al., 2013), and merger waves (Duchin and Schmidt, 2013), among others. We document the impact of a behavioral bias, specifically the anchoring bias, on investors’ perceptions of valuation levels and their reactions to announcements of mergers and acquisitions. As a unique feature of this study, we show not only the bias in the announcement period but also its correction in the longer term.
Page 7
ACCEPTED MANUSCRIPT
Second, our paper adds further evidence on the role of anchoring as a psychological bias in behavioral finance. One of the central themes in behavioral finance is to understand what psychological phenomena help shape investors’ demand for securities (Barberis et al., 1998; Shleifer, 2000; Baker, 2009; Baker and Wurgler, 2012). For the special case of firms making
CR IP T
acquisitions, this paper reports evidence that anchoring bias and its influence on valuation-level perceptions help shape investor demand for stocks. That is, for acquiring firms with stock prices well below their 52-week highs, investors seem more (less) inclined to bid up (sell down) prices in response to an acquisition announcement. Because the 52-week high is fundamentally
AN US
irrelevant, this anchoring bias in the initial market reaction is corrected in the longer term. This reversal suggests that the acquirer reference price effect is irrational, adding a new insight beyond Baker et al. (2012). In addition, we find that the anchoring bias has a stronger impact
M
among acquisitions with limited information about the targets, acquisitions involving greater uncertainty about acquirer valuation, and acquirers with higher levels of ownership by less
ED
sophisticated investors, consistent with the notion that investor bias is more likely to exist when the difficulty of estimating firm value is greatest and the forces of arbitrage are relatively limited.
PT
We develop the perceived valuation hypothesis in Section 2 and describe the sample and
CE
data in Section 3. Empirical results of acquirer announcement-period and long-run returns are presented in Section 4. We examine deal characteristics in Section 5. We check the robustness of
AC
the main results in Section 6 and conclude in Section 7. 2. The perceived valuation hypothesis A growing literature finds evidence of stocks’ past peak prices, especially the 52-week
high, being important in explaining investor and manager behavior. The general premise is that the 52-week high acts as an important reference point. This argument is supported in numerous
Page 8
ACCEPTED MANUSCRIPT
decision scenarios, including employee stock option exercising (Heath et al., 1999), momentum (George and Hwang, 2004), post-earnings announcement drift (George et al., 2014), investment and financing anomalies (George et al., 2018), and trading volume (Huddart et al., 2009). Additionally, Li and Yu (2012) find evidence of investor anchoring on the Dow 52-week high.
CR IP T
The results of these studies suggest that 52-week highs influence perceptions of valuation levels. For example, Baker et al. (2012) find a target reference price effect indicating that merger decision makers and target shareholders look to the target’s 52-week high as an indicator of valuation levels when evaluating offer premia. To the extent investors, managers, and other
AN US
decision agents’ perceptions of valuation levels for acquirers are also influenced by 52-week highs, the relation between current acquirer stock prices and their 52-week highs may influence mergers.
M
The adoption of the 52-week high price as a reference point has deep roots in the psychology literature, particularly the anchoring mechanism of Tversky and Kahneman (1974),
ED
who show that subjects are influenced by easily available reference points in their estimates of an unknown quantity. In real estate markets, for example, listing prices influence perceived values
PT
for both amateurs and real estate professionals (Northcraft and Neale, 1987), and past purchase
CE
price affects seller behavior (Genesove and Mayer, 2001). In legal disputes, damage awards are influenced by what is asked for in court (Hastie et al., 1999). Managers of initial public offering
AC
(IPO) firms measure their wealth losses and gains against some reference point (Loughran and Ritter, 2002). Firms manage earnings to exceed certain thresholds (DeGeorge et al., 1999). The number of options granted to CEOs tends to be rigid (Shue and Townsend, 2017). CEOs are more likely to issue equity due to stock performance since their arrival rather than inherited stock performance (Baker and Xuan, 2016). Mutual fund managers are more willing to sell past losers
Page 9
ACCEPTED MANUSCRIPT
if the losers are inherited (Jin and Scherbina, 2011). In a behavioral signaling model, Baker et al. (2016) argue that investors evaluate current dividends against some reference point established by the firm’s dividend history. The literature on human decision making identifies a variety of psychological
CR IP T
mechanisms associated with anchoring. Tversky and Kahneman (1974) propose an anchoringand-adjustment process whereby individuals rely on salient reference points as starting points in numerical estimations and adjust away from the reference point until the estimate falls within the range of plausible estimates (Strack and Mussweiler, 1997). Consequently, adjustments tend to
AN US
be insufficient and eventual estimates are biased toward the initial reference point. Other mechanisms include selective accessibility (Strack and Mussweiler, 1997), confirmatory hypothesis testing (Chapman and Johnson, 1994), and numeric priming (Oppenheimer et al.,
M
2008).
Regardless of the specific psychological mechanism involved, we consider the possibility
ED
that investors tend to anchor on the 52-week high in the evaluation of firms making acquisitions and that the 52-week high influences investors’ valuation-level perceptions. Absent any
PT
behavioral biases, we would expect the process of estimating the acquiring firm’s new stock
CE
price to start with the firm’s pre-acquisition price and then adjust away from that price to reflect the synergistic gains associated with the merger, the premium paid to the selling company, the
AC
method of payment, and other value-relevant information associated with the acquisition. Just as in the process of making a merger bid to a publicly traded firm (Baker et al., 2012) or assessing the impact of earnings announcements (George et al., 2014), however, investors face varying degrees of uncertainty and indeterminacy when valuing firms that are making acquisitions. When investors face a high degree of uncertainty, they may rely on heuristics to arrive at valuation-
Page 10
ACCEPTED MANUSCRIPT
level estimates. Thus, reference prices may play an important role in explaining the stock market’s response to an acquisition announcement. Specifically, if an acquirer’s 52-week high price influences some investors’ perceptions of valuation levels, these investors’ price adjustment processes at the announcement may be
CR IP T
influenced by the distance between the acquirer’s current stock price and its 52-week high. For acquirers with stock prices well below their 52-week highs, investors may perceive them as having a relatively low valuation. For acquirers with stock prices at or near their 52-week highs, on the other hand, investors may have a high perceived valuation of their stocks. Ultimately,
AN US
these arguments imply that acquirer announcement-period abnormal returns will be higher for acquirers with pre-announcement prices well below their 52-week highs than for acquirers with pre-announcement prices near their 52-week highs.
M
A further prediction of the perceived valuation hypothesis is that the influence of acquirer 52-week high on investor behavior should be more pronounced in acquisitions involving greater
ED
uncertainty and among acquisitions in which market participants have less information about the transaction. This prediction is based on evidence that reference points are more likely to
PT
influence numeric estimates when there is greater uncertainty (Jacowitz and Kahneman, 1995)
CE
and/or when subjects have less knowledge (Mussweiler and Strack, 2000). Specifically, we expect to observe a stronger acquirer reference price effect in acquisitions involving private
AC
targets, uncertainty about acquirer valuation, relatively large acquisitions, and non-cash forms of payment. Further, a substantial literature finds that individual investors are prone to behavioral biases (e.g., Odean, 1998; Barber and Odean, 2008) and that institutional investors are more sophisticated (e.g., Nofsinger and Sias, 1999; Bartov et al., 2000). Therefore, we expect a
Page 11
ACCEPTED MANUSCRIPT
stronger reference price effect in acquirers with higher individual (lower institutional) investor ownership. In addition to influencing acquirer abnormal returns, the perceived valuation hypothesis has implications for other deal characteristics. Conventional valuation levels have been shown to
CR IP T
influence merger decisions, the market response to those decisions, and their ultimate outcomes. For example, Dong et al. (2006) find that higher bidder valuation levels based on price-to-book value of equity and price-to-residual income value of equity are associated with a greater tendency to use stock as the means of payment and higher bid premia. Just as conventional
AN US
valuation levels may influence mergers, perceived valuation levels based on 52-week high prices may also influence mergers, especially under circumstances in which merger outcomes are being driven by decision agents who are faced with limited information and uncertainty. Following the
M
arguments in Shleifer and Vishny (2003) and Dong et al. (2006), the perceived valuation hypothesis implies that bidders with higher reference price ratios are associated with higher bid
ED
premia, higher target announcement returns, a greater (lower) likelihood of using stock (cash) as the sole consideration, a lower likelihood of being involved in contentious acquisitions, and a
PT
greater likelihood of successfully completing the deal.
CE
3. Sample and data
Our sample of acquisitions is drawn from the Securities Data Corporation (SDC). The
AC
deals are announced during the 1981–2014 period; the acquirers are publicly traded U.S. firms and have stock price data from CRSP and accounting data from Compustat; we exclude deals with transaction values lower than $1 million (as reported in SDC) or deals with relative sizes lower than 5% or higher than 200% of the acquiring firms’ equity; and deals are also excluded if
Page 12
ACCEPTED MANUSCRIPT
the acquirer owns more than 50% of the target prior to the announcement. We further require that the data necessary to conduct the main analyses are available. The final sample includes a total of 19,119 transactions. Panel A of Table 1 presents basic information about the sample. The average transaction value is $453.31 million; out of
CR IP T
these, 75% involve private targets (private firms or subsidiaries of other companies); 19% of the transactions are paid with stock, 23% by cash, and 58% with mixed instruments (cash, stock, debt, convertibles, earn-outs, etc.). Only 4% of the deals are tender offers, 2% are recorded as hostile, and 88% of deals are successful. The average acquirer reference price ratio is 0.81. As
AN US
defined below, this indicates that, on average, the acquirers’ pre-announcement stock price is 81% of its 52-week high.
Panel A of Table 1 also lists the year-by-year averages for these variables. Most of these
M
numbers vary over time, and there is a noticeable increase in the number of deals and tendency to make acquisitions with stock in the mid to late 1990s. This pattern is consistent with the merger
ED
wave over this period. There is, however, no apparent trend in the average reference price ratio. [Insert Table 1 here]
PT
The primary focus of our analysis is the impact of the acquirer’s reference price ratio
CE
(RPR, hereafter) on the stock market reaction to a merger announcement and deal characteristics. We define the RPR following George and Hwang (2004). Specifically, the RPR on day t is the
AC
ratio of the closing price Pt to the highest closing price over the previous 252 trading days (including day t), with prices adjusted by stock splits and dividends. RPR is between zero and one. A company with a stock price near its 52-week high will have an RPR close to one. In our main analysis, we choose the RPR as of the sixth day prior to the announcement, t – 6. The chart below illustrates the timing of measuring the RPR and the event window, where day t = 0 is the
Page 13
ACCEPTED MANUSCRIPT
announcement date. As shown in Panel B of Table 1, our sample has an average RPR of 0.81, a median of 0.86, a range of 0.01 to 1.0, and an interquartile range of 0.72 to 0.95. Event window [-5, +1]
CR IP T
Reference Price Ratio (RPR) measured
…
-6 -5 -4 -3 -2 -1 0 +1
…
We measure the acquirer’s cumulative abnormal return (CAR, hereafter) over the sevenday event window [-5, +1], where day 0 is the announcement date. The abnormal return is based
AN US
on a market model estimated over the [-370, -253] days relative to the announcement date. This approach follows Harford (1999). As shown in the robustness section, our main results are robust to alternative measures of abnormal returns.
Panel B in Table 1 shows the summary statistics of the main variables for our sample.
M
Our sample has an average acquirer CAR of 1.50%, with a median of 0.57%, ranging from -
ED
22.93% to 36.57%, and an interquartile range of -3.46% to 5.37%. The average numbers are similar to those reported in the literature based on similar samples of both publicly traded and
PT
private targets (e.g., Moeller et al., 2004). All variables are defined in Appendix A. 4. Acquirer announcement-period and long-run abnormal returns
CE
4.1.Acquirer cumulative abnormal returns and acquirer reference price ratio
AC
As an initial step in examining the potential relation between the acquirer RPR and acquirer CAR, we first separate our sample into acquisitions of publicly traded targets and private (private or subsidiary) targets and then sort each listing-status sample into two equal groups (by year) based on RPR. We categorize acquisitions by the target listing status for two reasons. First, the literature documents different acquirer CARs, and varying influences of other variables on acquirer CARs, in acquisitions of public versus private targets (e.g., Fuller et al., Page 14
ACCEPTED MANUSCRIPT
2002). In addition, to the extent the information environment associated with a public acquisition is better, we expect the tendency for investors to be influenced by reference prices to be lower than in private acquisitions, a natural setting for testing the perceived valuation hypothesis.5 The Low (High) RPR group includes acquisitions with acquirers’ RPR below (above) the
CR IP T
sample median within its listing-status group. Table 2 shows basic summary statistics of the two groups. Panel A of Table 2 shows that, in the whole sample, the Low RPR group has an average RPR of 0.677 and the High RPR group has an average RPR of 0.937, implying a difference of
of private targets (0.829 versus 0.800).
AN US
0.260. The RPR ratios for acquirers of public targets tend to be slightly higher than for acquirers
[Insert Table 2 here]
The first column in Panel B of Table 2 shows that the Low RPR group has an average
M
CAR of 2.021%. The High RPR group has an average CAR of 0.982%, implying a significant difference of -1.040% (t=-7.66).6 Combining the two differences in means between the Low RPR
ED
and High RPR groups from Panels A and B suggests that acquirer CARs increase by 4.00% (1.040%/0.260) if the RPR decreases from one to zero. Put differently, acquirers with pre-
PT
announcement prices at half of their 52-week highs earn about 2.0% higher abnormal returns, on
CE
average, than those with pre-announcement prices at their 52-week highs. For convenience, we refer to this negative relation between acquirer RPR and acquirer CAR as the reference price
AC
effect.7
5 Consistent with this view, Officer (2007) finds that private targets (private independent firms or subsidiaries of other firms) tend to receive lower premia, probably due to lower liquidity of such assets. 6 Unreported for brevity, the medians have a similar pattern and the Wilcoxon test on the distributions of the two groups yields a test statistic of -5.26, which is significant at the 1% level. 7 We obtain similar results based on reference price ratios of the 39-, 26-, and 13-week highs. This finding is consistent with Baker et al. (2012), who find that target offer prices cluster on multiple past peak prices. For simplicity, our main analysis focuses on the 52-week high as the reference price.
Page 15
ACCEPTED MANUSCRIPT
Fig.1 illustrates the acquirer cumulative average abnormal returns of the Low and High RPR groups from five days before to 20 days after the announcement date for acquisitions of private targets (Panel A) and acquisitions of public targets (Panel B). The end of day t – 6 is the starting date. For private targets, in addition to the salient gap between the two lines over the
CR IP T
event window of [-5, +1], the figure shows that the reference price effect persists for at least 20 days after the announcement. In the public subsample, there is a noticeable, but smaller gap between the two lines and the lines cross within ten days after the announcement.8 [Insert Fig. 1 here]
AN US
Panel B of Table 2 shows a substantially stronger reference price effect in the private subsample compared to the public subsample. Specifically, within the private sample (the 4th column), the high RPR acquirers earn an average CAR of 1.726% and low RPR acquirers earn
M
an average CAR of 2.916%, a significant difference of -1.190% (t=-7.48). In the public subsample (the 7th column), high RPR acquirers earn an average CAR of -1.201% and low RPR
ED
acquirers earn an average CAR of -0.618%, a smaller difference of -0.582% with weaker statistical significance (t=-2.36). This weaker reference price effect is consistent with the
PT
perceived valuation hypothesis as the information environment for deals involving public targets
CE
is much better than private targets and, as such, investors are influenced less by reference prices. Given the possibility that the reference price effect is driven by conventional valuation
AC
levels, we perform additional two-way sorts on market-to-book (M/B) ratios to determine whether the reference price effect exists within valuation-level categories. The second and third
8
Over the longer event window of [-5, +20] in the private subsample, the average acquirer CAR is 2.403% for the Low RPR group and 1.633% for the High RPR group, implying a difference of -0.770% (t=-2.37); in the public subsample, the average CAR is -2.149% for the Low RPR group and -1.756% for the High RPR group, implying a difference of 0.393% (t=0.78). Page 16
ACCEPTED MANUSCRIPT
columns in Panel B of Table 2 show these two-way sort results.9 They indicate that low RPR acquirers tend to earn higher CARs than the high RPR acquirers regardless of M/B level. For example, within the High M/B acquirers, those with a low RPR earn an average CAR of 1.787% while those with a high RPR earn an average CAR of 0.945%, a significant difference of -
CR IP T
0.842% (t=-4.18). The pattern is similar within the low M/B acquirer group. The 5th and 6th columns show a similar pattern in the private subsample. In the public subsample, however, the differences between the low and high RPR groups are substantially smaller. For example, within the High M/B group, the last column in Panel B of Table 2, low RPR acquirers earn a higher
AN US
CAR than high RPR acquirers (-0.932% vs. -1.600%), but the difference is only marginally significant (t=-1.78).
4.2.Multiple regressions of acquirer cumulative abnormal returns
M
Table 3 presents four ordinary least squares (OLS) regressions examining the impact of acquirer RPR on acquirer CAR. All regressions control for Fama and French (1997) industry and
ED
year fixed effects. To alleviate the impact of outliers we winsorize all continuous variables at the 1st and 99th percentiles. The t-statistics are based on standard errors clustered by firms. In the first
PT
regression, we regress acquirer CAR on acquirer RPR and a list of variables designed to control
CE
for various deal and firm characteristics. We include the most frequently controlled variables among the more recent studies in the M&A literature for which we have data (e.g., Fuller et al.,
AC
2002; Moeller et al., 2004, 2005; Masulis et al., 2007; Officer et al., 2009; Golubov et al., 2012; Harford et al., 2012; El-Khatib et al., 2015; Golubov et al., 2015). In the first regression, the coefficient on acquirer RPR is -5.487 (t=-8.56), indicating a
strong reference price effect. That is, acquirer CARs are higher (lower) when the acquirer’s pre-
9 Our proxy for valuation levels is market-to-book ratios for the univariate analysis shown in Panel B of Table 2. We consider other proxies in subsequent analysis.
Page 17
ACCEPTED MANUSCRIPT
announcement stock price is well below (near) its 52-week high even after controlling for the comprehensive list of firm and deal characteristics. It is worth noting that the reference price effect is economically significant. For a one standard deviation increase in the RPR (0.1844), the acquirer CAR decreases by 1.012% (=0.1844*5.487), or about 11% (=1.012/9.397) of its
CR IP T
standard deviation. By comparison, a one standard deviation change in firm size induces an 8.5% standard deviation change in acquirer CAR.
The coefficients on the control variables are largely consistent with the literature (Chang, 1998; Fuller et al., 2002; Faccio et al., 2006; Officer et al., 2009; Golubov et al., 2015). As
AN US
expected, private acquisitions earn higher acquirer returns. Notably, with both private and public deals included in the sample, the coefficient on stock as a method of payment is negative. However, using stock tends to earn higher acquirer returns in private transactions, as indicated by
M
the positive and significant coefficient on the interaction term that equals one for stock transactions involving private deals (e.g., Fuller et al., 2002; Slovin et al., 2005). Relative size
ED
has a positive and significant coefficient. M/B has a significant negative coefficient, consistent with the valuation hypothesis that acquirers with higher valuations earn lower returns at the
PT
M&A announcements. 10 Consistent with Moeller et al. (2004), larger acquirers earn lower
CE
acquisition returns. Acquirer returns are higher if an acquisition is preceded by a long dormant period (a year), consistent with Cai et al. (2011). The coefficient on tender offers is positive and
AC
significant, that on hostile deals is negative and significant, and that on toeholds is not
10 The variable M/B used in the regression analysis is measured at the fiscal year-end before the acquisition announcement. However, we find similar results when we measure M/B using stock price on day t-6 and book value equity reported in the most recent quarter prior to the announcement. Further, in unreported analysis, we use other proxies for conventional valuation levels, including a market-to-intrinsic value ratio, where intrinsic value is estimated using a residual income model following Edwards and Bell (1961) and Ohlson (1995), and a measure developed by Rhodes-Kropf et al. (2005) with the Hertzel and Li (2010) adjustment. In all cases, we find a reference price effect similar to what is reported in Table 3.
Page 18
ACCEPTED MANUSCRIPT
significant. Past return has a positive coefficient, consistent with the view that bidders with higher management quality earn greater acquisition returns. [Insert Table 3 here] The second and third models in Table 3 show the results of estimating the regression
CR IP T
model for the private and public subsamples separately. Consistent with reference prices being more important under circumstances involving greater uncertainty or a more opaque information environment, the reference price effect is more pronounced in the private subsample. The coefficient on RPR is -6.286 (t=-8.66) in the private subsample regression whereas it is -2.581
AN US
(t=-1.97) in the public subsample. The coefficients on the control variables are largely consistent with the results for the whole sample except as follows. Notably, the coefficient on stock as a method of payment is positive and statistically significant in the regression of private
M
acquisitions but negative (although not statistically significant) in the regression of public transactions. Cash deals involving public targets earn significantly higher acquirer returns.
ED
Relative size has a positive and significant coefficient in private acquisitions but the coefficient is negative and not statistically significant in public acquisitions. These results are consistent
PT
with the notion that investors perceive acquirers of private targets as getting a favorable deal and
CE
the benefit of the deal being amplified with the relative size of the target and by paying stock (Fuller et al., 2002). M/B has a negative coefficient, but it is not statistically significant in the
AC
subsamples.
In the fourth model of Table 3, we include the M/B and reference price ratio of the target
firm (Target Ln(M/B) and Target RPR, respectively) in the analysis. Due to the additional data requirement, this analysis is limited to the subsample of acquisitions involving public targets that are covered in CRSP. We find the coefficient on the acquirer’s RPR remains negative, but it is
Page 19
ACCEPTED MANUSCRIPT
only marginally significant (t=-1.92). The weaker acquirer reference price effect is consistent with the perceived valuation hypothesis as the information environment for these transactions is better and investors are thus less affected by reference prices. The coefficient on target M/B is negative and marginally significant (t=-1.70), consistent with Dong et al. (2006). The coefficient
CR IP T
on the target’s RPR is positive and significant (t=4.28). According to Dong et al. (2006), target valuation ratios can influence acquirer CARs via two potential mechanisms. On the one hand, to the extent targets are perceived as misvalued and an acquisition announcement triggers a reassessment of valuation levels, we expect a negative relation between target valuation levels
AN US
(either target RPR or target M/B) and acquirer CAR. On the other hand, if a higher target RPR is associated with lower bid premia (see Table 6 in Section 5.1), as indicated by Baker et al. (2012), and lower bid premia are associated with higher acquirer CARs, we expect a positive relation
M
between target RPR and acquirer CAR. The positive and significant coefficient on target RPR in the fourth model suggests that any negative influence from the reassessment of valuation levels
ED
is more than offset by the positive influence due to lower bid premia. To sum up, the first model in Table 3 establishes a strong and economically meaningful
PT
reference price effect in a multiple regression setting. By separating the sample into transactions
CE
involving public and private targets, the second, third, and fourth models in Table 3 provide evidence the reference price effect is stronger among private deals, consistent with reference
AC
prices having a greater influence when there is less information available. 11
11
Similar results are obtained when we exclude the potentially endogenous variables, stock, cash, tender offer, and hostility. Also, it should be noted that the results shown in Table 3 include both successful and unsuccessful deals. When the analysis is restricted to successful deals (those that are ultimately completed, according to SDC; n=16,806), the coefficient on RPR in the first model is -4.461 with a t-stat of -6.79. When the analysis is limited to unsuccessful deals (n=2,313), the coefficient on RPR is -12.891 with a t-stat of -6.03. This suggests that, although the reference price effect is stronger in deals that are unsuccessful, it is also strong in successful deals. Because the ultimate outcome of a given deal is not known at the time of announcement, however, we conduct the bulk of our analysis using all announced acquisitions. The results are similar regardless of whether we include unsuccessful deals. Page 20
ACCEPTED MANUSCRIPT
We also examine several alternative reference price ratios as in Baker et al. (2012). Specifically, we construct new reference price ratios relative to the 75th, 90th, and 95th percentiles, respectively, of acquirer prices over the 52-weeks preceding the sixth day before the announcement date. For convenience, we call them RPR75, RPR90, and RPR95, respectively.
CR IP T
They are naturally highly correlated with the original acquirer RPR, with correlation coefficients of 0.676, 0.881, and 0.956, respectively. When these alternative RPRs piecewise enter the regression of the full sample without the RPR, they all have significant negative coefficients. When these alternative RPRs piecewise enter the regression with the original RPR, however,
AN US
their coefficients become positive with weak statistical significance while the coefficients on the original RPR remain negative and significant.12
These falsification tests highlight the importance of the 52-week high, not some highly
M
correlated but less salient numbers that affect investors’ valuation perceptions. When the sample is separated into public and private acquisitions, the falsification tests for the private acquisitions
ED
yield very similar conclusions as the whole sample. In the subsample of public acquisitions and the subsample of public acquisitions covered in CRSP, however, the coefficients on RPR75,
PT
RPR90, and RPR95 are negative but not statistically significant when they piecewise enter the
CE
regression without RPR. When RPR is present, it is significant only in the regression with RPR75, consistent with reference prices playing a weaker role in public acquisitions. To the
AC
extent acquirer reference prices do play a role in these acquisitions, it seems to be the 52-week high.
Further, RPRs based on 39-, 26-, and 13-week highs yield similar results to those in the
regression that includes all acquisitions in Table 3. They indicate that, in addition to the 12
We also construct RPR99, the ratio of stock price to the 99 th percentiles of the past 252 daily stock prices. Due to the extremely high correlation (0.993) between RPR99 and the original RPR, the regression with them both present has serious multicollinearity problems. Neither coefficient is significant. Page 21
ACCEPTED MANUSCRIPT
commonly studied 52-week highs, some investors may also anchor on peak price levels over the past 39, 26, or 13 weeks (see also Baker et al., 2012). Regardless of the exact peak prices used as the anchor, the results support the same economic implication: Investor reaction to acquisition announcements seems influenced by recently observed peak prices as reference prices. These
CR IP T
unreported results are available upon request. 4.3.The role of uncertainty
The perceived valuation hypothesis predicts that the acquirer reference price effect is stronger among deals involving greater uncertainty, more opaque information environments, and
AN US
less sophisticated investors. Our analysis to this point finds support for this prediction by documenting a stronger reference price effect in acquisitions of private versus public targets. To further test these predictions, we examine the extent to which the reference price effect is
M
influenced by additional proxies for information uncertainty and individual investor ownership levels.
ED
We consider four proxies for information uncertainty: The volatility of acquirer stock returns, the number of analysts following the acquirer, the relative size of the acquisition, and an
PT
indicator variable for non-cash methods of payment. Stocks in general, including stocks of
CE
acquirers, are more difficult to value if their prices are more volatile. To the extent stock analysts produce valuation-relevant information, acquirers with fewer analysts following them have a
AC
more opaque information environment and are more difficult to value. Similarly, if the transaction value relative to the acquirer equity value is larger, the acquisition will have a greater impact on the acquirer’s stock value, posing greater uncertainty to the valuation process. Last, but not least, non-cash payment adds a level of complexity not present in cash deals and could
Page 22
ACCEPTED MANUSCRIPT
indicate the acquiring firm’s uncertainty regarding the value of the acquired assets (Hansen, 1987; Officer et al., 2009). The perceived valuation hypothesis predicts a stronger reference price effect for acquirers with more volatile stocks, for acquirers followed by fewer analysts, for deals of greater relative
CR IP T
size, and for deals that involve non-cash payment. We test these implications in multiple regressions by adding interaction terms. Further, because we document a substantial difference in reference price effects for acquisitions of private firms versus acquisitions of public firms, we analyze the influence of these proxies for information uncertainty separately for the two types of
AN US
acquisitions. Table 4 presents these regression results.13
In Models 1 and 6 of Table 4, we add both Sigma and RPR*Sigma to the baseline model, where Sigma measures volatility as in Zhang (2006). The full regression specifications follow
M
those of Table 3. For brevity, coefficients and t-statistics of some variables are not shown. For ease of interpretation we subtract the sample mean from the variable Sigma. The perceived
ED
valuation hypothesis implies a negative coefficient for RPR*Sigma. In private acquisitions, Model 1 of Table 4 reports the coefficient on RPR is -4.697 (t=-5.73), indicating a significant
PT
reference price effect for acquirers with average Sigma values and the coefficient on RPR*Sigma
CE
is -0.455 (t=-2.90), confirming the prediction. In public acquisitions (Model 6), the coefficients on RPR and RPR*Sigma are -2.927 (t=-1.97) and -0.534 (t=-1.94), respectively. These
AC
coefficients are also consistent with the perceived valuation hypothesis but notably weaker than in the sample of private acquisitions. Sigma itself has a positive and significant coefficient in the sample of private acquisitions but is not significant in the sample of public acquisitions.
13
For the subsample of public deals involving targets that are covered in CRSP, we also add the target valuation proxies (target RPR and target M/B) in these regressions and find very similar results to those for the “Public” subsample, although the statistical significance is expectedly weaker. For brevity, these results are not reported but available upon request. Page 23
ACCEPTED MANUSCRIPT
In Models 2 and 7 of Table 4, we add an interaction term, RPR*Few analysts, where Few analysts is a binary variable equal to one if the number of analysts following the acquirer over the six months prior to the announcement month is below the sample median. The perceived valuation hypothesis predicts a negative coefficient on this interaction term: Acquirers followed
CR IP T
by fewer analysts will have a stronger (i.e., more negative) reference price effect. In Model 2 of Table 4 the coefficient on RPR is -3.480 (t=-3.75) for private acquisitions, indicating a strong reference price effect for acquirers with an above-median number of analysts. That is, even for acquirers followed by many (above-median) analysts, there is a strong reference price effect in
AN US
acquisitions of private targets. For acquirers followed by few (below-median) analysts, the reference price effect is even stronger as the coefficient on RPR*Few analysts is a significant 4.822 (t=-4.04). We also report a positive coefficient for the binary variable Few analysts in the
M
sample of acquisitions involving private targets. In public acquisitions (Model 7), the coefficient on RPR is not statistically significant, suggesting there is little evidence of a reference price
ED
effect when a firm followed by many analysts acquires a publicly traded target, a circumstance in which investors are least likely influenced by reference prices when assessing acquirer stock
PT
price. The coefficient on RPR*Few analysts is -3.815 (t=-1.87), indicating a stronger reference
CE
price effect in acquisitions of public targets when the number of analysts following the acquirer is below the median.
AC
We add the interaction term between RPR and relative size to Models 3 and 8. The perceived valuation hypothesis predicts a negative coefficient on this interaction term. In private acquisitions, the coefficient on the interaction term is -6.623 (t=-3.42), further supporting the perceived valuation hypothesis. Similar to Model 1, the relative size variable is demeaned. Thus, the coefficient on RPR itself, -6.290 (t=-8.66) represents the reference price effect in private
Page 24
ACCEPTED MANUSCRIPT
acquisitions with an average relative size. In public acquisitions (Model 8), the coefficient on RPR is not significant, but the coefficient on the RPR*Rel. size variable is -5.461 (t=-2.10), indicating that, although there is no significant reference price effect in public acquisitions of an average relative size, there is an interaction effect that is consistent with the perceived valuation
CR IP T
hypothesis. Models 4 and 9 of Table 4 include an interaction term RPR*Non-cash. The perceived valuation hypothesis implies a negative coefficient on this term. Reported in Model 4 of Table 4, the coefficient on RPR*Non-cash is -3.364 (t=-2.40) in private acquisitions, confirming the prediction. The significant negative coefficient on RPR in Model 4 (-3.564 with t=-2.80) for
AN US
private acquisitions suggests there is a strong reference price effect even among cash-only transactions. In public acquisitions, the coefficient on RPR is -3.684 (t=-1.83), indicating an overall reference price effect, but the lack of a statistically significant coefficient on RPR*Non-
M
cash suggests that the reference price effect is unrelated to the method of payment in acquisitions of public targets.
ED
[Insert Table 4 here]
4.4.The role of investor sophistication
PT
To further explore the implications of the perceived valuation hypothesis, we examine
CE
whether less sophisticated investors are more likely to be influenced by the acquirer’s 52-week high price when estimating the valuation implications of an acquisition. Generally, empirical
AC
evidence suggests that individual investors are prone to behavioral biases (e.g., Odean, 1998; Barber and Odean, 2008) and that institutional investors tend to be more sophisticated (e.g., Nofsinger and Sias, 1999; Bartov et al., 2000). Therefore, we expect a stronger reference price effect when individual investor ownership of the acquirer is relatively high. Following other studies, we treat non-institutional investor ownership as a proxy for individual investor
Page 25
ACCEPTED MANUSCRIPT
ownership (e.g., Nofsinger and Sias, 1999; Gompers and Metrick, 2001; Sias and Whidbee, 2010). Institutional ownership data are from Thomson Reuters Institutional Holdings (13F). Total institutional ownership of the acquirer is measured at the most recent quarter-end prior to
CR IP T
the announcement month. The variable “Indiv. own.” is one minus the institutional ownership. If an acquirer’s institutional ownership data are missing it is assigned as zero and “Indiv. own.” is equal to one. Admittedly, this measure of investor sophistication is potentially problematic and is not necessarily indicative of investors who trade in an acquirer’s stock in the days surrounding
AN US
announcement dates. However, we suggest that the investing environment surrounding acquirer stocks with relatively high individual (low institutional) ownership is more exposed to potential behavioral biases.
M
The results of the regressions incorporating individual ownership levels are shown in Models 5 and 10 of Table 4. We demean the Indiv. own. variable and add both Indiv. own. and
ED
RPR*Indiv. own. to the baseline model. The perceived valuation hypothesis implies a negative coefficient for RPR*Indiv. own. In private acquisitions (Model 5), the coefficient on RPR is
PT
negative and significant, suggesting a strong reference price effect for acquirers with average
CE
individual investor ownership; the coefficient on the interaction term is -0.112 (t=-5.40), indicating that the reference price effect is stronger when individual investors own a greater
AC
percentage of acquiring firms’ stock. The coefficient on Indiv. own. is positive and significant. In public acquisitions (Model 10), the coefficients on RPR and the interaction term are -2.187 (t=1.69) and -0.103 (t=-2.82), respectively, suggesting a marginally significant overall reference price effect and that the effect increases with acquirer individual ownership.
Page 26
ACCEPTED MANUSCRIPT
Overall, the evidence in Table 4 is consistent with the perceived valuation hypothesis. The reference price effect is stronger in acquisitions involving opaque information environments, when there is greater uncertainty about acquirer valuation, and when less sophisticated investors hold a higher percentage of acquirer stocks.
CR IP T
4.5.Long-run acquirer returns
To further distinguish between the perceived valuation hypothesis and other potential explanations for the reference price effect, we examine longer horizon returns. If the reference price effect is being driven by perceived valuation levels based on 52-week high prices, an
AN US
irrational behavioral bias, we expect the short-horizon return pattern documented above to be reversed over the long term as the bias is eventually corrected, and we should observe a positive relation between long-horizon acquirer returns and acquirer RPR. On the other hand, if the
M
reference price effect is capturing the influence of some unidentified fundamental factor associated with acquisitions, then the short-horizon return pattern should be permanent and there
ED
should be no relation between RPR and long-horizon returns. Finally, if the RPR is acting as another proxy for conventional valuation levels, which are associated with poor long-horizon
PT
acquirer returns (Rau and Vermaelen, 1998; Fu et al., 2013), we should observe a negative
CE
relation between long-horizon returns and the RPR. Measuring long-horizon returns is inflicted with methodological issues and no consensus
AC
optimal method is proffered in the literature (e.g., Lyon et al., 1999; Mitchell and Stafford, 2000). Further, tests of price efficiency inevitably suffer from a bad-model problem (Fama, 1998). We first follow the two approaches commonly adopted in the literature: buy-and-hold abnormal returns (BHAR), and calendar-time portfolio returns (CTPR). To avoid the bad-model problem, we also examine subsequent earnings announcement abnormal returns. While none of
Page 27
ACCEPTED MANUSCRIPT
the approaches is perfect, the collective evidence provides a more comprehensive view of longrun returns. As in Table 2, we form two equal subsamples (by year) on pre-announcement acquirer RPR and examine their average BHAR and CTPR. The holding period for both methodologies is
CR IP T
one year following the month of announcement. To alleviate the influence of outliers we winsorize the BHARs at the 1st and 99th percentiles, although results are similar without winsorization. Panel A of Table 5 presents the BHAR results for the whole sample and the Low and High RPR subsamples. The first column shows that the acquirers in our sample earn -
AN US
3.625% over the one-year period following the announcement month.14 The magnitude is in line with existing studies (e.g., Bouwman et al., 2009; Ma et al., 2011; Duchin and Schmidt, 2013). More striking are the different BHARs of the Low and High RPR groups. As defined in
M
Table 2, Low RPR (High RPR) represents lower (higher) RPR acquirers. The mean one-year BHAR is -5.889% for Low RPR and -1.369% for high RPR acquirers, resulting in a difference of
ED
4.520%. The return difference in the BHAR clearly exhibits a reversal of the short-term return pattern shown in Table 2. Recall that Low RPR acquirers earn significantly higher
PT
announcement-period abnormal returns, but the long-horizon results shown in Table 5 indicate
CE
that the Low RPR acquirers earn significantly lower long-horizon abnormal returns than the High RPR acquirers.
AC
Panel B of Table 5 shows the Fama and French (2015) five-factor alphas of the calendartime portfolios for the whole sample of acquirers and for the Low and High RPR subsamples. 14 The t-statistics used here to determine statistical significance assume independence across acquisitions deals. In reality, however, these deals are clustered and thus BHARs might be cross-correlated. To address the issue of sample clustering and cross-correlation, the literature (e.g., Lyon et al., 1999; Mitchell and Stafford, 2000) suggests the use of bootstrapping to derive the empirical distribution. This approach is adopted in Bouwman et al. (2009). We follow their approach and find that the t-stat for the whole sample BHAR is -9.15 (the empirical distribution indicates statistical significance at the 1% level) for the one-year BHAR. Regardless of how the statistical significance is measured, the pattern of reversal in BHAR is clear.
Page 28
ACCEPTED MANUSCRIPT
The measurement period is one year following the month of announcement. As the numbers in Panel B are monthly alphas and those in Panel A are BHAR, the magnitude of the numbers in Panel B is on a smaller scale. For the whole sample, the monthly alpha is -0.057%. The alphas are -0.165% for the Low RPR group and 0.017% for High RPR, with a difference of 0.181%
CR IP T
(t=1.29). The second (third) columns of Panels A and B present similar evidence for the subsample of acquisitions involving only private (public) targets. [Insert Table 5 here]
To further shed light on the long-term stock price behavior following M&A events, we
AN US
examine the abnormal returns surrounding the earnings announcements over the subsequent oneyear period. This approach is motivated by the anomalies literature. For example, Chopra et al. (1992) find that long-term poorly performing stocks exhibit an overreaction pattern in the long-
M
term subsequent time period as well as in short-term abnormal returns around quarterly earnings announcements, suggesting that part of the long-term return pattern is reflected in the short-term
ED
earnings announcement periods. Jegadeesh and Titman (1993) use a similar approach to study the momentum anomaly; La Porta et al. (1997) use this approach to study the value premium;
PT
and Titman et al. (2004) use it to study the capital investment anomaly. Engelberg et al. (2017)
CE
report that, for 97 return anomalies, anomaly returns are seven times higher on quarterly earnings announcement days than days of no news, suggesting that earnings announcements preserve the
AC
return pattern over the long term. The central premise applied to our context is that the mispricing following M&A events is corrected partly in the market response to subsequent earnings announcements. If, according to the perceived valuation hypothesis, high RPR acquirers earn higher returns than low RPR acquirers over the long horizon, we expect to observe the same pattern in the abnormal returns surrounding the earnings announcements over the
Page 29
ACCEPTED MANUSCRIPT
contemporaneous period. Because earnings announcement-period abnormal returns are measured over a very short time (three-day event window in our analysis), the bad-model problem is less serious (Fama, 1998). Specifically, for each acquisition we record the four earnings announcement dates (from
CR IP T
Compustat quarterly file) over the one-year period following the M&A announcement. For each of the earnings announcements, we measure the abnormal returns over the [-1, +1] window, adjusted by the CRSP equal-weighted market returns (results are similar when adjusted by valueweighted market returns). We examine the four-quarter averages. Panel C presents the results.
AN US
For the whole sample, the average earnings announcement-period return is negative (positive) for the low (high) RPR group, leading to a significant difference of 0.204 (t=3.16). The private and public subsamples exhibit a similar pattern.
M
For comparison, Panels A, B, and C also show results for subsamples formed on two conventional valuation proxies, M/B and RKRV. RKRV is based on a measure developed by
ED
Rhodes-Kropf et al. (2005) with the Hertzel and Li (2010) adjustment. The RKRV approach generates three measures of valuation depending on their regression specifications. We show
PT
results based on their most sophisticated model, but we find similar results using their other two
CE
specifications.15 Appendix B contains the detailed definition. Across three samples (the whole sample, and the private and public subsamples), two conventional valuation proxies (M/B and
AC
RKRV), and three long-horizon return measures (BHAR, CTPR, and subsequent earnings announcement returns), Table 5 generates a total of 18 (3x2x3) return differentials between the high and low valuation groups, among which 17 are negative, with 12 significantly so, and the 15
Unreported for brevity, we find six of the nine BHAR, CTPR, and earnings announcement return differentials between the high and low valuation groups are negative when subsamples are formed on sorts using a valuation level proxy based on a market-to-intrinsic value ratio, where intrinsic value is estimated using a residual income model following Edwards and Bell (1961) and Ohlson (1995), and more specifically, Dong et al. (2006) and Ma et al. (2011). However, none of the return differentials is statistically significant. Page 30
ACCEPTED MANUSCRIPT
remaining positive one is not statistically significant. The evidence based on M/B and RKRV in general is consistent with conventional valuation theory in that overvalued acquirers earn lower returns in the long term. To explore the relation between long-horizon acquirer returns and acquirer RPR (along
CR IP T
with conventional valuation-level proxies) further and account for the impact of other deal and firm characteristics, we also estimate OLS regressions of BHARs on acquirer RPR and the list of control variables used in the baseline regressions of acquirer CAR in Table 3. The results are presented in Panel D of Table 5. For the sample with all targets, the first column of Panel D
AN US
shows a positive and significant coefficient on acquirer RPR and a negative, marginally significant coefficient on acquirer M/B. This pattern carries over to the regressions for the private, public, and CRSP subsamples as well, shown in the last three columns of Panel D.
M
Consistent with Panel A, acquirer RPR and our conventional valuation-level measure, M/B, have exactly opposite influences on long-term returns, albeit without a statistically significant
ED
coefficient on M/B in the public and CRSP subsamples. Specifically, in the full sample and private subsample, acquirers with higher M/B earn lower long-term returns than those with lower
PT
M/B, consistent with conventional valuation theory and the market-timing theory of mergers
CE
(Shleifer and Vishny, 2003). At the same time, however, acquirers with higher RPRs earn higher long-term returns than those with lower RPRs in the full sample and all subsamples, consistent
AC
with the perceived valuation hypothesis.16 The sharp contrast between the results based on RPR and those based on M/B and RKRV
suggests that RPR is different from conventional valuation proxies. The evidence supports the 16
Similar results are obtained when we exclude the potentially endogenous variables, stock, cash, tender offer, and hostility. Likewise, the coefficients on RPR remain positive and significant with similar magnitude when we replace M/B with RKRV or our market-to-intrinsic value estimate. Finally, we also find similar results when we measure M/B by using stock price on day t-6 and book value equity reported in the most recent quarter prior to the announcement. Page 31
ACCEPTED MANUSCRIPT
perceived valuation hypothesis that investors irrationally view acquirers’ 52-week highs as a gauge of “fair” stock value. Consequently, their reactions to acquisition announcements are influenced by the 52-week highs in the short-term, but the influence is eventually corrected in the
5. Reference prices and deal characteristics
CR IP T
long-term.
The analysis to this point indicates that the acquirer’s RPR influences acquirer stock prices at the announcement, particularly when there is less information or greater uncertainty about the impact of the acquisition on firm value. In this section, we examine the impact of
AN US
acquirer RPR on deal characteristics, including bid premia and target announcement returns in Section 5.1 and the method of payment and deal combativeness in Section 5.2. When analyzing the influence of acquirer RPR on deal characteristics, we also consider
M
the influence of acquirer M/B as well as the RPR and M/B of the target wherever possible. Accordingly, we interpret our results along with those reported in Dong et al. (2006) and,
ED
wherever relevant, those in Baker et al. (2012). Specifically, the evidence in Dong et al. (2006) indicates that higher acquirer valuation, as measured by M/B, is related to a greater likelihood of
PT
using stock as the method of payment, lower likelihood of using cash, and a lower likelihood of
CE
acquiring through tender offers. Higher target valuation is associated with a greater likelihood of using stock, lower likelihood of using cash, lower likelihood of using tender offers and deal
AC
hostility, and a higher probability of deal success. 5.1. Bid premia and target announcement-period abnormal returns Table 6 presents five OLS regressions of bid premium on acquirer RPR, acquirer M/B,
target RPR, target M/B, and the other control variables used in Table 3 after excluding the potential endogenous variables, stock, cash, tender offer, and hostility. The deals in this analysis
Page 32
ACCEPTED MANUSCRIPT
are limited to those with target information available on CRSP and with available information on bid premia. The first column shows the regression results for all deals with available data. The second and third columns show results for acquisitions sorted on the acquirers’ sigma to allow for the possibility that uncertainty about acquirer valuation may influence the impact of valuation
CR IP T
levels on bid premia.17 Similarly, although the method of payment is potentially endogenous, the fourth and fifth columns present results for the cash and non-cash subsamples to allow for the possibility that acquirer valuation levels may not influence bid premia when cash is the only method of payment.
AN US
The first regression in Table 6 indicates that the coefficient on acquirer RPR is positive but not statistically significant in the whole CRSP sample. In the subsample of deals with high uncertainty about acquirer valuation levels (High sigma) and the subsample of deals involving
M
non-cash forms of payment (Non-cash), however, the coefficients on acquirer RPR are positive and at least marginally significant. This suggests that when the valuation level of the acquirer’s
ED
stock is difficult to determine or the valuation level of the acquirer is important to the bid premium negotiation process, the acquirer’s 52-week high influences bid premium negotiations.
PT
When the valuation level of the acquirer is easier to determine or is not directly relevant to the
AC
CE
negotiation process, however, the acquirer’s RPR is less likely to influence bid premiums: The
17
Bid premia are determined by managers, consultants, and other relatively sophisticated decision agents, and these decision agents are likely to have access to information not available to acquirers’ investors. To the extent these parties are influenced by the bidders’ 52-week high prices, it is likely to depend more directly on the degree of uncertainty they face in estimating and negotiating the ‘fair’ bid premium. The most direct measure of uncertainty for these decision makers is the acquirers’ sigma. Thus, for the purposes of analyzing the impact of the acquirers’ RPR on bid premia, we focus on the acquirers’ sigma as our proxy for the information environment. The other information environment measures (number of analysts following the acquirer, relative size of the acquisition, and individual investor ownership) reflect the information environment faced by acquirers’ investors but not necessarily the information environment faced by the decision agents involved in negotiating an acquisition bid. Not surprisingly, we do not find any significant differential results on acquirer RPR in subsamples formed on these other uncertainty measures. Unreported for brevity, these results are available upon request. Page 33
ACCEPTED MANUSCRIPT
acquirer RPR coefficient is positive but not significant in the Low sigma subsample and is negative and marginally significant in the Cash subsample.18 Notably, target RPR is negatively related to bid premia in the whole sample as well as the subsamples of cash deals and in transactions with low acquirer sigma. This is consistent with the
CR IP T
perceived valuation hypothesis in that targets with pre-announcement prices close to their 52week highs may be perceived as having relatively high valuation levels. Accordingly, they receive a lower premium. This result is also consistent with the argument put forward by Baker et al. (2012). That is, if the decision makers in merger transactions implicitly use targets’ past
AN US
peak prices as reference points, bid premia will be negatively related to target RPRs. When there is more uncertainty about the value of acquirer shares (High sigma) and when the acquirer’s valuation level is directly relevant to bid premium negotiations (Non-cash), however, the
M
coefficients on target RPR are not statistically significant.
Bid premia are higher when acquirer M/B is higher in the whole CRSP sample. Acquirer
ED
M/B is not significant in the Cash subsample, indicating that conventional acquirer valuation levels play no role in explaining bid premia for cash acquisitions, as expected. In non-cash deals,
PT
however, bid premia are positively related to acquirer M/B, consistent with Dong et al. (2006).
CE
Further, acquirer M/B has at least a marginally significant influence on bid premia in both the Low sigma and High sigma samples. Target M/B has a negative coefficient in the whole CRSP
AC
sample and each subsample, but it is not statistically significant in the Low sigma sample. These results add new evidence beyond Baker et al. (2012). A potential interpretation is
that, in negotiating “fair” bid premia, target managers focus on their own firms’ 52-week highs when the value of the payment is known or has relatively low uncertainty associated with it. In 18
One possible interpretation of the negative coefficient on acquirer RPR in the cash subsample is that, to avoid being paid with overvalued stock, target managers are willing to accept a lower premium in exchange for cash payment when acquirers’ perceived valuation levels are high. Page 34
ACCEPTED MANUSCRIPT
acquisitions involving non-cash payment or when there is high uncertainty regarding the value of the acquirer’s shares, on the other hand, determining the fairness of a bid depends more critically on the valuation of the acquirer’s stock rather than how close the offer price is to the target’s past peak price and, therefore, target managers focus on acquirer RPRs (in addition to conventional
CR IP T
valuation levels) in these bid premium negotiations. With more uncertainty about the value of acquirer shares, target managers may demand higher bid premia from acquirers with higher RPRs or be willing to accept lower bid premia from acquirers with lower RPRs. [Insert Table 6 here]
AN US
Table 7 presents five OLS regressions of target CAR on acquirer RPR, acquirer M/B, target RPR, target M/B, and the other control variables used in Table 3, after excluding the potential endogenous variables, stock, cash, tender offer, and hostility. Similar to the analysis of
M
bid premia, the first regression is based on all transactions with available data on targets from CRSP. The second and third columns show results for acquisitions sorted on acquirers’ sigma
ED
and the last two columns present results for the Cash and Non-cash subsamples.19 Acquirer RPR has positive and significant coefficients in the whole sample, in deals of higher uncertainty about
PT
acquirer valuation, and the non-cash deals. To the extent that acquirer RPR positively affects bid
CE
premia, as shown in Table 6, this positive impact is reflected in target CAR. That is, when the valuation level of the acquirer is directly relevant to the bid premium negotiation process (Non-
AC
cash) or the valuation level of the acquirer is more uncertain (High sigma), the acquirer’s RPR positively influences the bid premium and this carries over to a positive influence on the target’s CAR.
19
For the sake of brevity, we include only the results of subsamples sorted according to the acquirers’ sigma. In unreported results, we find a positive and significant coefficient on the acquirers’ RPR in all subsamples sorted on other information environment proxies, similar to the results in the first model of Table 7. Page 35
ACCEPTED MANUSCRIPT
Target RPR and target M/B have negative and significant coefficients in all five models. The results concerning the influence of target M/B on target CAR are consistent with Dong et al. (2006). The magnitudes of the target RPR coefficients appear larger in the Low sigma subsample than in the High sigma subsample, and larger in the Cash subsample than in the Non-cash
CR IP T
subsample. As discussed in Dong et al. (2006), target valuation can affect target CAR via two possible mechanisms. First, as shown in Table 6, targets with higher RPR (or M/B) receive lower premia, particularly among Cash and Low sigma subsamples. Thus, targets with higher RPR (or M/B) earn lower announcement abnormal returns, and this relation is stronger among the Cash
AN US
and Low sigma subsamples. Second, the acquisition announcement triggers a reassessment of valuation levels for acquirer and target shares, and targets with relatively high valuation levels tend to experience a negative reassessment. The coefficients on target RPR in Table 7 are
M
consistent with both mechanisms.
[Insert Table 7 here]
ED
5.2.Method of payment and deal combativeness
Table 8 presents Probit regression models on the method of payment (Stock and Cash) in
PT
Panel A and deal combativeness (Tender offers, Hostile, and Success) in Panel B for the
CE
subsamples of deals involving private targets, public targets, and targets covered in CRSP, respectively.20 We exclude method of payment and deal combativeness variables from the list of
AC
independent variables to avoid potential endogeneity. Our results in Panel A on the acquirer and target M/B ratios are in general consistent with Dong et al. (2006). The first three columns show that the acquirer’s M/B is positively related to the use of stock as the method of payment, consistent with Dong et al. (2006) and valuation theory. Further, in the CRSP subsample, the 20
The samples in Dong et al. (2006) and Baker et al. (2012) are limited to acquisitions of target firms with sufficient data available in CRSP. Therefore, direct comparisons to their results are limited to our CRSP sample results. In addition, (unreported) results from a Logit analysis lead to the same conclusions. Page 36
ACCEPTED MANUSCRIPT
target’s M/B also carries a positive and significant coefficient. The coefficient on the acquirer’s RPR is not statistically significant except within the CRSP subsample that also includes the target’s RPR. In this model, the coefficient on the acquirer’s RPR is positive and marginally significant while the coefficient on the target’s RPR is negative and significant. Thus, there is
CR IP T
some evidence that the acquirer’s RPR positively influences the decision to use stock as the method of payment, consistent with the perceived valuation hypothesis, but the negative coefficient associated with target RPR seems inconsistent with RPR acting as an indicator of valuation levels.
AN US
The last three columns of Panel A in Table 8 present Probit regressions on the use of cash. The pattern of the coefficients on acquirer and target M/B is generally consistent (in the opposite direction) with the coefficients in the first three columns. The coefficient on acquirer
M
RPR is positive in the private subsample and negative in the public and CRSP subsamples but none of the coefficients is statistically significant. The coefficient on the target RPR is positive
ED
and significant. Overall, the results in Panel A offer mixed evidence concerning whether the RPRs for the bidder and target influence the method of payment.
PT
Panel B of Table 8 reports Probit regression results on indicators of deal combativeness,
CE
specifically whether the mode of acquisition is a tender offer, whether the bid is met with hostility, and whether the deal is ultimately successful. Consistent with Dong et al. (2006),
AC
targets with higher M/B are less likely to be hostile and more likely to complete the deal. To the extent that tender offers are mostly paid in cash, bidders with low perceived
valuation should be more likely to make tender offers than bidders with high perceived valuation. However, we find that acquirer RPR is insignificant in the Tender offer regressions. In addition, the perceived valuation hypothesis does not have a specific prediction on how acquirer
Page 37
ACCEPTED MANUSCRIPT
misvaluation affects deal hostility. On the one hand, an overvalued acquirer is expected to offer a high premium, reducing the likelihood of deal combativeness. On the other hand, perceived acquirer overvaluation increases the probability of target hostility. We observe a significant negative effect of acquirer RPR but a positive effect of acquirer M/B on deal hostility in the
significance in the Hostile regression for the private subsample.
CR IP T
CRSP subsample, yet acquirer RPR is insignificant and acquirer M/B is negative with marginal
The coefficients on acquirer RPR in the deal success regressions are positive and significant in the private and public subsamples, and positive but not significant in the CRSP
AN US
subsample. Again, this provides weak evidence that the 52-week high influences perceived valuation levels. Interestingly, the regression analysis in Dong et al. (2006) does not find a statistically significant role for acquirer valuation levels in explaining deal success, but in the
M
public and CRSP subsamples we find a negative and statistically significant impact of acquirer M/B on deal success when acquirer RPR is also included in the regression. In the CRSP
ED
subsample, target M/B is positively related to deal success, consistent with Dong et al. (2006);
PT
the coefficient of target RPR is insignificant. [Insert Table 8 here]
CE
In summary, although acquirer and target RPRs influence acquirer and target CARs and bid premia in a manner consistent with the perceived valuation hypothesis, reference prices have
AC
little consistent impact on the method of payment and deal combativeness. We suggest that because stock market reactions depend on investor perceptions of acquirer and target valuations, reference prices affect acquirer and target CARs. Reference prices also affect bid premia because bid premia are the result of negotiations between target and acquirer managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and
Page 38
ACCEPTED MANUSCRIPT
deal combativeness, on the other hand, are primarily managerial decisions by either the target or the acquirer, and if managers are more informed of their own firms' true valuations, these decisions will be less affected by reference prices. 6. Robustness checks
CR IP T
We examine the robustness of the relation between acquirer RPR and acquirer CAR by considering alternative measures of CAR, alternative measures of RPR, and in subsamples of time periods. Following the analysis of Table 3, we estimate regressions using these alternatives for the overall sample with all targets and the subsamples involving only private and public
AN US
targets.21 The same general pattern emerges throughout this section. That is, the reference price effect is strong in the overall sample and the private subsample but weaker in the public subsample, a result implied by the perceived valuation hypothesis. The full regression
M
specifications follow those of Table 3. For brevity, we only show the coefficients and t-statistics of the acquirer RPR.
ED
6.1.Alternative methods of estimating CAR
In our main analysis, we use the market model approach to estimate abnormal returns
PT
over the event window of [-5, +1]. To check robustness, we estimate three models following,
CE
respectively, Moeller et al. (2004) for CAR 1, a simple market-adjusted abnormal return for CAR 2, and Lang et al. (1991) for CAR 3.22
AC
We use each of these alternative CARs to estimate the baseline regressions, with results reported in Panel A of Table 9. In all regressions, the RPR coefficients are negative and
21
Further analysis indicates that results based on the subsample of transactions involving targets covered in CRSP are robust to the variations in the measurement of acquirer CAR, acquirer RPR, and in time subsamples, though with weaker statistical significance. Unreported for brevity, these results are available upon request. 22 Using additional alternative models (Fuller et al., 2002; Faccio et al., 2006; Officer et al., 2009), as adopted in recent studies, yields similar results. Page 39
ACCEPTED MANUSCRIPT
significant at the 1% level in the overall sample and the subsample involving private targets. As expected, the reference price effect is weaker in acquisitions of public targets. It is possible that the reference price effect exists even without announcements of mergers and acquisitions. That is, there could be a general pattern between RPR and stock
CR IP T
returns. To test this possibility, we re-estimate the acquirers’ CAR using a benchmark formed on reference price ratios. Specifically, for each acquirer in an acquisition deal, we identify a set of control firms whose reference price ratios at day t-6 are similar (i.e., same percentile) to the acquirer and who do not make acquisitions. We then use their average returns over the event
AN US
window [-5, +1] of the acquirer as the benchmark. Thus, the abnormal returns to the acquirer constructed using this approach have already filtered out the “general” return pattern associated with reference price ratios. Using this RPR-controlled cumulative abnormal return as the
M
dependent variable in our baseline regressions yields similarly strong results. Thus, the reference price effect we find is not a general return pattern. Rather, it is associated with the
ED
announcements of mergers and acquisitions.
[Insert Table 9 here]
PT
6.2.Alternative measures of RPR
CE
In the main analysis, we measure the acquirer RPR as of the sixth day prior to the acquisition announcement. To test whether our results are sensitive to this choice, we estimate
AC
three alternative measures of RPR: RPR on the 11th day prior to the announcement, RPR at the month-end prior to the announcement month, and RPR quintiles (0 to 4). We then estimate the baseline regression models by replacing the RPR with each of the three alternative RPR measures. Results are presented in Panel B of Table 9. All RPR coefficients are negative and
Page 40
ACCEPTED MANUSCRIPT
significant for regressions based on the overall sample and the subsample involving private targets. Throughout the paper, we only consider the 52-week high. To the extent that the 52-week low is also salient, it might play a role. To test this possibility, we first add a variable P/L, which
CR IP T
is the ratio of stock price at day t-6 to L, the 52-week low, in the baseline regressions. We find that the coefficients of P/L are not significant and the RPR coefficients are -5.495 (t=-8.53), 6.294 (t=-8.61), and -2.651 (t=-2.01) in the regressions based on the whole sample, the subsample of private targets, and the subsample of public targets, respectively. Further, we
AN US
construct an alternative reference price ratio (P-L)/(H-L), where P is the stock price on day t-6, and H and L are the 52-week high and low prices, respectively (see Lee and Yerramilli, 2016). We find that this new measure has a correlation coefficient of 0.843 with our RPR. When we use
M
this alternative variable in place of the RPR in the baseline regressions, the coefficients on this variable are negative and significant. When both RPR and (P-L)/(H-L) are included in the
ED
regressions, the coefficients on RPR are -8.468 (t=-7.89), -9.949 (t=-8.04), and -3.115 (t=-1.51) in the regressions based on the whole sample, the subsample of private targets, and the
PT
subsample of public targets, respectively. By contrast, the coefficients on (P-L)/(H-L) become
CE
positive.
As another possibility, investors might react differently to announcements of mergers and
AC
acquisitions because the overall market is near or far below its 52-week high. To examine this possibility, we construct the reference price ratio for the Standard & Poor’s (S&P) 500 index, or RPR of SP. We find that firm-level RPR is positively correlated with RPR of SP, with a correlation coefficient of 0.287. When we replace the firm-level RPR with the RPR of SP in our baseline regressions, however, the coefficients are largely positive with weak statistical
Page 41
ACCEPTED MANUSCRIPT
significance. When both firm- and market-level RPRs are in the regressions, the coefficients on the firm-level RPR are -5.802 (t=-8.85), -6.495 (t=-8.80), and -3.236 (t=-2.39), and those on the market-level RPR are 4.943 (t=3.00), 3.490 (t=1.83), and 8.169 (t=2.66) in the regressions based on the whole sample, the subsample of private targets, and the subsample of public targets,
CR IP T
respectively. Likewise, considering the 52-week low of the S&P 500 level in the market-level RPR yields virtually identical results. Thus, the reference price effect we document in this paper is not due to the contemporaneous overall market index relative to its 52-week high or low. 6.3.The reference price effect over time
AN US
We examine whether the reference price effect holds over time. If, for example, investors learn about this bias and correct it over time, we expect the reference price effect to decay over time. To test this possibility, we form three subsamples by the year of announcements: 1981–
M
1990 (1980s), 1991–2000 (1990s), and 2001–2014 (2000s) and estimate the baseline regression models for each of the subsamples. The regression results are listed in Panel C of Table 9. The
ED
general pattern of a stronger reference price effect among acquisitions of private targets than acquisitions of public targets remains. This pattern also holds for each of the three subperiods.
PT
The effect appears stronger over the 1990s in the subsample of private targets.
CE
Further (unreported) analyses are conducted to determine whether investor inattention interacts with the reference price effect. Louis and Sun (2010) document that acquisitions
AC
announced on Fridays receive less investor attention. In subsamples of deals on Fridays and deals on other days we find that the reference price effect exists in both subsamples with similar coefficients as in the baseline regressions.
Page 42
ACCEPTED MANUSCRIPT
7. Conclusion We examine whether acquirer 52-week high prices influence perceived valuation levels, market responses to acquisition announcements, and deal characteristics. In a comprehensive sample of mergers and acquisitions, we document a significant acquirer reference price effect:
CR IP T
Acquirers earn higher (lower) announcement-period returns when their pre-announcement stock prices are well below (near) their 52-week highs. The acquirer reference price effect is stronger in acquisitions of private targets, deals involving greater uncertainty (more volatile acquirers, acquirers followed by fewer analysts, greater relative size, and non-cash payment), and acquirers
AN US
with greater individual investor ownership. Further, the announcement-period abnormal return pattern is reversed in the subsequent year, suggesting that the influence of 52-week high prices on perceived valuation levels is the result of an irrational behavioral bias.
M
We further report that acquirer reference prices affect bid premia and target announcement returns when the uncertainty of acquirer valuation is high and when non-cash
ED
payment is made for the acquisition. By contrast, we find little consistent evidence regarding whether reference prices (of either the acquirer or the target) affect method of payment or deal
PT
combativeness. A possible interpretation is that stock market reactions depend on investor
CE
perceptions of acquirer and target valuations and that investors have less information than managers. Therefore, investors are more likely to rely on heuristics when assessing the valuation
AC
implications of acquisitions. Reference prices also affect bid premia because bid premia are the result of negotiations between target and acquirer managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and deal combativeness, on the other hand, are primarily decisions by the acquirer and target managers, and if acquirer
Page 43
ACCEPTED MANUSCRIPT
and target managers are more informed of their own firms’ true valuations, these decisions are less likely affected by reference prices. While we focus on M&A events, the anchoring bias and its influence on perceived valuations might also be at work in other corporate events. Such a general pattern would suggest
CR IP T
a theory on how past peak prices affect investors’ reactions to news. The extent to which this conjecture squares with data is left for future research. Regardless of the general pattern, however, our paper provides strong evidence of anchoring bias influencing perceived valuation
AC
CE
PT
ED
M
AN US
levels and investors’ reactions to M&A events.
Page 44
ACCEPTED MANUSCRIPT
Appendix A. Variable definitions
AC
CE
PT
ED
M
AN US
CR IP T
BHAR is calculated using the reference portfolio approach. Each month, we sort all NYSE common stocks into size quintiles based on their month end market capitalization. These quintiles are further sorted into quintiles using book-to-market ratios, which are further sorted into quintiles based on returns over the past 12 months. We then place Amex and Nasdaq firms into one of these 125 portfolios based on their month-end size, book-tomarket ratios, and past 12-month returns. After excluding firms that made significant acquisitions in the past two-year period, we calculate the monthly return for each of the 125 reference portfolios by averaging the monthly returns across all stocks in each portfolio. These portfolio returns are then used as benchmarks to calculate buy-and-hold abnormal returns for our sample firms. The BHAR is measured over the one-year period following the month of announcement. Bid premium is the premium the target firm receives relative to the price four weeks prior to the acquisition announcement, as reported in the SDC. CAR is the cumulative abnormal returns to the acquirer, measured over the event window [–5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. Cash is equal to one if the total consideration is paid in cash. Cross border is equal to one if the acquirer and target come from different countries. Dormant > 1 yr is a dummy variable equal to one if the current transaction is at least one year apart from the previous transaction made by peers in the same CRSP four-digit Standard Industrial Classification code (SIC) industry. The definition follows Cai et al. (2011). Few analysts is binary equal to one if the number of analysts following the acquirer (over the past six months prior to the acquisition announcement month) is below the sample median (by year). Hostile is equal to one if the SDC classifies the deal as hostile or unsolicited. Indiv. own. is one minus the institutional ownership measure, which is the total number of shares owned by institutional investors divided by the number of shares outstanding at the end of the most recent reporting quarter. A missing value for institutional ownership is assigned as zero and the Indiv. own. is assigned as one. Leverage is total assets minus book value of equity, normalized by total assets, all measured at the year-end before announcement. Ln(M/B) is the natural logarithm of the acquirer’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Non-cash is equal to one if at least part of the payment is not in cash. Past return is the raw return over the 12-month period prior to the announcement month. Private is equal to one if the SDC classifies the target public status as subsidiary or private firm. Rel. size is the ratio of the transaction value excluding assumed liability, divided by market capitalization of the acquirer, measured at the beginning of the calendar year in which the announcement occurs. RKRV is defined in Appendix B. RPR is the acquirer’s stock price as of t – 6, the sixth day prior to the announcement, divided by the highest closing price over the past 252 days relative to t – 6. Same industry (or Same ind.) is a dummy variable equal to one if the acquirer and acquired assets share the same two-digit Standard Industrial Classification code (SIC). Page 45
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Sigma is the weekly return standard deviation of the acquirer over the past 52 weeks prior to the acquisition announcement month, as in Zhang (2006). Size is the natural logarithm of market capitalization as of the calendar year-end prior to the announcement in constant $million as of year 2014. Stock is equal to one if the total consideration is paid in stock. Success is equal to one if the transaction is eventually completed, as recorded in the SDC. Target CAR is the cumulative abnormal returns to the target, measured over the event window [– 5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. Target Ln(M/B) is the natural logarithm of the target firm’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Target RPR is the RPR ratio of the target firm (if covered in CRSP) as of the sixth day prior to the announcement date. RPR is the ratio of stock price to the 52-week high price level. Tender offer is equal to one if the deal is classified as a tender offer. Toehold is equal to one if the acquirer owns 5% or more of the target before the announcement.
Page 46
ACCEPTED MANUSCRIPT
Appendix B. RKRV definition We follow Rhodes-Kropf, Robinson, and Viswanathan (2005) and Hertzel and Li (2010) to construct the valuation measure RKRV. All firms covered in the CRSP/Compustat merged database from fiscal years 1977 to 2014 are used in the estimation.
M
AN US
CR IP T
Step 1: For each fiscal year t (from 1977 to 2014) and industry j (of the total 12 Fama and French industries), the following cross-sectional regression is estimated, ( ) , where is the natural logarithm of market value of equity, or log(shrout*prc) from CRSP, measured three months following the fiscal year-end month; is a vector of firm-specific accounting variables at fiscal year t, [ ]; is a vector of conditional accounting multiples [ ]; is the natural logarithm of book equity (Compustat annual item CEQ); is the natural logarithm of net income if it is positive (Compustat annual item NI); is the natural logarithm of the absolute value of net income if it is negative; and is the leverage, equal to 1 – MVE/MVA, where MVE is CRSP shrout*prc, and MVA is MVE plus total assets (Compustat AT) minus deferred taxes (Compustat TXDB) minus book equity (Compustat CEQ). The regressions generate a time series of estimated accounting multiples ̂ .
ED
Step 2: For each fiscal year t (from 1978 to 2014) and industry j, the trailing 20-year moving ∑ averages of the accounting multiples are calculated as ̅ ̂ , where is the total number of accounting multiples available in the trailing 20-year period. ( ̂ ); ( ̂ ) ( ̅ )
CE
PT
Step 3: The firm-specific error is The time series sector error is The long-run value to book is Thus, the RKRV valuation measure is
(
̅ ); and . (
̅ )
AC
Step 4: Each M&A transaction is matched with the acquirer’s RKRV variable such that the announcement month is at least four months, but at most 15 months (inclusive) later than the fiscal year-end month.
Page 47
ACCEPTED MANUSCRIPT
References Akbulut, M., 2013. Do overvaluation-driven stock acquisitions really benefit acquirer shareholders? Journal of Financial and Quantitative Analysis 48 (4), 1025–1055. Amihud, Y., Lev, B., 1981. Risk reduction as a managerial motive for conglomerate mergers. Bell Journal of Economics 12 (2), 605–617.
CR IP T
Andrade, G., Mitchell, M., Stafford, E., 2001. New evidence and perspectives on mergers. Journal of Economic Perspectives 15 (2), 103–120. Baker, M., 2009. Capital market-driven corporate finance. Annual Review of Financial Economics 1, 181–205. Baker, M., Mendel, B., Wurgler, J., 2016. Dividends as reference points: A behavioral signaling approach. Review of Financial Studies 29 (3), 697–738.
AN US
Baker, M., Pan, X., Wurgler, J., 2012. The effect of reference point prices on mergers and acquisitions. Journal of Financial Economics 106 (1), 49–71. Baker, M., Wurgler, J., 2012. Behavioral corporate finance: An updated survey. In George Constantinides, Milton Harris, and Rene Stulz (Ed.), The Handbook of the Economics of Finance: Empirical Corporate Finance, vol. 2. Elsevier, North Holland., pp. 357–424.
M
Baker, M., Xuan, Y., 2016. Under new management: Equity issues and the attribution of past returns. Journal of Financial Economics 121 (1), 66–78.
ED
Barber, B., Odean, T., 2008. All that glitters: The effect of attention and news on the buying behavior of individual and institutional investors. Review of Financial Studies 21 (2), 785–818.
PT
Barberis, N., Shleifer, A., Vishny, R., 1998. A model of investor sentiment. Journal of Financial Economics 49 (3), 307–343. Bartov, E., Radhakrishnan, S., Krinsky, I., 2000. Investor sophistication and patterns in stock returns after earnings announcements. Accounting Review 75 (1), 43–63.
CE
Bereskin, F., Byun, S., Officer, M.S., Oh, J., 2017. The effect of cultural similarity on mergers and acquisitions. Journal of Financial and Quantitative Analysis (forthcoming).
AC
Betton, S., Eckbo, B.E., Thorburn, K., 2008. Corporate takeovers. In: Eckbo, B.E. (Ed.), The Handbook of Corporate Finance: Empirical Corporate Finance, vol. 2., Elsevier, North Holland, pp. 291–421. Bouwman, C., Fuller, K., Nain, A., 2009. Market valuation and acquisition quality: Empirical evidence. Review of Financial Studies 22 (2), 633–679. Cai, J., Song, M.H., Walkling, R.A., 2011. Anticipation, acquisitions and bidder returns: industry shocks and the transfer of information across rivals. Review of Financial Studies 24 (7), 2242–2285. Chang, S., 1998. Takeovers of privately held targets, methods of payment and bidder returns. Journal of Finance 53 (2), 773–784. Page 48
ACCEPTED MANUSCRIPT
Chapman, G., Johnson, E., 1994. The limits of anchoring. Journal of Behavioral Decision Making 7 (4), 223–242. Chopra, N., Lakonishok, J., Ritter, J., 1992. Measuring abnormal returns: Do stocks overreact? Journal of Financial Economics 31 (2), 235–268. Danbolt, J., Siganos, A., Vagenas-Nanos, E., 2015. Investor sentiment and bidder announcement abnormal returns. Journal of Corporate Finance 33, 164–179.
CR IP T
DeGeorge, F., Patel, J., Zeckhauser, R.,1999. Earnings management to exceed thresholds. The Journal of Business 72 (1), 1–33. Dong, M., Hirshleifer, D., Richardson, S., Teoh, S.H., 2006. Does investor (mis)valuation drive the takeover market? Journal of Finance 61 (2), 725–762.
AN US
Driessen, J., Lin, T., Van Hemert, O., 2013. How the 52-week high and low affect option-implied volatilities and stock return moments. Review of Finance 17, 369–401. Duchin, R., Schmidt, B., 2013. Riding the merger wave: Uncertainty, reduced monitoring, and bad acquisitions. Journal of Financial Economics 107 (1), 69–88. Edwards, E.O., Bell, P.W., 1961. The Theory and Measurement of Business Income. University of California Press, Berkeley, CA.
M
El-Khatib, R., Fogel, K., Jandik, T., 2015. CEO network centrality and merger performance. Journal of Financial Economics 116 (2), 349–382. Engelberg, J., McLean, D., Pontiff, J., 2017. Anomalies and news. Journal of Finance (forthcoming).
ED
Faccio, M., McConnell, J., Stolin, D., 2006. Returns to acquirers of listed and unlisted targets. Journal of Financial and Quantitative Analysis 41 (1), 197–220.
PT
Fama, E.F., 1998. Market efficiency, long-term returns, and behavioral finance. Journal of Financial Economics 49 (3), 283–306.
CE
Fama, E.F., French, K.R.,1997. Industry costs of equity. Journal of Financial Economics 43 (2), 153–194. Fama, E.F., French, K.R., 2015. A five-factor asset pricing model. Journal of Financial Economics 116 (1), 1–22.
AC
Fu, F., Lin, L., Officer, M.S., 2013. Acquisitions driven by stock overvaluation: Are they good deals? Journal of Financial Economics 109 (1), 24–39. Fuller, K., Netter, J., Stegemoller, M., 2002. What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. Journal of Finance 57 (4), 1763–1793. Genesove, D., Mayer, C., 2001. Nominal loss aversion and seller behavior: Evidence from the housing market. Quarterly Journal of Economics 116 (4), 1233–1260. George, T.J., Hwang, C., 2004. The 52-week high and momentum investing. Journal of Finance 59 (5), 2145–2176. Page 49
ACCEPTED MANUSCRIPT
George, T.J., Hwang, C., Li, Y., 2014. Anchoring, the 52-week high and post earnings announcement drift. Unpublished working paper. University of Houston, Nanyang Technological University, and University of Cambridge. George, T.J., Hwang, C., Li, Y., 2018. The 52-week high, q-theory, and the cross section of stock returns. Journal of Financial Economics 128 (1), 148–163.
CR IP T
Golubov, A., Petmezas, D., Travlos, N.G., 2012. When it pays to pay your investment banker: new evidence on the role of financial advisors in M&As. Journal of Finance 67 (1), 271–312. Golubov, A., Yawson, A., Zhang, H., 2015. Extraordinary acquirers. Journal of Financial Economics 116 (2), 314–330. Gompers, P.A., Metrick, A., 2001. Institutional investors and equity prices. Quarterly Journal of Economics 116 (1), 229–259.
AN US
Hansen, R., 1987. A theory for the choice of exchange medium in mergers and acquisitions. Journal of Business 60 (1), 75–96. Harford, J., 1999. Corporate cash reserves and acquisitions. Journal of Finance 54 (6), 1969–1997. Harford, J., Humphery-Jenner, M., Powell, R., 2012. The sources of value destruction in acquisitions by entrenched managers. Journal of Financial Economics 106 (2), 247–261.
M
Hastie, R., Schkade, D., Payne, J., 1999. Juror judgment in civil cases: Effects of plaintiff’s requests and plaintiff’s identity on punitive damage awards. Law and Human Behavior 23 (4), 445–470.
ED
Heath, C., Huddart, S., Lang, M., 1999. Psychological factors and stock option exercise. Quarterly Journal of Economics 114 (2), 601–627.
PT
Hertzel, M.G., Li, Z., 2010. Behavioral and rational explanations of stock price performance around SEOs: Evidence from a decomposition of market-to-book ratios. Journal of Financial and Quantitative Analysis 45 (4), 935–958.
CE
Huddart, S., Lang, M., Yetman, M., 2009. Volume and price patterns around a stock’s 52-week highs and lows: theory and evidence. Management Science 55 (1), 16–31.
AC
Jacowitz, K.E., Kahneman, D., 1995. Measures of anchoring in estimation tasks. Personality and Social Psychology Bulletin 21 (11), 1161–1166. Jarrell, G.A., Brickley, J.A., Netter, J.M., 1988. The market for corporate control: The empirical evidence since 1980. Journal of Economic Perspectives 2 (1), 49–68. Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance 48 (1), 65–91. Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76 (2), 323–329.
Page 50
ACCEPTED MANUSCRIPT
Jensen, M., Ruback, R., 1983. The market for corporate control: The scientific evidence. Journal of Financial Economics 11 (1-4), 5–50. Jin, L., Scherbina, A., 2011. Inheriting losers. Review of Financial Studies 24 (3), 786–820. Kahneman, D., 2011. Thinking, Fast and Slow. Farrar, Straus and Giroux, New York.
CR IP T
La Porta, R., Lakonishok, J., Shleifer, A., Vishny, R., 1997. Good news for value stocks: further evidence on market efficiency. Journal of Finance 52 (2), 859–874. Lang, L., Stulz, R., Walkling, R.A., 1989. Managerial performance, Tobin’s q, and the gains from successful tender offers. Journal of Financial Economics 24 (1), 137–154. Lang, L., Stulz, R., Walkling, R.A., 1991. A test of the free cash flow hypothesis: The case of bidder returns. Journal of Financial Economics 29 (2), 315–335.
AN US
Lee, S., Yerramilli, V., 2016. Relative values, announcement timing, and shareholder returns in mergers and acquisitions. Unpublished working paper. University of Houston. Li, J., Yu, J., 2012. Investor attention, psychological anchors, and stock return predictability. Journal of Financial Economics 104 (2), 401–419. Loughran, T., Ritter, J., 2002. Why don't issuers get upset about leaving money on the table in IPOs? Review of Financial Studies 15 (2), 413–444.
M
Louis, H., Sun, A., 2010. Investor inattention and the market reaction to merger announcements. Management Science 56 (10), 1781–1793.
ED
Lyon, J.D., Barber, B.M., Tsai, C., 1999. Improved methods for tests of long-run abnormal stock returns. Journal of Finance 54 (1), 165–201.
PT
Ma, Q., Whidbee, D.A., Zhang, W., 2011. Value, valuation, and the long-run performance of merged firms. Journal of Corporate Finance 17 (1), 1–17.
CE
Malmendier, U., Tate, G., 2008. Who makes acquisitions? CEO overconfidence and the market's reaction. Journal of Financial Economics 89 (1), 20–43. Masulis, R.W., Wang, C., Xie, F., 2007. Corporate governance and acquirer returns. Journal of Finance 62 (4), 1851–1889.
AC
Matsusaka, J. G., 1993. Takeover motives during the conglomerate merger wave. Rand Journal of Economics 24 (3), 357–379. Mitchell, M., Stafford, E., 2000. Managerial decisions and long-term stock price performance. Journal of Business 73 (3), 287–329. Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2004. Firm size and the gains from acquisitions. Journal of Financial Economics 73 (2), 201–228. Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2005. Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. Journal of Finance 60 (2), 757–782. Page 51
ACCEPTED MANUSCRIPT
Morck, R., Shleifer, A., Vishny, R.W., 1990. Do managerial objectives drive bad acquisitions? Journal of Finance 45 (1), 31–48. Mussweiler, T., Strack, F., 2000. Numeric judgments under uncertainty: The role of knowledge in anchoring. Journal of Experimental Social Psychology 36 (5), 495–518.
CR IP T
Nofsinger, J. R., Sias, R.W., 1999. Herding and feedback trading by institutional and individual investors. Journal of Finance 54 (6), 2263–2295. Northcraft, G.B., Neale, M.A., 1987. Experts, amateurs, and real estate: An anchoring-and-adjustment perspective on property pricing decisions. Organizational Behavior and Human Decision Processes 39(1), 84–97. Odean, T., 1998. Are investors reluctant to realize their losses? Journal of Finance 53 (5), 1775–1798.
AN US
Officer, M., 2007. The price of corporate liquidity: Acquisition discount for unlisted targets. Journal of Financial Economics 83 (3), 571–598. Officer, M., Poulsen, A., Stegemoller, M., 2009. Target–firm information asymmetry and acquirer returns. Review of Finance 13 (3), 467–493. Ohlson, J.A., 1995. Earnings, book values, and dividends in equity valuation. Contemporary Accounting Research 11 (2), 661–687.
M
Oppenheimer, D., LeBoeuf, R., Brewer, N., 2008. Anchors aweigh: A demonstration of cross-modality anchoring and magnitude priming. Cognition 106 (1), 13–26.
ED
Rau, P. R., Vermaelen, T., 1998. Glamour, value and the post-acquisition performance of acquiring firms. Journal of Financial Economics 49 (2), 223–253.
PT
Rhodes-Kropf, M., Viswanathan, S., 2004. Market valuation and merger waves. Journal of Finance 59 (6), 2685–2718.
CE
Rhodes-Kropf, M., Robinson, D.T., Viswanathan, S., 2005. Valuation waves and merger activity: The empirical evidence. Journal of Financial Economics 77 (3), 561–603. Roll, R., 1986. The hubris hypothesis of corporate takeovers. Journal of Business 59 (2), 197–216.
AC
Shleifer, A., 2000. Inefficient Markets: An Introduction to Behavioral Finance. Oxford: Oxford University Press, New York. Shleifer, A., Vishny, R.W., 2003. Stock market driven acquisitions. Journal of Financial Economics 70 (3), 295–311. Shue, K., Townsend, R., 2017. Growth through rigidity: An explanation for the rise in CEO pay. Journal of Financial Economics 123 (1), 1–21. Sias, R., Whidbee, D.A., 2010. Insider trades and demand by institutional and individual investors. Review of Financial Studies 23 (4), 1544–1595.
Page 52
ACCEPTED MANUSCRIPT
Slovin, M.B., Sushka, M.E., Polonchek, J.A., 2005. Methods of payment in asset sales: Contracting with equity versus cash. Journal of Finance 60 (5), 2385–2407. Strack, F., Mussweiler, T., 1997. Explaining the enigmatic anchoring effect: Mechanisms of selective accessibility. Journal of Personality and Social Psychology 73 (3), 437–446. Titman, S., Wei, K.C.J., Xie, F., 2004. Capital investments and stock returns. Journal of Financial and Quantitative Analysis 39 (4), 677–700.
CR IP T
Tversky, A., Kahneman, D., 1974. Judgement under uncertainty: heuristics and biases. Science 185 (4157), 1124–1131.
AC
CE
PT
ED
M
AN US
Zhang, X.F., 2006. Information uncertainty and stock returns. Journal of Finance 61(1), 105–136.
Page 53
ACCEPTED MANUSCRIPT
PT
Stock 0.19
Cash 0.23
0.01 0.00 0.00 0.02 0.19 0.23 0.25 0.15 0.24 0.20 0.28 0.32 0.28 0.31 0.34 0.31 0.27 0.27 0.28 0.29 0.22 0.14 0.14 0.10 0.10 0.07 0.08 0.08
0.02 0.00 0.01 0.04 0.36 0.24 0.22 0.21 0.25 0.22 0.19 0.15 0.19 0.19 0.19 0.14 0.17 0.18 0.19 0.21 0.22 0.25 0.27 0.28 0.32 0.34 0.35 0.33
Tender Hostile Success RPR 0.04 0.02 0.88 0.81
AN US
0.68 0.73 0.83 0.72 0.55 0.63 0.59 0.66 0.70 0.79 0.74 0.76 0.80 0.75 0.73 0.74 0.74 0.76 0.72 0.70 0.69 0.78 0.75 0.78 0.79 0.78 0.75 0.75
ED
312.46 140.71 89.95 194.20 277.84 218.29 181.04 180.99 188.92 166.49 112.06 112.00 185.00 152.21 232.11 325.64 293.67 595.03 559.41 1027.34 658.65 275.51 256.64 366.00 599.38 590.15 609.35 688.86
CE
221 236 341 370 278 339 276 335 394 322 387 452 603 766 910 985 1,242 1,348 972 759 623 570 590 694 657 681 591 416
AC
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Private 0.75
M
Panel A: Sample description Transaction Year Nobs value ($m) All 19,119 453.31
CR IP T
Table 1 Sample description and summary statistics. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. Panel A lists the overall and year-by-year averages of sample characteristics. Transaction value ($million) is as reported in SDC; Private is the percentage of transactions involving Private targets (private firms or subsidiaries); Stock (Cash) is the percentage of deals paid for purely with stock (cash); Tender is the percentage of deals that are tender offers; Hostile is the percentage of deals involving target resistance; Success is the percentage of deals that are eventually completed; RPR is the average RPR ratio of the acquirers. RPR ratio is the acquiring firm’s closing stock price on the sixth day prior to the announcement date divided by the stock’s 52-week high. Panel B presents the main variables’ summary statistics. The variables listed in Panel B are defined in Appendices A and B.
Page 54
0.13 0.12 0.07 0.08 0.14 0.12 0.08 0.13 0.08 0.04 0.03 0.02 0.03 0.03 0.04 0.03 0.04 0.03 0.05 0.05 0.05 0.04 0.04 0.03 0.02 0.02 0.03 0.05
0.08 0.06 0.04 0.02 0.04 0.04 0.08 0.07 0.05 0.04 0.03 0.02 0.01 0.02 0.03 0.02 0.02 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.02 0.00 0.01 0.03
0.79 0.83 0.90 0.84 0.71 0.86 0.81 0.78 0.76 0.83 0.84 0.87 0.88 0.87 0.86 0.87 0.90 0.89 0.90 0.89 0.90 0.91 0.93 0.91 0.92 0.90 0.89 0.83
0.83 0.79 0.89 0.79 0.89 0.86 0.81 0.74 0.87 0.75 0.81 0.81 0.83 0.79 0.84 0.84 0.83 0.78 0.73 0.73 0.71 0.74 0.83 0.85 0.83 0.84 0.83 0.69
ACCEPTED MANUSCRIPT
2009 2010 2011 2012 2013 2014
331 418 442 489 508 573
866.73 560.83 776.13 556.45 584.73 1533.99
0.74 0.78 0.84 0.82 0.80 0.80
0.15 0.07 0.06 0.05 0.07 0.11
0.28 0.39 0.36 0.37 0.37 0.33
0.06 0.03 0.02 0.03 0.02 0.02
0.02 0.02 0.03 0.01 0.01 0.02
0.90 0.90 0.90 0.93 0.92 0.89
0.71 0.84 0.83 0.83 0.90 0.86
Mean 1.50 0.81 -3.62 6.30 0.73 0.32 0.30 0.32 0.58 0.79 44.11 21.23
Median 0.57 0.86 -7.74 6.30 0.65 0.18 0.31 0.18 0.52 0.87 35.74 18.17
Std. dev. 9.40 0.18 50.41 2.01 0.73 0.70 0.63 0.35 0.74 0.21 41.95 21.02
Min Q1 -22.93 -3.46 0.01 0.72 -113.35 -32.71 1.84 4.84 -0.80 0.25 -0.72 -0.06 -1.51 -0.06 0.05 0.09 -1.12 0.12 0.13 0.69 -42.95 19.01 -22.14 7.30
AN US
N 19,119 19,119 17,218 19,119 19,119 19,119 18,996 19,119 3,435 3,887 3,624 3,678
AC
CE
PT
ED
M
Variables CAR RPR BHAR Size Ln(M/B) Past return RKRV Rel. size Target Ln(M/B) Target RPR Bid premium Target CAR
CR IP T
Panel B: Summary statistics
Page 55
Q3 5.37 0.95 17.96 7.69 1.12 0.49 0.65 0.40 0.98 0.96 60.74 31.70
Max 36.57 1.00 211.62 11.12 3.19 3.81 2.16 1.72 3.03 1.00 231.72 94.91
CR IP T
ACCEPTED MANUSCRIPT
AN US
Table 2 Acquirer cumulative abnormal returns (CAR) by RPR groups. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. The sample is two-way sorted yearly on the target listing status (private vs. public) and the acquirer RPR ratio. The Low and High RPR subsamples are formed by the sample medians of acquirer RPR ratios within each listing-status group. Panel A shows the mean RPR values and the number of observations for the whole sample and the subsamples of private and public targets, respectively. In Panel B, the sample is three-way sorted on target listing status, acquirer RPR, and acquirer M/B ratios. For the whole sample and the subsamples, Panel B reports the mean values of the acquirer cumulative abnormal returns (CAR). Row “H – L” presents the differences in CAR between the High RPR and Low RPR groups, with t-statistics in square brackets. Acquirer CAR is the cumulative abnormal returns to the acquirer, measured over the event window [–5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. RPR is the acquirer’s stock price as of t – 6, the sixth day prior to the announcement, divided by the highest closing price over the past 252 days relative to t – 6. Acquirer M/B is the acquirer’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
All targets 19,119 9,543 9,576
Panel B: Acquirer CAR (%) by target listing status, acquirer RPR, and acquirer M/B Groups All targets Private targets All Low High All Low M/B M/B M/B M/B M/B All RPR 1.501a 1.624a 1.378a 2.320a 2.364a Low RPR (L) 2.021a 2.272a 1.787a 2.916a 3.124a a a a a High RPR (H) 0.982 1.017 0.945 1.726 1.648a
High M/B 2.276a 2.719a 1.808a
All M/B -0.911a -0.618a -1.201a
H-L t(H - L)
-0.911 [-3.89]a
-0.582 [-2.36]b
-1.255 [-6.89]a
AC
CE
-1.040 [-7.66]a
PT
ED
M
Panel A: Acquirer RPR levels RPR groups Average acquirer RPR level All targets Private targets Public targets All RPR 0.807 0.800 0.829 Low RPR (L) 0.677 0.665 0.713 High RPR (H) 0.937 0.934 0.944
-0.842 [-4.18]a
-1.190 [-7.48]a
Page 56
-1.476 [-6.86]a
No. of obs. Private targets 14,270 7,128 7,142
Public targets 4,849 2,415 2,434
Public targets Low M/B -0.564a -0.278 -0.827a -0.549 [-1.72]c
High M/B -1.255a -0.932a -1.600a -0.668 [-1.78]c
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Table 3 Baseline acquirer CAR regressions. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. This table presents the coefficients and t-statistics (in parentheses) for four OLS regressions based on the whole sample (All), the subsample of deals involving only private targets (Private), the subsample of deals involving only public targets (Public), and the subsample of deals involving targets that are covered in CRSP (CRSP). Estimated t-statistics are based on robust standard errors clustered by firm. The dependent variable is the acquirer CAR as measured over the seven-day event window [-5, +1]. All regressions include Fama and French (1997) industry and year fixed effects. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 57
ACCEPTED MANUSCRIPT
Stock Cash Private Stock*Private Rel. size Size Leverage Dormant > 1 yr
M
Same industry Tender offer
ED
Hostile
CE
Past return
PT
Toehold Cross border
Private -6.286 (-8.66)a -0.168 (-1.12) 0.709 (2.05)b -0.138 (-0.74)
3.873 (11.33)a -0.249 (-3.97)a -0.670 (-1.51) 0.512 (2.04)b -0.079 (-0.43) 2.751 (0.82) -2.919 (-1.53) 0.566 (0.56) -0.352 (-1.32) 0.792 (4.24)a
Public -2.581 (-1.97)b -0.460 (-1.63) -0.402 (-1.31) 1.187 (3.74)a
CRSP -3.100 (-1.92)c -0.228 (-0.67) 0.078 (0.21) 1.429 (3.64)a
-0.486 (-1.34) -0.579 (-6.54)a 1.252 (1.49) 1.091 (2.59)a 0.637 (2.24)b 0.651 (1.77)c -0.148 (-0.36) -0.479 (-0.88) 0.413 (0.77) 0.774 (2.13)b
-0.900 (-2.19)b -0.466 (-4.19)a 0.961 (1.02) 1.276 (2.66)a 0.589 (1.75)c 0.801 (1.88)c 0.371 (0.79) -0.207 (-0.33) -1.876 (-1.56) 0.713 (1.68)c 4.607 (4.28)a -0.524 (-1.70)c 3.969 (2.21)b 3,336 0.048
CR IP T
Ln(M/B)
All -5.487 (-8.56)a -0.285 (-2.14)b -1.208 (-4.06)a 0.145 (0.91) 2.321 (10.60)a 2.125 (5.04)a 2.216 (8.80)a -0.398 (-7.69)a -0.081 (-0.21) 0.628 (2.88)a 0.107 (0.67) 1.058 (2.22)b -0.759 (-1.89)c 0.047 (0.10) -0.187 (-0.77) 0.777 (4.70)a
AN US
Variables RPR
Target RPR
AC
Target Ln(M/B) Intercept N Adj. R2
7.284 (9.46)a 19,119 0.064
8.711 (10.45)a 14,270 0.054
Page 58
7.776 (5.03)a 4,849 0.049
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 4 Acquirer CAR regressions with interaction terms. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. This table presents the coefficients and t-statistics (in parentheses) for ten OLS regressions, with the first five based on the subsample of acquisitions involving private targets and the last five based on the subsample of acquisitions involving public targets. Estimated t-statistics are based on robust standard errors clustered by firm. The dependent variable is the acquirer CAR as measured over the seven-day event window [-5, +1]. The continuous interaction variables Sigma, Rel. size, and Indiv. own. are demeaned. The full regression specifications follow those of Table 3. For brevity, coefficients and t-statistics of all other variables are not shown. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Variables Private targets Public targets Model 1 2 3 4 5 Model 6 7 8 9 10 RPR -4.697 -3.480 -6.290 -3.564 -5.226 -2.927 -0.356 -1.829 -3.684 -2.187 (-5.73)a (-3.75)a (-8.66)a (-2.80)a (-7.36)a (-1.97)b (-0.20) (-1.38) (-1.83)c (-1.69)c RPR*Sigma -0.455 -0.534 (-2.90)a (-1.94)c RPR*Few analysts -4.822 -3.815 (-4.04)a (-1.87)c RPR*Rel. size -6.623 -5.461 (-3.42)a (-2.10)b RPR*Non-cash -3.364 1.324 (-2.40)b (0.61) RPR*Indiv. own. -0.112 -0.103 (-5.40)a (-2.82)a Sigma 0.364 0.241 (3.01)a (1.08) Few analysts 3.807 2.939 (3.69)a (1.63) Indiv. own. 0.094 0.092 (5.21)a (2.87)a Rel. size 3.857 3.844 8.969 3.859 3.841 -0.447 -0.518 4.023 -0.493 -0.492 (11.31)a (11.27)a (5.53)a (11.29)a (11.27)a (-1.23) (-1.42) (1.77)c (-1.36) (-1.36) N 14,270 14,270 14,270 14,270 14,270 4,849 4,849 4,849 4,849 4,849 Adj. R2 0.056 0.056 0.057 0.055 0.058 0.052 0.050 0.051 0.049 0.053 Page 59
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 5 Long-term acquirer returns. The sample includes a total of 17,218 acquisitions made by U.S. acquirers over the 1981–2014 period, for which the BHAR data are available. In the whole sample and the two subsamples formed on target listing status, the observations are further sorted yearly into two equal groups based on acquirer RPR, acquirer M/B, or acquirer RKRV. For each of the subsamples, Panel A presents the average BHAR, Panel B presents the Fama and French (2015) five-factor monthly alphas of calendar-time portfolio returns (CTPR), and Panel C presents the average abnormal returns of the four earnings announcements over the subsequent one-year period. Both BHAR and CTPR returns are measured over the one-year period starting from the month after the announcement. For each of the subsequent quarterly earnings announcements, the abnormal returns are measured over the [-1,+1] window, adjusted by the CRSP equal-weighted market returns. Rows “H – L” show the return differences of the two groups formed on acquirer RPR, acquirer M/B, or acquirer RKRV, with t-statistics in square brackets. Panel D presents regressions of BHAR based on the whole sample (All), the private and public subsamples, and the subsample of deals involving targets covered in CRSP, respectively. Estimated t-statistics (in parentheses) are based on robust standard errors clustered by firm. All regressions include Fama and French (1997) industry and year fixed effects. All variables are defined in Appendices A and B. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 60
CR IP T
ACCEPTED MANUSCRIPT
Panel A: Acquirer buy-and-hold abnormal returns (BHAR) (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -3.625a -3.388a -4.278a -3.625a -3.388a Low (L) -5.889a -5.589a -6.720a -2.792a -2.470a a b b a High (H) -1.369 -1.194 -1.854 -4.455 -4.305a
Public -4.278a -3.683a -4.868a
H–L t(H - L)
-1.835 [-1.95]c
-1.185 [-0.93]
Panel B: Acquirer Fama and French (2015) five-factor monthly alphas (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -0.057 -0.034 -0.194b -0.057 -0.034 Low (L) -0.165 -0.124 -0.342b 0.061 0.086 High (H) 0.017 0.011 -0.052 -0.171 -0.157
Public -0.194b -0.199c -0.195
4.395 [4.68]a
4.866 [3.83]a
0.003 [0.02]
Panel C: Acquirer average earnings announcement-period abnormal returns (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -0.006 0.010 -0.050 -0.006 0.010 Low (L) -0.108b -0.080 -0.185b 0.139a 0.186a High (H) 0.096b 0.100b 0.085 -0.150a -0.165a
Public -0.050 0.009 -0.109
H–L t(H – L)
-0.118 [-1.13]
0.135 [0.87]
-0.231b [-2.30]b
0.180 [2.27]b
0.270 [2.58]a
-0.289 [-4.47]a
AC
CE
0.204 [3.16]a
PT
ED
0.181 [1.29]
M
-0.242b [-2.02]b
H-L t(H – L)
0.290c [1.76]c
-1.663 [-2.16]b
AN US
4.520 [5.89]a
Page 61
-0.351 [-4.42]a
Sort by RKRV All targets Private Public -3.628a -3.400a -4.256a -2.776a -2.684a -3.028a a a -4.477 -4.113 -5.476a -1.701 [-2.21]b
-1.429 [-1.52]
-2.448 [-1.92]c
Sort by RKRV All targets Private Public -0.063 -0.041 -0.198b 0.146 0.192 -0.065 -0.272a -0.271b -0.333a -0.419a [-4.05]a
-0.464a [-4.05]a
-0.268c [-1.95]c
Sort by RKRV All targets Private Public -0.005 0.012 -0.050 0.056 0.067 0.028 -0.066 -0.043 -0.127c -0.122 [-1.89]c
-0.110 [-1.39]
-0.154 [-1.47]
ACCEPTED MANUSCRIPT
ED
PT
CE
CRSP 20.677 (2.59)a -1.098 (-0.63) -2.069 (-1.07) 5.389 (2.48)b
0.329 (0.20) 0.501 (1.17) -4.092 (-0.94) 0.200 (0.09) 4.511 (2.96)a -1.198 (-0.60) 1.107 (0.45) 0.177 (0.05) 0.937 (0.34) -4.882 (-3.04)a
0.973 (0.50) 0.439 (0.81) -4.295 (-0.83) -1.391 (-0.56) 5.014 (2.77)a -0.579 (-0.25) -0.528 (-0.19) -3.127 (-0.81) -5.270 (-1.06) -3.700 (-1.78)c 3.465 (0.65) -1.831 (-1.25) -30.506 (-3.39)a 3,257 0.030
AN US
-1.167 (-0.67) -0.769 (-1.92)c 5.381 (1.86)c 2.033 (1.41) 3.945 (3.78)a 3.970 (0.50) -8.702 (-1.12) -1.723 (-0.39) -1.627 (-1.06) -2.830 (-2.50)b
Public 24.813 (4.01)a -1.139 (-0.81) -2.827 (-1.80)c 4.008 (2.28)b
CR IP T
Private 20.752 (5.03)a -1.674 (-1.75)c -4.075 (-2.11)b 0.861 (0.77)
M
Panel D: Regressions of acquirer BHAR Variables All RPR 20.803 (5.86)a Ln(M/B) -1.447 (-1.77)c Stock -3.214 (-2.14)b Cash 1.324 (1.42) Private 0.747 (0.66) Stock*Private -0.845 (-0.38) Rel. size -1.019 (-0.80) Size -0.421 (-1.37) Leverage 2.905 (1.17) Dormant > 1 yr 1.628 (1.36) Same industry 4.039 (4.58)a Tender offer 0.694 (0.36) Hostile 0.529 (0.22) Toehold -0.400 (-0.14) Cross border -1.181 (-0.88) Past return -3.134 (-3.18)a Target RPR
AC
Target Ln(M/B) Intercept N Adj. R2
-20.965 (-4.65)a 17,218 0.016
-18.531 (-3.56)a 12,637 0.015
Page 62
-27.530 (-3.92)a 4,581 0.030
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Table 6 Bid premium. The sample includes a total of 2,757 acquisitions made by U.S. acquirers over the 1981–2014 period involving targets covered in CRSP that have data on bid premium and target announcement returns. This table presents the coefficients and t-statistics (in parentheses) for five OLS regressions on bid premium, one for the whole sample and four on the subsamples. The Low (High) sigma subsample includes deals with acquirer sigma below (above) the sample median; the Cash subsample includes all deals with cash as the sole consideration of payment; the Non-cash subsample includes all deals with at least some payment not in cash. All regressions include Fama and French (1997) industry and year fixed effects. Estimated tstatistics are based on robust standard errors clustered by firm. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Whole sample Low sigma High sigma Cash Non-cash RPR 9.814 0.454 17.782 -25.359 15.649 b c (1.42) (0.04) (2.06) (-1.82) (1.96)c Ln(M/B) 4.207 3.499 3.961 2.996 4.726 (2.83)a (1.66)c (1.84)c (1.12) (2.61)a Target RPR -10.132 -16.713 -2.620 -27.785 -8.614 (-1.96)c (-1.88)c (-0.40) (-3.04)a (-1.33) Target Ln(M/B) -6.130 -2.886 -8.808 -5.025 -6.492 a a c (-3.98) (-1.44) (-3.90) (-1.94) (-3.41)a Rel. size -3.830 -5.670 -2.643 -2.102 -3.306 b b (-2.21) (-2.43) (-1.02) (-0.52) (-1.64) Size 0.013 0.012 -0.142 1.472 -0.260 (0.02) (0.02) (-0.16) (1.26) (-0.42) Leverage -4.088 -4.235 -1.966 9.348 -7.336 (-0.89) (-0.58) (-0.31) (1.17) (-1.31) Dormant > 1 yr 2.846 4.752 -0.071 -1.127 4.870 (1.14) (1.40) (-0.02) (-0.25) (1.54) Same industry -2.379 -5.043 2.093 -4.337 -0.926 (-1.44) (-2.41)b (0.82) (-1.49) (-0.45) Toehold -3.346 -1.114 -3.948 -5.257 -1.805 (-0.90) (-0.21) (-0.73) (-1.02) (-0.36) Cross border 3.456 2.447 10.138 4.381 4.965 (0.57) (0.32) (1.13) (0.61) (0.69) Past return 5.220 -1.689 4.652 -1.822 6.191 a b (3.07) (-0.41) (2.53) (-0.51) (3.29)a Intercept 41.395 60.131 26.776 74.611 34.827 (5.18)a (4.16)a (2.31)b (5.07)a (3.61)a N 2,757 1,371 1,386 637 2,120 Adj. R2 0.090 0.128 0.068 0.096 0.094
Page 63
ACCEPTED MANUSCRIPT
Rel. size Size Leverage Dormant > 1 yr Same industry
CE
Toehold Cross border
AC
Past return Intercept N Adj. R2
High sigma 21.564 (4.81)a 0.129 (0.13) -17.449 (-4.83)a -7.428 (-7.50)a -6.449 (-5.01)a 0.876 (1.73)c -3.174 (-0.97) -3.000 (-1.50) 3.189 (2.30)b -1.458 (-0.58) 2.086 (0.45) 0.899 (1.09) 27.363 (4.43)a 1,386 0.143
AN US
Target Ln(M/B)
Low sigma 2.060 (0.27) 1.628 (1.46) -33.898 (-6.97)a -3.445 (-3.39)a -8.836 (-7.09)a 0.346 (0.88) 0.793 (0.20) 2.714 (1.47) -1.410 (-1.15) -5.811 (-3.00)a 0.449 (0.08) 1.445 (0.55) 49.850 (5.91)a 1,371 0.230
M
Target RPR
ED
Ln(M/B)
Whole sample 17.556 (4.75)a 0.951 (1.26) -21.890 (-7.75)a -5.707 (-8.12)a -7.363 (-8.15)a 0.697 (2.29)b -1.622 (-0.68) 0.004 (0.00) 0.128 (0.14) -3.483 (-2.25)b -0.362 (-0.11) 1.444 (1.71)c 29.738 (6.74)a 2,757 0.153
PT
Variables RPR
CR IP T
Table 7 Target CAR. The sample includes a total of 2,757 acquisitions made by U.S. acquirers over the 1981–2014 period involving targets covered in CRSP that have data on bid premium and target CAR. This table presents the coefficients and t-statistics (in parentheses) for five OLS regressions on target CAR, one for the whole sample and four on the subsamples. The Low (High) sigma subsample includes deals with acquirer sigma below (above) the sample median; the Cash subsample includes all deals with cash as the sole consideration of payment; the Non-cash subsample includes all deals with at least some payment not in cash. All regressions include Fama and French (1997) industry and year fixed effects. Estimated t-statistics are based on robust standard errors clustered by firm. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 64
Cash -0.165 (-0.02) -1.058 (-0.75) -46.359 (-7.93)a -3.571 (-2.34)b -5.764 (-2.36)b 1.233 (1.80)c 0.373 (0.08) -0.099 (-0.04) -1.559 (-0.91) -1.556 (-0.54) 4.150 (0.59) -2.225 (-1.02) 63.423 (6.75)a 637 0.235
Non-cash 19.132 (4.56)a 2.050 (2.35)b -19.402 (-5.98)a -5.924 (-7.52)a -6.290 (-6.21)a 0.670 (2.01)b -1.300 (-0.46) 0.023 (0.01) 1.388 (1.31) -4.687 (-2.62)a -0.787 (-0.21) 2.189 (2.39)b 22.744 (4.54)a 2,120 0.136
ACCEPTED MANUSCRIPT
Table 8 Probabilistic regressions of deal characteristics. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981 –2014 period, with 14,270 involving private targets, 4,849 involving public targets, and 3,336 involving targets covered in CRSP. Panels A and B present 15 Probit regression models with the dependent variables shown in the top rows and samples in the second rows. Year fixed effects are included but not reported. All variables are defined in Appendix A. Estimated t-statistics are in parentheses. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Private -0.038 (-0.34) 0.198 (9.50)a
Stock Public 0.243 (1.34) 0.163 (4.98)a
-0.041 (-3.60)a 0.136 (3.06)a 0.129 (2.10)b 4.616 (8.38)a -0.306 (-5.87)a 0.180 (5.96)a 0.512 (2.61)a -0.216 (-4.07)a 0.062 (2.73)a -1.574 (-12.64)a 0.148 14,270 12,498 1,772
-0.032 (-2.41)b 0.116 (2.34)b 0.552 (6.29)a 7.909 (8.00)a -0.348 (-4.85)a 0.191 (4.36)a -0.300 (-2.81)a -0.318 (-4.08)a 0.057 (1.34) -0.612 (-3.67)a 0.141 4,849 2,939 1,910
RPR Ln(M/B) Target RPR
Leverage Sigma Dormant > 1 yr Same industry Toehold
CE
Cross border
ED
Rel. size
PT
Size
AC
Past return Intercept
McFadden's LRI N N(dep. var. = 0) N(dep. var. = 1)
M
Target Ln(M/B)
CRSP 0.384 (1.68)c 0.080 (1.87)c -0.582 (-3.99)a 0.152 (3.77)a -0.028 (-1.58) 0.126 (2.12)b 0.743 (6.71)a 9.503 (7.26)a -0.351 (-4.18)a 0.155 (2.95)a -0.369 (-2.97)a 0.115 (0.71) 0.077 (1.43) -0.192 (-0.85) 0.146 3,336 1,982 1,354
Private 0.105 (1.02) -0.067 (-3.47)a
AN US
Variables
CR IP T
Panel A: Method of payment
Page 65
0.065 (6.97)a -0.149 (-3.48)a -0.065 (-1.22) -3.666 (-6.65)a 0.083 (2.29)b 0.090 (3.68)a -0.679 (-3.07)a 0.037 (0.96) -0.049 (-2.02)b -0.868 (-10.15)a 0.077 14,270 10,933 3,337
Cash Public -0.169 (-0.79) -0.108 (-2.87)a
-0.015 (-1.01) -0.561 (-9.09)a -0.803 (-8.23)a -7.302 (-5.98)a 0.197 (2.79)a -0.303 (-6.52)a 0.301 (2.87)a 0.424 (5.64)a -0.079 (-1.43) 0.007 (0.04) 0.138 4,849 3,799 1,050
CRSP -0.356 (-1.32) -0.038 (-0.79) 0.658 (4.01)a -0.108 (-2.37)b -0.056 (-2.77)a -0.720 (-9.57)a -0.918 (-7.43)a -7.088 (-4.42)a 0.176 (2.09)b -0.346 (-6.15)a 0.283 (2.26)b -0.139 (-0.78) -0.047 (-0.70) -0.235 (-0.92) 0.154 3,336 2,599 737
ED
PT
CE
AC
Hostile Public -0.580 (-2.12)b 0.061 (1.28)
CRSP -0.701 (-2.12)b 0.122 (2.04)b 0.141 (0.67) -0.291 (-4.96)a 0.101 (4.08)a 0.576 (7.45)a -0.273 (-1.76)c -4.345 (-2.22)b 0.131 (1.29) -0.043 (-0.60) 0.864 (6.99)a 0.220 (1.08) 0.047 (0.59) -2.161
AN US
Private 0.045 (0.07) -0.222 (-1.71)c
0.175 (3.50)a 0.770 (5.07)a -0.013 (-0.04) -3.788 (-0.98) -0.457 (-1.63) 0.107 (0.78) 1.665 (6.48)a -0.195 (-0.54) -0.039 (-0.22) -8.918
M
Panel B: Tender offer, hostility, and deal success Variables Tender offer Private Public CRSP RPR 0.127 -0.135 -0.251 (0.23) (-0.63) (-0.96) Ln(M/B) -0.298 -0.030 -0.053 (-2.79)a (-0.76) (-1.06) Target RPR 0.117 (0.70) Target Ln(M/B) -0.060 (-1.32) Size 0.115 0.132 0.118 (2.69)a (7.94)a (5.58)a Rel. size 0.535 0.055 -0.009 (3.74)a (0.91) (-0.13) Leverage -0.656 -1.066 -1.065 (-2.52)b (-9.90)a (-8.07)a Sigma -3.141 -3.186 -3.332 (-0.99) (-2.55)b (-2.11)b Dormant > 1 yr -0.356 0.277 0.310 (-1.58) (3.83)a (3.74)a Same industry -0.293 -0.179 -0.210 (-2.44)b (-3.62)a (-3.63)a Toehold 1.414 0.438 0.420 (5.82)a (4.32)a (3.53)a Cross border 0.042 0.943 0.327 (0.15) (12.69)a (1.94)c Past return 0.153 -0.099 -0.032 (1.30) (-1.82)c (-0.51) Intercept -8.126 -1.827 -1.516
CR IP T
ACCEPTED MANUSCRIPT
Page 66
0.106 (5.31)a 0.618 (9.37)a -0.219 (-1.69)c -3.831 (-2.40)b 0.159 (1.82)c -0.092 (-1.50) 0.745 (7.02)a 0.341 (3.53)a -0.028 (-0.40) -2.340
Private 0.191 (1.66)c -0.020 (-0.91)
Success Public 0.416 (2.13)b -0.061 (-1.75)c
0.051 (4.50)a -0.230 (-5.24)a -0.407 (-6.34)a -2.355 (-4.07)a 0.033 (0.71) 0.040 (1.34) -0.144 (-0.80) -0.119 (-2.45)b 0.031 (1.18) 1.190
0.032 (2.19)b -0.396 (-7.68)a 0.168 (1.78)c -1.041 (-0.99) -0.033 (-0.46) 0.132 (2.85)a -0.307 (-3.10)a -0.233 (-2.97)a 0.039 (0.83) 0.395
CRSP 0.292 (1.18) -0.125 (-2.71)a 0.177 (1.13) 0.166 (3.77)a 0.024 (1.25) -0.418 (-6.68)a 0.138 (1.15) -0.485 (-0.35) -0.029 (-0.35) 0.118 (2.10)b -0.418 (-3.59)a -0.293 (-1.75)c 0.029 (0.50) 0.243
(-8.09)a 0.140 4,849 4,077 772
(-5.36)a 0.099 3,336 2,780 556
(-27.30)a 0.348 14,270 14,227 43
(-9.22)a 0.112 4,849 4,480 369
(-6.66)a 0.125 3,336 3,041 295
AC
CE
PT
ED
M
AN US
McFadden's LRI N N(dep. var. = 0) N(dep. var. = 1)
(-30.78)a 0.400 14,270 14,195 75
CR IP T
ACCEPTED MANUSCRIPT
Page 67
(10.62)a 0.043 14,270 1,384 12,886
(2.18)b 0.065 4,849 929 3,920
(0.99) 0.084 3,336 648 2,688
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 9 Robustness checks. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period, with 14,270 (4,849) involving private (public) targets. For the whole sample and subsamples, this table presents regressions of acquirer CAR by varying CARs in Panel A, varying RPRs in Panel B, and subperiods in Panel C. CAR 1 is estimated as in Moeller et al. (2004), CAR 2 is estimated over the [-5, +1] window with return adjusted by market return, and CAR 3 follows Lang et al. (1991). RPR 1 is measured on the 11 th day prior to the announcement, RPR 2 is measured at the month-end prior to the announcement, and RPR 3 is the RPR quintiles (0 to 4). The three subperiods are defined as 1981–1990 (1980s), 1991–2000 (1990s), and 2001–2014 (2000s) based on the year of announcements. The full regression specifications follow those of Table 3. For brevity, coefficients and t-statistics (in parentheses) of all other variables are not shown. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Panel A: Robustness by alternative CARs Variables All targets Private targets Public targets CAR 1 CAR 2 CAR 3 CAR 1 CAR 2 CAR 3 CAR 1 CAR 2 CAR 3 RPR -4.477 -5.989 -4.938 -5.143 -6.661 -5.728 -1.584 -3.590 -2.168 (-8.88)a (-9.50)a (-6.65)a (-9.35)a (-9.25)a (-6.74)a (-1.48) (-2.89)a (-1.46) N 19,119 19,119 19,119 14,270 14,270 14,270 4,849 4,849 4,849 Adj. R2 0.073 0.070 0.057 0.054 0.060 0.047 0.073 0.051 0.054 Panel B: Robustness by alternative RPRs Variables All targets Private targets Public targets RPR 1 RPR 2 RPR 3 RPR 1 RPR 2 RPR 3 RPR 1 RPR 2 RPR 3 RPR -4.245 -3.508 -0.410 -5.027 -4.018 -0.508 -1.319 -2.015 -0.158 (-6.57)a (-5.69)a (-5.66)a (-6.88)a (-5.68)a (-5.95)a (-1.00) (-1.63) (-1.19) N 19,119 19,119 19,119 14,270 14,270 14,270 4,849 4,849 4,849 Adj. R2 0.061 0.060 0.059 0.051 0.049 0.048 0.048 0.049 0.048 Panel C: Robustness by subperiods Variables All targets Private targets Public targets 1980s 1990s 2000s 1980s 1990s 2000s 1980s 1990s 2000s RPR -5.811 -7.725 -3.390 -6.467 -8.452 -3.947 -4.219 -3.881 -2.032 (-4.42)a (-7.12)a (-3.82)a (-4.11)a (-6.88)a (-4.04)a (-1.82)c (-1.89)c (-0.95) N 3,112 8,424 7,583 2,154 6,247 5,869 958 2,177 1,714 Adj. R2 0.070 0.078 0.050 0.056 0.072 0.037 0.074 0.044 0.075
Page 68
ACCEPTED MANUSCRIPT
Panel A: Acquisitions of private targets
3.00 2.50
CR IP T
2.00
1.50
Low RPR High RPR
1.00 0.50
Days relative to announcement date
0.00 -6
-4
-2
0
2
Panel B: Acquisitions of public targets
8
10
12
14
16
18
20
Low RPR High RPR
CE
PT
-0.50
-1.50
6
ED
0.00
-1.00
4
M
0.50
-2.00
Days relative to announcement date
-2.50
AC
Cumulative abnormal return (CAR, in %)
1.00
AN US
Cumulative abnormal return (CAR, in %)
3.50
-6
-4
-2
0
2
4
6
8
10
12
14
16
18
20
Fig. 1. Acquirer cumulative abnormal returns and reference price ratios. The sample includes 14,270 acquisitions of private targets (Panel A) and 4,849 acquisitions of public targets (Panel B) over the 1981–2014 period made by U.S. acquirers. Within each subsample, acquirers are yearly sorted into two equal groups by their reference price ratios as of the sixth day prior to the announcement date. Low RPR (High RPR) represents the group with reference price ratios below (above) the subsample median. The abnormal returns (in %) are based on a market model estimated from days [-370, -253] relative to the announcement date. Page 69
Acquirer Reference Prices and Acquisition Performance Qingzhong Ma , David A. Whidbee , Wei Zhang PII: DOI: Reference:
S0304-405X(18)30280-0 https://doi.org/10.1016/j.jfineco.2018.10.004 FINEC 2976
To appear in:
Journal of Financial Economics
Received date: Revised date: Accepted date:
27 February 2017 5 February 2018 5 March 2018
Please cite this article as: Qingzhong Ma , David A. Whidbee , Wei Zhang , Acquirer Reference Prices and Acquisition Performance, Journal of Financial Economics (2018), doi: https://doi.org/10.1016/j.jfineco.2018.10.004
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Acquirer Reference Prices and Acquisition Performance*
CR IP T
Qingzhong Ma David A. Whidbee Wei Zhang This version: 2018 03 18 Abstract In a comprehensive sample of mergers and acquisitions, we find a reference price effect: Acquirers earn higher (lower) announcement-period returns when their
AN US
pre-announcement stock prices are well below (near) their 52-week highs. This reference price effect is stronger in acquisitions of private targets, deals involving greater uncertainty, and acquirers with greater individual investor ownership, and it is reversed in the subsequent year. Further, acquirer reference prices affect bid
M
premia and target announcement-period returns in deals with greater uncertainty in acquirer valuation. The overall evidence is consistent with investors irrationally
ED
using 52-week high prices as a measure of acquirer valuation.
valuation.
PT
Keywords: Mergers; Acquisitions; Anchoring; Reference point; Perceived
*
CE
JEL classification: G31, G34
AC
Ma: 325 Tehama Hall, College of Business, California State University, Chico, CA. Tel: 530 898 6182; email: [email protected]. Whidbee (corresponding author): Todd Hall Addition 470, PO Box 644746, Carson College of Business, Washington State University, Pullman, WA. Tel: 509 335 3098; email: [email protected]; fax:509 335 3857. Zhang: 465 Tehama Hall, College of Business, California State University, Chico, CA. Tel: 530 898 6544; email: [email protected]. We are very grateful to Bill Schwert (the editor) and to Ming Dong (the referee) for thoughtful comments and suggestions that substantially improved the paper. We also appreciate discussions, comments, and suggestions from Mehmet Akbulut, Malcolm Baker (AFA discussant), Dalen Chiang, Jim Downing, Levon Goukasian, Jarrad Harford, Kateryna Holland (FMA discussant), Christine Hsu, Emily Jian Huang, Yoonsoo Nam, Micah Officer, Richard Ponarul, Sergey Smirnov, Hai Tran (CCFC discussant), Stephen Treanor, and seminar participants at California State University at Chico, Southwest University of Finance and Economics, Chengdu, China, Washington State University, California Corporate Finance Conference (CCFC) 2016 at Loyola Marymount University, the Financial Management Association (FMA) 2016 Annual Meeting, and the American Finance Association (AFA) 2017 Annual Meeting. All remaining errors are the authors’. Earlier versions of the paper were circulated under the title “Anchoring and Acquisitions.”
ACCEPTED MANUSCRIPT
1. Introduction The stock market’s reaction to announcements of mergers and acquisitions (M&A) has been extensively examined. 1 Most existing studies assume investors rationally process the information available at the time of the announcement and incorporate it into stock prices. In
CR IP T
recent years, however, a growing body of literature challenges the notion that stock prices rationally reflect public information in a timely manner. Several studies find evidence that a stock price’s proximity to past peak prices affects investor behavior. George and Hwang (2004), for example, find that a stock price’s proximity to its 52-week high predicts returns better than
AN US
traditional momentum strategies.2 Baker et al. (2012) find strong evidence that offer premia paid to publicly traded targets are significantly affected by the 52-week high prices of the targets, suggesting that reference prices play an important role in perceived valuation levels of targets in
M
acquisitions. In this paper, we investigate whether 52-week high prices play a role in the perceived valuation levels of acquirers at the M&A announcements.3
ED
In a comprehensive sample of 19,119 acquisitions over the 1981–2014 period involving both public and private targets, we find that bidders with pre-announcement stock prices well
PT
below their 52-week highs earn significantly higher announcement-period abnormal returns than
CE
those with prices at or near their 52-week highs. That is, the acquirer’s reference price ratio (or RPR, the ratio of the pre-announcement stock price to its 52-week high price) is negatively 1
AC
The literature is vast and is comprehensively surveyed in Jensen and Ruback (1983), Jarrell et al. (1988), Andrade et al. (2001), and Betton et al. (2008). 2 Additional evidence includes Heath et al. (1999) who find a strong tendency for company executives to exercise stock options when the underlying stock price moves past its 52-week high. Huddart et al. (2009) document a significant increase in trading volume when a stocks’ price crosses its 52-week high. Driessen et al. (2013) examine implied volatilities derived from option prices and find that implied volatilities decrease for stocks approaching their 52-week high, but then increase after prices cross this threshold. 3 Many of the studies examining the influence of the 52-week high on investor behavior also consider the role of the 52-week low and often find that the historic low plays a statistically significant role. However, the 52-week high seems to have a stronger influence. In addition, most acquisitions are made by firms with stock prices closer to their 52-week high than their low. In our sample, for example, 65% of acquirers have stock prices nearer to their 52-week high than their 52-week low. Therefore, we focus on the role of the 52-week high in the bulk of our analysis. Nevertheless, we also consider the role of the 52-week low in our robustness checks. Page 1
ACCEPTED MANUSCRIPT
related to the acquirer’s announcement-period abnormal returns. We call this result the reference price effect.4 Like other studies examining how the 52-week high affects investor behavior, we focus on the anchoring phenomenon, which is considered one of “the most reliable and robust” results of experimental psychology (Kahneman, 2011). We refer to the possibility that anchoring
CR IP T
influences investors’ responses to merger announcements and merger outcomes as the perceived valuation hypothesis.
According to the anchoring phenomenon, individuals commonly rely on salient, even if fundamentally irrelevant, anchors in forming beliefs or norms (Tversky and Kahneman, 1974). If
AN US
the 52-week high is a meaningful price reference point for investors, as suggested by the empirical evidence, it may influence investors’ perceptions of valuation levels in mergers in much the same way as other valuation levels. Typically, valuation levels refer to the relation
M
between current stock prices and their book values or estimated fundamental values. These conventional valuation levels influence market reactions to acquisition announcements and deal
ED
outcomes (e.g., Rhodes-Kropf et al., 2005; Dong et al., 2006). We hypothesize that the relation between the current stock price and recent peak values, such as the 52-week high, may also
PT
influence mergers and acquisitions. Specifically, if the acquirer's stock price is near its 52-week
CE
high, investors may perceive the stock as having a relatively high valuation level and therefore be reluctant to bid up the price and be more inclined to sell down the price in response to an
AC
acquisition announcement. By contrast, if the acquirer's pre-announcement stock price is well below the 52-week high, investors may perceive the acquirer's valuation to be low and therefore be more willing to bid up rather than sell down the stock price.
4
To be clear, our focus is on acquirer reference prices, as compared to the target reference price effect documented in Baker et al. (2012). For convenience, we use the term “reference price effect” to refer to the acquirer reference price effect. Whenever necessary to avoid confusion or to be more precise, we use the more explicit term “acquirer reference price effect.” Further, we use the terms acquirer and bidder interchangeably throughout. Page 2
ACCEPTED MANUSCRIPT
An additional implication of the anchoring phenomenon is that anchors are more likely to influence numeric estimates when subjects have less knowledge (Mussweiler and Strack, 2000) and/or there is greater uncertainty (Jacowitz and Kahneman, 1995). Therefore, the reference price effect should be more important in acquisitions with limited information about the target,
CR IP T
transactions that involve greater uncertainty, acquirers with an opaque information environment, and acquirers owned by less sophisticated investors. Consistent with this implication, we find a stronger reference price effect among acquisitions of private targets (including independent private firms and subsidiaries of other companies) than acquisitions of public targets. Further, the
AN US
reference price effect is stronger among acquirers with higher stock return volatility, acquirers that are followed by fewer analysts, acquisitions that are relatively large, and acquirers with higher individual investor ownership levels.
M
A possible alternative explanation for our findings is that the reference price effect is due to the impact of conventional valuation levels. When a bidder’s stock price is near its 52-week
ED
high, its stock price might also be high relative to its fundamental or intrinsic value. In the context of acquisitions, overvalued acquirers earn lower returns (e.g., Shleifer and Vishny, 2003;
PT
Rhodes-Kropf and Viswanathan, 2004). However, the reference price effect remains strong even
CE
after controlling for a host of firm and deal characteristics, including a variety of proxies for stock valuation level, suggesting the reference price effect is not explained by conventional
AC
valuation levels or other firm or deal characteristics. Further, in acquisitions of private targets, we find a significant reference price effect even among acquisitions paid entirely with cash, a sample least likely to be influenced by conventional valuation levels. Yet another possibility is that the reference price effect acts as a proxy for some fundamental but unidentified factor associated with acquisition decisions. For example, fueled by
Page 3
ACCEPTED MANUSCRIPT
high stock prices relative to their 52-week highs, managers might develop hubris or become overconfident (Roll, 1986; Malmendier and Tate, 2008) and make poor acquisition decisions. Higher stock prices might also give managers greater leverage over monitoring mechanisms, such as the board of directors and the market for corporate control, and protect them when they
CR IP T
engage in activities that benefit themselves at the expense of shareholder value (e.g., Amihud and Lev, 1981; Jensen, 1986; Morck et al., 1990; Fu et al., 2013; Akbulut, 2013; Duchin and Schmidt, 2013). As another possibility, when a company’s stock price is near its 52-week high, investors might have anticipated the acquisition and incorporated such an anticipation into the
AN US
price already. Given that acquirers on average earn positive returns, at least in our sample, such an anticipation may lead to lower announcement returns for anticipated acquisitions. Finally, bidders with stock prices near their 52-week highs may, indeed, be overvalued and the reference
M
price effect may be a rational response to acquisition announcements by overvalued acquirers. These indirect effects are plausible and may be reflected in the market’s reaction to the merger
ED
announcement.
If this is the case, however, we would expect the market’s reaction to these merger
PT
announcements to be permanent. By contrast, if the announcement-period return pattern is being
CE
driven by anchoring, an irrational behavioral bias, the return pattern is likely reversed over the long horizon when the bias is eventually corrected. To further distinguish the perceived valuation
AC
hypothesis from alternative explanations, we examine the long-horizon abnormal returns to merger announcements and find that the announcement-period return pattern is reversed over the subsequent one-year period. Specifically, the long-horizon abnormal returns for acquirers with pre-announcement stock prices well below their 52-week highs are lower than those for acquirers with pre-announcement stock prices near their 52-week highs. While statistical significance
Page 4
ACCEPTED MANUSCRIPT
varies, this conclusion holds for buy-and-hold abnormal returns in portfolio analysis and in regressions. The same pattern of return reversal is also observed in short-window earnings announcements over the one-year period following the M&A announcements. This finding is inconsistent with the reference price effect acting as a proxy for some unidentified fundamental
CR IP T
factor associated with merger decisions or as a proxy for conventional valuation levels. Rather, our results are more consistent with investors’ initial responses to acquisition announcements being irrationally influenced by perceived valuation levels relative to 52-week high prices.
The reference price effect survives a battery of robustness checks and falsification tests. It
AN US
is robust to alternative models of estimating abnormal returns and alternative reference price ratio measures. The effect is strong in the 1980s, 1990s, and the new millennium. We also examine the effects of bidder and target reference prices on bid premium and
M
target announcement return in the subsample of deals involving targets that are covered in the Center for Research in Security Prices (CRSP). We find that bidders with pre-announcement
ED
prices closer to their 52-week highs pay higher bid premia, although the effect is significant only among non-cash deals and deals involving more uncertainty. This is consistent with the
PT
interpretation that in deal negotiations, acquirers perceived to be overvalued are compelled to
CE
offer high premia to their targets. Also, as discussed in Dong et al. (2006), perceived acquirer overvaluation can influence target announcement returns for two reasons: Acquirer overvaluation
AC
should give rise to a higher bid premium, and bid announcement should alert the market to reassess the pre-existing (acquirer and target) misvaluation. Indeed, we observe that acquirers with stock prices closer to their 52-week highs are associated with higher target announcement returns in the whole CRSP sample and among the subsamples of non-cash deals and deals involving more uncertainty.
Page 5
ACCEPTED MANUSCRIPT
The 52-week high price of the target also has a significant impact on bid premium and target announcement return. Consistent with Baker et al. (2012), we find that targets with stock prices well below their 52-week highs are associated with higher bid premia in the whole CRSP sample, suggesting that targets perceived to be undervalued are able to negotiate high premia.
CR IP T
However, we find that the target reference price effect on bid premium is significant only in the subsample of cash deals or when acquirer uncertainty is low. A possible interpretation in the context of reference prices is that, when the uncertainty of bidder valuation is high (in deals with non-cash payment or high acquirer stock volatility), the target firm mainly uses the bidder
AN US
reference price to negotiate a high premium. When acquirer valuation uncertainty is low or in cash deals, on the other hand, the target primarily relies on the target price distance to its 52week high in the bid premium negotiation. Similarly, we find the ratio of target stock price to
M
target 52-week high exerting a negative impact on target announcement return in the CRSP sample, with a stronger effect in subsamples of low acquirer and deal uncertainty.
ED
In contrast to the reference price influence on target and bidder announcement-period returns and bid premia, we find little consistent evidence that reference prices (for either the
PT
bidder or target) affect the method of payment or deal combativeness. A possible interpretation is
CE
that stock market reactions depend on investor perceptions of acquirer and target valuations, and investors have less information than managers. Therefore, investors are more likely to rely on
AC
heuristics when assessing the valuation implications of an acquisition announcement. Similarly, even though an offer premium is a managerial decision, it is commonly the result of negotiations between acquirer and target managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and deal combativeness, on the other hand, are primarily decisions by the acquirer and target managers and, if acquirer and target managers are
Page 6
ACCEPTED MANUSCRIPT
more informed of their own firms’ true valuations, these decisions will be less affected by their stocks’ reference prices. The paper contributes to two important areas in the finance literature. First, we contribute to the M&A literature by documenting a new, non-fundamental factor in affecting the acquirer
CR IP T
and target investor reactions to merger announcements and bid premia. Our paper adds to the evidence that behavioral factors influence the takeover market. The literature already documents such behavioral factors as investor misvaluation (Dong et al., 2006) and managerial overconfidence (Malmendier and Tate, 2008). More recently, Danbolt et al. (2015) find a
AN US
positive relation between a measure of investor sentiment and acquirer announcement-period abnormal returns. Otherwise, most existing studies interpret market reactions at the announcements as reflecting expected changes in firm fundamentals. Part of the relatively recent
M
literature in this vein focuses on management performance (Lang et al., 1989), asset relatedness (Matsusaka, 1993), corporate governance (Masulis et al., 2007), free cash flows (Lang et al.,
ED
1991; Harford, 1999), firm size (Moeller et al., 2004), chief executive officer (CEO) network centrality (El-Khatib et al., 2015), target asset uncertainty (Officer et al., 2009), dormant period
PT
(Cai et al., 2011), management entrenchment (Harford et al., 2012), and cultural similarity
CE
(Bereskin et al., 2017). Studies of the long-term performance of mergers emphasize the role of over-extrapolation (Rau and Vermaelen, 1998), market-wide valuation (Bouwman et al., 2009),
AC
firm-level valuation (Akbulut, 2013; Fu et al., 2013), and merger waves (Duchin and Schmidt, 2013), among others. We document the impact of a behavioral bias, specifically the anchoring bias, on investors’ perceptions of valuation levels and their reactions to announcements of mergers and acquisitions. As a unique feature of this study, we show not only the bias in the announcement period but also its correction in the longer term.
Page 7
ACCEPTED MANUSCRIPT
Second, our paper adds further evidence on the role of anchoring as a psychological bias in behavioral finance. One of the central themes in behavioral finance is to understand what psychological phenomena help shape investors’ demand for securities (Barberis et al., 1998; Shleifer, 2000; Baker, 2009; Baker and Wurgler, 2012). For the special case of firms making
CR IP T
acquisitions, this paper reports evidence that anchoring bias and its influence on valuation-level perceptions help shape investor demand for stocks. That is, for acquiring firms with stock prices well below their 52-week highs, investors seem more (less) inclined to bid up (sell down) prices in response to an acquisition announcement. Because the 52-week high is fundamentally
AN US
irrelevant, this anchoring bias in the initial market reaction is corrected in the longer term. This reversal suggests that the acquirer reference price effect is irrational, adding a new insight beyond Baker et al. (2012). In addition, we find that the anchoring bias has a stronger impact
M
among acquisitions with limited information about the targets, acquisitions involving greater uncertainty about acquirer valuation, and acquirers with higher levels of ownership by less
ED
sophisticated investors, consistent with the notion that investor bias is more likely to exist when the difficulty of estimating firm value is greatest and the forces of arbitrage are relatively limited.
PT
We develop the perceived valuation hypothesis in Section 2 and describe the sample and
CE
data in Section 3. Empirical results of acquirer announcement-period and long-run returns are presented in Section 4. We examine deal characteristics in Section 5. We check the robustness of
AC
the main results in Section 6 and conclude in Section 7. 2. The perceived valuation hypothesis A growing literature finds evidence of stocks’ past peak prices, especially the 52-week
high, being important in explaining investor and manager behavior. The general premise is that the 52-week high acts as an important reference point. This argument is supported in numerous
Page 8
ACCEPTED MANUSCRIPT
decision scenarios, including employee stock option exercising (Heath et al., 1999), momentum (George and Hwang, 2004), post-earnings announcement drift (George et al., 2014), investment and financing anomalies (George et al., 2018), and trading volume (Huddart et al., 2009). Additionally, Li and Yu (2012) find evidence of investor anchoring on the Dow 52-week high.
CR IP T
The results of these studies suggest that 52-week highs influence perceptions of valuation levels. For example, Baker et al. (2012) find a target reference price effect indicating that merger decision makers and target shareholders look to the target’s 52-week high as an indicator of valuation levels when evaluating offer premia. To the extent investors, managers, and other
AN US
decision agents’ perceptions of valuation levels for acquirers are also influenced by 52-week highs, the relation between current acquirer stock prices and their 52-week highs may influence mergers.
M
The adoption of the 52-week high price as a reference point has deep roots in the psychology literature, particularly the anchoring mechanism of Tversky and Kahneman (1974),
ED
who show that subjects are influenced by easily available reference points in their estimates of an unknown quantity. In real estate markets, for example, listing prices influence perceived values
PT
for both amateurs and real estate professionals (Northcraft and Neale, 1987), and past purchase
CE
price affects seller behavior (Genesove and Mayer, 2001). In legal disputes, damage awards are influenced by what is asked for in court (Hastie et al., 1999). Managers of initial public offering
AC
(IPO) firms measure their wealth losses and gains against some reference point (Loughran and Ritter, 2002). Firms manage earnings to exceed certain thresholds (DeGeorge et al., 1999). The number of options granted to CEOs tends to be rigid (Shue and Townsend, 2017). CEOs are more likely to issue equity due to stock performance since their arrival rather than inherited stock performance (Baker and Xuan, 2016). Mutual fund managers are more willing to sell past losers
Page 9
ACCEPTED MANUSCRIPT
if the losers are inherited (Jin and Scherbina, 2011). In a behavioral signaling model, Baker et al. (2016) argue that investors evaluate current dividends against some reference point established by the firm’s dividend history. The literature on human decision making identifies a variety of psychological
CR IP T
mechanisms associated with anchoring. Tversky and Kahneman (1974) propose an anchoringand-adjustment process whereby individuals rely on salient reference points as starting points in numerical estimations and adjust away from the reference point until the estimate falls within the range of plausible estimates (Strack and Mussweiler, 1997). Consequently, adjustments tend to
AN US
be insufficient and eventual estimates are biased toward the initial reference point. Other mechanisms include selective accessibility (Strack and Mussweiler, 1997), confirmatory hypothesis testing (Chapman and Johnson, 1994), and numeric priming (Oppenheimer et al.,
M
2008).
Regardless of the specific psychological mechanism involved, we consider the possibility
ED
that investors tend to anchor on the 52-week high in the evaluation of firms making acquisitions and that the 52-week high influences investors’ valuation-level perceptions. Absent any
PT
behavioral biases, we would expect the process of estimating the acquiring firm’s new stock
CE
price to start with the firm’s pre-acquisition price and then adjust away from that price to reflect the synergistic gains associated with the merger, the premium paid to the selling company, the
AC
method of payment, and other value-relevant information associated with the acquisition. Just as in the process of making a merger bid to a publicly traded firm (Baker et al., 2012) or assessing the impact of earnings announcements (George et al., 2014), however, investors face varying degrees of uncertainty and indeterminacy when valuing firms that are making acquisitions. When investors face a high degree of uncertainty, they may rely on heuristics to arrive at valuation-
Page 10
ACCEPTED MANUSCRIPT
level estimates. Thus, reference prices may play an important role in explaining the stock market’s response to an acquisition announcement. Specifically, if an acquirer’s 52-week high price influences some investors’ perceptions of valuation levels, these investors’ price adjustment processes at the announcement may be
CR IP T
influenced by the distance between the acquirer’s current stock price and its 52-week high. For acquirers with stock prices well below their 52-week highs, investors may perceive them as having a relatively low valuation. For acquirers with stock prices at or near their 52-week highs, on the other hand, investors may have a high perceived valuation of their stocks. Ultimately,
AN US
these arguments imply that acquirer announcement-period abnormal returns will be higher for acquirers with pre-announcement prices well below their 52-week highs than for acquirers with pre-announcement prices near their 52-week highs.
M
A further prediction of the perceived valuation hypothesis is that the influence of acquirer 52-week high on investor behavior should be more pronounced in acquisitions involving greater
ED
uncertainty and among acquisitions in which market participants have less information about the transaction. This prediction is based on evidence that reference points are more likely to
PT
influence numeric estimates when there is greater uncertainty (Jacowitz and Kahneman, 1995)
CE
and/or when subjects have less knowledge (Mussweiler and Strack, 2000). Specifically, we expect to observe a stronger acquirer reference price effect in acquisitions involving private
AC
targets, uncertainty about acquirer valuation, relatively large acquisitions, and non-cash forms of payment. Further, a substantial literature finds that individual investors are prone to behavioral biases (e.g., Odean, 1998; Barber and Odean, 2008) and that institutional investors are more sophisticated (e.g., Nofsinger and Sias, 1999; Bartov et al., 2000). Therefore, we expect a
Page 11
ACCEPTED MANUSCRIPT
stronger reference price effect in acquirers with higher individual (lower institutional) investor ownership. In addition to influencing acquirer abnormal returns, the perceived valuation hypothesis has implications for other deal characteristics. Conventional valuation levels have been shown to
CR IP T
influence merger decisions, the market response to those decisions, and their ultimate outcomes. For example, Dong et al. (2006) find that higher bidder valuation levels based on price-to-book value of equity and price-to-residual income value of equity are associated with a greater tendency to use stock as the means of payment and higher bid premia. Just as conventional
AN US
valuation levels may influence mergers, perceived valuation levels based on 52-week high prices may also influence mergers, especially under circumstances in which merger outcomes are being driven by decision agents who are faced with limited information and uncertainty. Following the
M
arguments in Shleifer and Vishny (2003) and Dong et al. (2006), the perceived valuation hypothesis implies that bidders with higher reference price ratios are associated with higher bid
ED
premia, higher target announcement returns, a greater (lower) likelihood of using stock (cash) as the sole consideration, a lower likelihood of being involved in contentious acquisitions, and a
PT
greater likelihood of successfully completing the deal.
CE
3. Sample and data
Our sample of acquisitions is drawn from the Securities Data Corporation (SDC). The
AC
deals are announced during the 1981–2014 period; the acquirers are publicly traded U.S. firms and have stock price data from CRSP and accounting data from Compustat; we exclude deals with transaction values lower than $1 million (as reported in SDC) or deals with relative sizes lower than 5% or higher than 200% of the acquiring firms’ equity; and deals are also excluded if
Page 12
ACCEPTED MANUSCRIPT
the acquirer owns more than 50% of the target prior to the announcement. We further require that the data necessary to conduct the main analyses are available. The final sample includes a total of 19,119 transactions. Panel A of Table 1 presents basic information about the sample. The average transaction value is $453.31 million; out of
CR IP T
these, 75% involve private targets (private firms or subsidiaries of other companies); 19% of the transactions are paid with stock, 23% by cash, and 58% with mixed instruments (cash, stock, debt, convertibles, earn-outs, etc.). Only 4% of the deals are tender offers, 2% are recorded as hostile, and 88% of deals are successful. The average acquirer reference price ratio is 0.81. As
AN US
defined below, this indicates that, on average, the acquirers’ pre-announcement stock price is 81% of its 52-week high.
Panel A of Table 1 also lists the year-by-year averages for these variables. Most of these
M
numbers vary over time, and there is a noticeable increase in the number of deals and tendency to make acquisitions with stock in the mid to late 1990s. This pattern is consistent with the merger
ED
wave over this period. There is, however, no apparent trend in the average reference price ratio. [Insert Table 1 here]
PT
The primary focus of our analysis is the impact of the acquirer’s reference price ratio
CE
(RPR, hereafter) on the stock market reaction to a merger announcement and deal characteristics. We define the RPR following George and Hwang (2004). Specifically, the RPR on day t is the
AC
ratio of the closing price Pt to the highest closing price over the previous 252 trading days (including day t), with prices adjusted by stock splits and dividends. RPR is between zero and one. A company with a stock price near its 52-week high will have an RPR close to one. In our main analysis, we choose the RPR as of the sixth day prior to the announcement, t – 6. The chart below illustrates the timing of measuring the RPR and the event window, where day t = 0 is the
Page 13
ACCEPTED MANUSCRIPT
announcement date. As shown in Panel B of Table 1, our sample has an average RPR of 0.81, a median of 0.86, a range of 0.01 to 1.0, and an interquartile range of 0.72 to 0.95. Event window [-5, +1]
CR IP T
Reference Price Ratio (RPR) measured
…
-6 -5 -4 -3 -2 -1 0 +1
…
We measure the acquirer’s cumulative abnormal return (CAR, hereafter) over the sevenday event window [-5, +1], where day 0 is the announcement date. The abnormal return is based
AN US
on a market model estimated over the [-370, -253] days relative to the announcement date. This approach follows Harford (1999). As shown in the robustness section, our main results are robust to alternative measures of abnormal returns.
Panel B in Table 1 shows the summary statistics of the main variables for our sample.
M
Our sample has an average acquirer CAR of 1.50%, with a median of 0.57%, ranging from -
ED
22.93% to 36.57%, and an interquartile range of -3.46% to 5.37%. The average numbers are similar to those reported in the literature based on similar samples of both publicly traded and
PT
private targets (e.g., Moeller et al., 2004). All variables are defined in Appendix A. 4. Acquirer announcement-period and long-run abnormal returns
CE
4.1.Acquirer cumulative abnormal returns and acquirer reference price ratio
AC
As an initial step in examining the potential relation between the acquirer RPR and acquirer CAR, we first separate our sample into acquisitions of publicly traded targets and private (private or subsidiary) targets and then sort each listing-status sample into two equal groups (by year) based on RPR. We categorize acquisitions by the target listing status for two reasons. First, the literature documents different acquirer CARs, and varying influences of other variables on acquirer CARs, in acquisitions of public versus private targets (e.g., Fuller et al., Page 14
ACCEPTED MANUSCRIPT
2002). In addition, to the extent the information environment associated with a public acquisition is better, we expect the tendency for investors to be influenced by reference prices to be lower than in private acquisitions, a natural setting for testing the perceived valuation hypothesis.5 The Low (High) RPR group includes acquisitions with acquirers’ RPR below (above) the
CR IP T
sample median within its listing-status group. Table 2 shows basic summary statistics of the two groups. Panel A of Table 2 shows that, in the whole sample, the Low RPR group has an average RPR of 0.677 and the High RPR group has an average RPR of 0.937, implying a difference of
of private targets (0.829 versus 0.800).
AN US
0.260. The RPR ratios for acquirers of public targets tend to be slightly higher than for acquirers
[Insert Table 2 here]
The first column in Panel B of Table 2 shows that the Low RPR group has an average
M
CAR of 2.021%. The High RPR group has an average CAR of 0.982%, implying a significant difference of -1.040% (t=-7.66).6 Combining the two differences in means between the Low RPR
ED
and High RPR groups from Panels A and B suggests that acquirer CARs increase by 4.00% (1.040%/0.260) if the RPR decreases from one to zero. Put differently, acquirers with pre-
PT
announcement prices at half of their 52-week highs earn about 2.0% higher abnormal returns, on
CE
average, than those with pre-announcement prices at their 52-week highs. For convenience, we refer to this negative relation between acquirer RPR and acquirer CAR as the reference price
AC
effect.7
5 Consistent with this view, Officer (2007) finds that private targets (private independent firms or subsidiaries of other firms) tend to receive lower premia, probably due to lower liquidity of such assets. 6 Unreported for brevity, the medians have a similar pattern and the Wilcoxon test on the distributions of the two groups yields a test statistic of -5.26, which is significant at the 1% level. 7 We obtain similar results based on reference price ratios of the 39-, 26-, and 13-week highs. This finding is consistent with Baker et al. (2012), who find that target offer prices cluster on multiple past peak prices. For simplicity, our main analysis focuses on the 52-week high as the reference price.
Page 15
ACCEPTED MANUSCRIPT
Fig.1 illustrates the acquirer cumulative average abnormal returns of the Low and High RPR groups from five days before to 20 days after the announcement date for acquisitions of private targets (Panel A) and acquisitions of public targets (Panel B). The end of day t – 6 is the starting date. For private targets, in addition to the salient gap between the two lines over the
CR IP T
event window of [-5, +1], the figure shows that the reference price effect persists for at least 20 days after the announcement. In the public subsample, there is a noticeable, but smaller gap between the two lines and the lines cross within ten days after the announcement.8 [Insert Fig. 1 here]
AN US
Panel B of Table 2 shows a substantially stronger reference price effect in the private subsample compared to the public subsample. Specifically, within the private sample (the 4th column), the high RPR acquirers earn an average CAR of 1.726% and low RPR acquirers earn
M
an average CAR of 2.916%, a significant difference of -1.190% (t=-7.48). In the public subsample (the 7th column), high RPR acquirers earn an average CAR of -1.201% and low RPR
ED
acquirers earn an average CAR of -0.618%, a smaller difference of -0.582% with weaker statistical significance (t=-2.36). This weaker reference price effect is consistent with the
PT
perceived valuation hypothesis as the information environment for deals involving public targets
CE
is much better than private targets and, as such, investors are influenced less by reference prices. Given the possibility that the reference price effect is driven by conventional valuation
AC
levels, we perform additional two-way sorts on market-to-book (M/B) ratios to determine whether the reference price effect exists within valuation-level categories. The second and third
8
Over the longer event window of [-5, +20] in the private subsample, the average acquirer CAR is 2.403% for the Low RPR group and 1.633% for the High RPR group, implying a difference of -0.770% (t=-2.37); in the public subsample, the average CAR is -2.149% for the Low RPR group and -1.756% for the High RPR group, implying a difference of 0.393% (t=0.78). Page 16
ACCEPTED MANUSCRIPT
columns in Panel B of Table 2 show these two-way sort results.9 They indicate that low RPR acquirers tend to earn higher CARs than the high RPR acquirers regardless of M/B level. For example, within the High M/B acquirers, those with a low RPR earn an average CAR of 1.787% while those with a high RPR earn an average CAR of 0.945%, a significant difference of -
CR IP T
0.842% (t=-4.18). The pattern is similar within the low M/B acquirer group. The 5th and 6th columns show a similar pattern in the private subsample. In the public subsample, however, the differences between the low and high RPR groups are substantially smaller. For example, within the High M/B group, the last column in Panel B of Table 2, low RPR acquirers earn a higher
AN US
CAR than high RPR acquirers (-0.932% vs. -1.600%), but the difference is only marginally significant (t=-1.78).
4.2.Multiple regressions of acquirer cumulative abnormal returns
M
Table 3 presents four ordinary least squares (OLS) regressions examining the impact of acquirer RPR on acquirer CAR. All regressions control for Fama and French (1997) industry and
ED
year fixed effects. To alleviate the impact of outliers we winsorize all continuous variables at the 1st and 99th percentiles. The t-statistics are based on standard errors clustered by firms. In the first
PT
regression, we regress acquirer CAR on acquirer RPR and a list of variables designed to control
CE
for various deal and firm characteristics. We include the most frequently controlled variables among the more recent studies in the M&A literature for which we have data (e.g., Fuller et al.,
AC
2002; Moeller et al., 2004, 2005; Masulis et al., 2007; Officer et al., 2009; Golubov et al., 2012; Harford et al., 2012; El-Khatib et al., 2015; Golubov et al., 2015). In the first regression, the coefficient on acquirer RPR is -5.487 (t=-8.56), indicating a
strong reference price effect. That is, acquirer CARs are higher (lower) when the acquirer’s pre-
9 Our proxy for valuation levels is market-to-book ratios for the univariate analysis shown in Panel B of Table 2. We consider other proxies in subsequent analysis.
Page 17
ACCEPTED MANUSCRIPT
announcement stock price is well below (near) its 52-week high even after controlling for the comprehensive list of firm and deal characteristics. It is worth noting that the reference price effect is economically significant. For a one standard deviation increase in the RPR (0.1844), the acquirer CAR decreases by 1.012% (=0.1844*5.487), or about 11% (=1.012/9.397) of its
CR IP T
standard deviation. By comparison, a one standard deviation change in firm size induces an 8.5% standard deviation change in acquirer CAR.
The coefficients on the control variables are largely consistent with the literature (Chang, 1998; Fuller et al., 2002; Faccio et al., 2006; Officer et al., 2009; Golubov et al., 2015). As
AN US
expected, private acquisitions earn higher acquirer returns. Notably, with both private and public deals included in the sample, the coefficient on stock as a method of payment is negative. However, using stock tends to earn higher acquirer returns in private transactions, as indicated by
M
the positive and significant coefficient on the interaction term that equals one for stock transactions involving private deals (e.g., Fuller et al., 2002; Slovin et al., 2005). Relative size
ED
has a positive and significant coefficient. M/B has a significant negative coefficient, consistent with the valuation hypothesis that acquirers with higher valuations earn lower returns at the
PT
M&A announcements. 10 Consistent with Moeller et al. (2004), larger acquirers earn lower
CE
acquisition returns. Acquirer returns are higher if an acquisition is preceded by a long dormant period (a year), consistent with Cai et al. (2011). The coefficient on tender offers is positive and
AC
significant, that on hostile deals is negative and significant, and that on toeholds is not
10 The variable M/B used in the regression analysis is measured at the fiscal year-end before the acquisition announcement. However, we find similar results when we measure M/B using stock price on day t-6 and book value equity reported in the most recent quarter prior to the announcement. Further, in unreported analysis, we use other proxies for conventional valuation levels, including a market-to-intrinsic value ratio, where intrinsic value is estimated using a residual income model following Edwards and Bell (1961) and Ohlson (1995), and a measure developed by Rhodes-Kropf et al. (2005) with the Hertzel and Li (2010) adjustment. In all cases, we find a reference price effect similar to what is reported in Table 3.
Page 18
ACCEPTED MANUSCRIPT
significant. Past return has a positive coefficient, consistent with the view that bidders with higher management quality earn greater acquisition returns. [Insert Table 3 here] The second and third models in Table 3 show the results of estimating the regression
CR IP T
model for the private and public subsamples separately. Consistent with reference prices being more important under circumstances involving greater uncertainty or a more opaque information environment, the reference price effect is more pronounced in the private subsample. The coefficient on RPR is -6.286 (t=-8.66) in the private subsample regression whereas it is -2.581
AN US
(t=-1.97) in the public subsample. The coefficients on the control variables are largely consistent with the results for the whole sample except as follows. Notably, the coefficient on stock as a method of payment is positive and statistically significant in the regression of private
M
acquisitions but negative (although not statistically significant) in the regression of public transactions. Cash deals involving public targets earn significantly higher acquirer returns.
ED
Relative size has a positive and significant coefficient in private acquisitions but the coefficient is negative and not statistically significant in public acquisitions. These results are consistent
PT
with the notion that investors perceive acquirers of private targets as getting a favorable deal and
CE
the benefit of the deal being amplified with the relative size of the target and by paying stock (Fuller et al., 2002). M/B has a negative coefficient, but it is not statistically significant in the
AC
subsamples.
In the fourth model of Table 3, we include the M/B and reference price ratio of the target
firm (Target Ln(M/B) and Target RPR, respectively) in the analysis. Due to the additional data requirement, this analysis is limited to the subsample of acquisitions involving public targets that are covered in CRSP. We find the coefficient on the acquirer’s RPR remains negative, but it is
Page 19
ACCEPTED MANUSCRIPT
only marginally significant (t=-1.92). The weaker acquirer reference price effect is consistent with the perceived valuation hypothesis as the information environment for these transactions is better and investors are thus less affected by reference prices. The coefficient on target M/B is negative and marginally significant (t=-1.70), consistent with Dong et al. (2006). The coefficient
CR IP T
on the target’s RPR is positive and significant (t=4.28). According to Dong et al. (2006), target valuation ratios can influence acquirer CARs via two potential mechanisms. On the one hand, to the extent targets are perceived as misvalued and an acquisition announcement triggers a reassessment of valuation levels, we expect a negative relation between target valuation levels
AN US
(either target RPR or target M/B) and acquirer CAR. On the other hand, if a higher target RPR is associated with lower bid premia (see Table 6 in Section 5.1), as indicated by Baker et al. (2012), and lower bid premia are associated with higher acquirer CARs, we expect a positive relation
M
between target RPR and acquirer CAR. The positive and significant coefficient on target RPR in the fourth model suggests that any negative influence from the reassessment of valuation levels
ED
is more than offset by the positive influence due to lower bid premia. To sum up, the first model in Table 3 establishes a strong and economically meaningful
PT
reference price effect in a multiple regression setting. By separating the sample into transactions
CE
involving public and private targets, the second, third, and fourth models in Table 3 provide evidence the reference price effect is stronger among private deals, consistent with reference
AC
prices having a greater influence when there is less information available. 11
11
Similar results are obtained when we exclude the potentially endogenous variables, stock, cash, tender offer, and hostility. Also, it should be noted that the results shown in Table 3 include both successful and unsuccessful deals. When the analysis is restricted to successful deals (those that are ultimately completed, according to SDC; n=16,806), the coefficient on RPR in the first model is -4.461 with a t-stat of -6.79. When the analysis is limited to unsuccessful deals (n=2,313), the coefficient on RPR is -12.891 with a t-stat of -6.03. This suggests that, although the reference price effect is stronger in deals that are unsuccessful, it is also strong in successful deals. Because the ultimate outcome of a given deal is not known at the time of announcement, however, we conduct the bulk of our analysis using all announced acquisitions. The results are similar regardless of whether we include unsuccessful deals. Page 20
ACCEPTED MANUSCRIPT
We also examine several alternative reference price ratios as in Baker et al. (2012). Specifically, we construct new reference price ratios relative to the 75th, 90th, and 95th percentiles, respectively, of acquirer prices over the 52-weeks preceding the sixth day before the announcement date. For convenience, we call them RPR75, RPR90, and RPR95, respectively.
CR IP T
They are naturally highly correlated with the original acquirer RPR, with correlation coefficients of 0.676, 0.881, and 0.956, respectively. When these alternative RPRs piecewise enter the regression of the full sample without the RPR, they all have significant negative coefficients. When these alternative RPRs piecewise enter the regression with the original RPR, however,
AN US
their coefficients become positive with weak statistical significance while the coefficients on the original RPR remain negative and significant.12
These falsification tests highlight the importance of the 52-week high, not some highly
M
correlated but less salient numbers that affect investors’ valuation perceptions. When the sample is separated into public and private acquisitions, the falsification tests for the private acquisitions
ED
yield very similar conclusions as the whole sample. In the subsample of public acquisitions and the subsample of public acquisitions covered in CRSP, however, the coefficients on RPR75,
PT
RPR90, and RPR95 are negative but not statistically significant when they piecewise enter the
CE
regression without RPR. When RPR is present, it is significant only in the regression with RPR75, consistent with reference prices playing a weaker role in public acquisitions. To the
AC
extent acquirer reference prices do play a role in these acquisitions, it seems to be the 52-week high.
Further, RPRs based on 39-, 26-, and 13-week highs yield similar results to those in the
regression that includes all acquisitions in Table 3. They indicate that, in addition to the 12
We also construct RPR99, the ratio of stock price to the 99 th percentiles of the past 252 daily stock prices. Due to the extremely high correlation (0.993) between RPR99 and the original RPR, the regression with them both present has serious multicollinearity problems. Neither coefficient is significant. Page 21
ACCEPTED MANUSCRIPT
commonly studied 52-week highs, some investors may also anchor on peak price levels over the past 39, 26, or 13 weeks (see also Baker et al., 2012). Regardless of the exact peak prices used as the anchor, the results support the same economic implication: Investor reaction to acquisition announcements seems influenced by recently observed peak prices as reference prices. These
CR IP T
unreported results are available upon request. 4.3.The role of uncertainty
The perceived valuation hypothesis predicts that the acquirer reference price effect is stronger among deals involving greater uncertainty, more opaque information environments, and
AN US
less sophisticated investors. Our analysis to this point finds support for this prediction by documenting a stronger reference price effect in acquisitions of private versus public targets. To further test these predictions, we examine the extent to which the reference price effect is
M
influenced by additional proxies for information uncertainty and individual investor ownership levels.
ED
We consider four proxies for information uncertainty: The volatility of acquirer stock returns, the number of analysts following the acquirer, the relative size of the acquisition, and an
PT
indicator variable for non-cash methods of payment. Stocks in general, including stocks of
CE
acquirers, are more difficult to value if their prices are more volatile. To the extent stock analysts produce valuation-relevant information, acquirers with fewer analysts following them have a
AC
more opaque information environment and are more difficult to value. Similarly, if the transaction value relative to the acquirer equity value is larger, the acquisition will have a greater impact on the acquirer’s stock value, posing greater uncertainty to the valuation process. Last, but not least, non-cash payment adds a level of complexity not present in cash deals and could
Page 22
ACCEPTED MANUSCRIPT
indicate the acquiring firm’s uncertainty regarding the value of the acquired assets (Hansen, 1987; Officer et al., 2009). The perceived valuation hypothesis predicts a stronger reference price effect for acquirers with more volatile stocks, for acquirers followed by fewer analysts, for deals of greater relative
CR IP T
size, and for deals that involve non-cash payment. We test these implications in multiple regressions by adding interaction terms. Further, because we document a substantial difference in reference price effects for acquisitions of private firms versus acquisitions of public firms, we analyze the influence of these proxies for information uncertainty separately for the two types of
AN US
acquisitions. Table 4 presents these regression results.13
In Models 1 and 6 of Table 4, we add both Sigma and RPR*Sigma to the baseline model, where Sigma measures volatility as in Zhang (2006). The full regression specifications follow
M
those of Table 3. For brevity, coefficients and t-statistics of some variables are not shown. For ease of interpretation we subtract the sample mean from the variable Sigma. The perceived
ED
valuation hypothesis implies a negative coefficient for RPR*Sigma. In private acquisitions, Model 1 of Table 4 reports the coefficient on RPR is -4.697 (t=-5.73), indicating a significant
PT
reference price effect for acquirers with average Sigma values and the coefficient on RPR*Sigma
CE
is -0.455 (t=-2.90), confirming the prediction. In public acquisitions (Model 6), the coefficients on RPR and RPR*Sigma are -2.927 (t=-1.97) and -0.534 (t=-1.94), respectively. These
AC
coefficients are also consistent with the perceived valuation hypothesis but notably weaker than in the sample of private acquisitions. Sigma itself has a positive and significant coefficient in the sample of private acquisitions but is not significant in the sample of public acquisitions.
13
For the subsample of public deals involving targets that are covered in CRSP, we also add the target valuation proxies (target RPR and target M/B) in these regressions and find very similar results to those for the “Public” subsample, although the statistical significance is expectedly weaker. For brevity, these results are not reported but available upon request. Page 23
ACCEPTED MANUSCRIPT
In Models 2 and 7 of Table 4, we add an interaction term, RPR*Few analysts, where Few analysts is a binary variable equal to one if the number of analysts following the acquirer over the six months prior to the announcement month is below the sample median. The perceived valuation hypothesis predicts a negative coefficient on this interaction term: Acquirers followed
CR IP T
by fewer analysts will have a stronger (i.e., more negative) reference price effect. In Model 2 of Table 4 the coefficient on RPR is -3.480 (t=-3.75) for private acquisitions, indicating a strong reference price effect for acquirers with an above-median number of analysts. That is, even for acquirers followed by many (above-median) analysts, there is a strong reference price effect in
AN US
acquisitions of private targets. For acquirers followed by few (below-median) analysts, the reference price effect is even stronger as the coefficient on RPR*Few analysts is a significant 4.822 (t=-4.04). We also report a positive coefficient for the binary variable Few analysts in the
M
sample of acquisitions involving private targets. In public acquisitions (Model 7), the coefficient on RPR is not statistically significant, suggesting there is little evidence of a reference price
ED
effect when a firm followed by many analysts acquires a publicly traded target, a circumstance in which investors are least likely influenced by reference prices when assessing acquirer stock
PT
price. The coefficient on RPR*Few analysts is -3.815 (t=-1.87), indicating a stronger reference
CE
price effect in acquisitions of public targets when the number of analysts following the acquirer is below the median.
AC
We add the interaction term between RPR and relative size to Models 3 and 8. The perceived valuation hypothesis predicts a negative coefficient on this interaction term. In private acquisitions, the coefficient on the interaction term is -6.623 (t=-3.42), further supporting the perceived valuation hypothesis. Similar to Model 1, the relative size variable is demeaned. Thus, the coefficient on RPR itself, -6.290 (t=-8.66) represents the reference price effect in private
Page 24
ACCEPTED MANUSCRIPT
acquisitions with an average relative size. In public acquisitions (Model 8), the coefficient on RPR is not significant, but the coefficient on the RPR*Rel. size variable is -5.461 (t=-2.10), indicating that, although there is no significant reference price effect in public acquisitions of an average relative size, there is an interaction effect that is consistent with the perceived valuation
CR IP T
hypothesis. Models 4 and 9 of Table 4 include an interaction term RPR*Non-cash. The perceived valuation hypothesis implies a negative coefficient on this term. Reported in Model 4 of Table 4, the coefficient on RPR*Non-cash is -3.364 (t=-2.40) in private acquisitions, confirming the prediction. The significant negative coefficient on RPR in Model 4 (-3.564 with t=-2.80) for
AN US
private acquisitions suggests there is a strong reference price effect even among cash-only transactions. In public acquisitions, the coefficient on RPR is -3.684 (t=-1.83), indicating an overall reference price effect, but the lack of a statistically significant coefficient on RPR*Non-
M
cash suggests that the reference price effect is unrelated to the method of payment in acquisitions of public targets.
ED
[Insert Table 4 here]
4.4.The role of investor sophistication
PT
To further explore the implications of the perceived valuation hypothesis, we examine
CE
whether less sophisticated investors are more likely to be influenced by the acquirer’s 52-week high price when estimating the valuation implications of an acquisition. Generally, empirical
AC
evidence suggests that individual investors are prone to behavioral biases (e.g., Odean, 1998; Barber and Odean, 2008) and that institutional investors tend to be more sophisticated (e.g., Nofsinger and Sias, 1999; Bartov et al., 2000). Therefore, we expect a stronger reference price effect when individual investor ownership of the acquirer is relatively high. Following other studies, we treat non-institutional investor ownership as a proxy for individual investor
Page 25
ACCEPTED MANUSCRIPT
ownership (e.g., Nofsinger and Sias, 1999; Gompers and Metrick, 2001; Sias and Whidbee, 2010). Institutional ownership data are from Thomson Reuters Institutional Holdings (13F). Total institutional ownership of the acquirer is measured at the most recent quarter-end prior to
CR IP T
the announcement month. The variable “Indiv. own.” is one minus the institutional ownership. If an acquirer’s institutional ownership data are missing it is assigned as zero and “Indiv. own.” is equal to one. Admittedly, this measure of investor sophistication is potentially problematic and is not necessarily indicative of investors who trade in an acquirer’s stock in the days surrounding
AN US
announcement dates. However, we suggest that the investing environment surrounding acquirer stocks with relatively high individual (low institutional) ownership is more exposed to potential behavioral biases.
M
The results of the regressions incorporating individual ownership levels are shown in Models 5 and 10 of Table 4. We demean the Indiv. own. variable and add both Indiv. own. and
ED
RPR*Indiv. own. to the baseline model. The perceived valuation hypothesis implies a negative coefficient for RPR*Indiv. own. In private acquisitions (Model 5), the coefficient on RPR is
PT
negative and significant, suggesting a strong reference price effect for acquirers with average
CE
individual investor ownership; the coefficient on the interaction term is -0.112 (t=-5.40), indicating that the reference price effect is stronger when individual investors own a greater
AC
percentage of acquiring firms’ stock. The coefficient on Indiv. own. is positive and significant. In public acquisitions (Model 10), the coefficients on RPR and the interaction term are -2.187 (t=1.69) and -0.103 (t=-2.82), respectively, suggesting a marginally significant overall reference price effect and that the effect increases with acquirer individual ownership.
Page 26
ACCEPTED MANUSCRIPT
Overall, the evidence in Table 4 is consistent with the perceived valuation hypothesis. The reference price effect is stronger in acquisitions involving opaque information environments, when there is greater uncertainty about acquirer valuation, and when less sophisticated investors hold a higher percentage of acquirer stocks.
CR IP T
4.5.Long-run acquirer returns
To further distinguish between the perceived valuation hypothesis and other potential explanations for the reference price effect, we examine longer horizon returns. If the reference price effect is being driven by perceived valuation levels based on 52-week high prices, an
AN US
irrational behavioral bias, we expect the short-horizon return pattern documented above to be reversed over the long term as the bias is eventually corrected, and we should observe a positive relation between long-horizon acquirer returns and acquirer RPR. On the other hand, if the
M
reference price effect is capturing the influence of some unidentified fundamental factor associated with acquisitions, then the short-horizon return pattern should be permanent and there
ED
should be no relation between RPR and long-horizon returns. Finally, if the RPR is acting as another proxy for conventional valuation levels, which are associated with poor long-horizon
PT
acquirer returns (Rau and Vermaelen, 1998; Fu et al., 2013), we should observe a negative
CE
relation between long-horizon returns and the RPR. Measuring long-horizon returns is inflicted with methodological issues and no consensus
AC
optimal method is proffered in the literature (e.g., Lyon et al., 1999; Mitchell and Stafford, 2000). Further, tests of price efficiency inevitably suffer from a bad-model problem (Fama, 1998). We first follow the two approaches commonly adopted in the literature: buy-and-hold abnormal returns (BHAR), and calendar-time portfolio returns (CTPR). To avoid the bad-model problem, we also examine subsequent earnings announcement abnormal returns. While none of
Page 27
ACCEPTED MANUSCRIPT
the approaches is perfect, the collective evidence provides a more comprehensive view of longrun returns. As in Table 2, we form two equal subsamples (by year) on pre-announcement acquirer RPR and examine their average BHAR and CTPR. The holding period for both methodologies is
CR IP T
one year following the month of announcement. To alleviate the influence of outliers we winsorize the BHARs at the 1st and 99th percentiles, although results are similar without winsorization. Panel A of Table 5 presents the BHAR results for the whole sample and the Low and High RPR subsamples. The first column shows that the acquirers in our sample earn -
AN US
3.625% over the one-year period following the announcement month.14 The magnitude is in line with existing studies (e.g., Bouwman et al., 2009; Ma et al., 2011; Duchin and Schmidt, 2013). More striking are the different BHARs of the Low and High RPR groups. As defined in
M
Table 2, Low RPR (High RPR) represents lower (higher) RPR acquirers. The mean one-year BHAR is -5.889% for Low RPR and -1.369% for high RPR acquirers, resulting in a difference of
ED
4.520%. The return difference in the BHAR clearly exhibits a reversal of the short-term return pattern shown in Table 2. Recall that Low RPR acquirers earn significantly higher
PT
announcement-period abnormal returns, but the long-horizon results shown in Table 5 indicate
CE
that the Low RPR acquirers earn significantly lower long-horizon abnormal returns than the High RPR acquirers.
AC
Panel B of Table 5 shows the Fama and French (2015) five-factor alphas of the calendartime portfolios for the whole sample of acquirers and for the Low and High RPR subsamples. 14 The t-statistics used here to determine statistical significance assume independence across acquisitions deals. In reality, however, these deals are clustered and thus BHARs might be cross-correlated. To address the issue of sample clustering and cross-correlation, the literature (e.g., Lyon et al., 1999; Mitchell and Stafford, 2000) suggests the use of bootstrapping to derive the empirical distribution. This approach is adopted in Bouwman et al. (2009). We follow their approach and find that the t-stat for the whole sample BHAR is -9.15 (the empirical distribution indicates statistical significance at the 1% level) for the one-year BHAR. Regardless of how the statistical significance is measured, the pattern of reversal in BHAR is clear.
Page 28
ACCEPTED MANUSCRIPT
The measurement period is one year following the month of announcement. As the numbers in Panel B are monthly alphas and those in Panel A are BHAR, the magnitude of the numbers in Panel B is on a smaller scale. For the whole sample, the monthly alpha is -0.057%. The alphas are -0.165% for the Low RPR group and 0.017% for High RPR, with a difference of 0.181%
CR IP T
(t=1.29). The second (third) columns of Panels A and B present similar evidence for the subsample of acquisitions involving only private (public) targets. [Insert Table 5 here]
To further shed light on the long-term stock price behavior following M&A events, we
AN US
examine the abnormal returns surrounding the earnings announcements over the subsequent oneyear period. This approach is motivated by the anomalies literature. For example, Chopra et al. (1992) find that long-term poorly performing stocks exhibit an overreaction pattern in the long-
M
term subsequent time period as well as in short-term abnormal returns around quarterly earnings announcements, suggesting that part of the long-term return pattern is reflected in the short-term
ED
earnings announcement periods. Jegadeesh and Titman (1993) use a similar approach to study the momentum anomaly; La Porta et al. (1997) use this approach to study the value premium;
PT
and Titman et al. (2004) use it to study the capital investment anomaly. Engelberg et al. (2017)
CE
report that, for 97 return anomalies, anomaly returns are seven times higher on quarterly earnings announcement days than days of no news, suggesting that earnings announcements preserve the
AC
return pattern over the long term. The central premise applied to our context is that the mispricing following M&A events is corrected partly in the market response to subsequent earnings announcements. If, according to the perceived valuation hypothesis, high RPR acquirers earn higher returns than low RPR acquirers over the long horizon, we expect to observe the same pattern in the abnormal returns surrounding the earnings announcements over the
Page 29
ACCEPTED MANUSCRIPT
contemporaneous period. Because earnings announcement-period abnormal returns are measured over a very short time (three-day event window in our analysis), the bad-model problem is less serious (Fama, 1998). Specifically, for each acquisition we record the four earnings announcement dates (from
CR IP T
Compustat quarterly file) over the one-year period following the M&A announcement. For each of the earnings announcements, we measure the abnormal returns over the [-1, +1] window, adjusted by the CRSP equal-weighted market returns (results are similar when adjusted by valueweighted market returns). We examine the four-quarter averages. Panel C presents the results.
AN US
For the whole sample, the average earnings announcement-period return is negative (positive) for the low (high) RPR group, leading to a significant difference of 0.204 (t=3.16). The private and public subsamples exhibit a similar pattern.
M
For comparison, Panels A, B, and C also show results for subsamples formed on two conventional valuation proxies, M/B and RKRV. RKRV is based on a measure developed by
ED
Rhodes-Kropf et al. (2005) with the Hertzel and Li (2010) adjustment. The RKRV approach generates three measures of valuation depending on their regression specifications. We show
PT
results based on their most sophisticated model, but we find similar results using their other two
CE
specifications.15 Appendix B contains the detailed definition. Across three samples (the whole sample, and the private and public subsamples), two conventional valuation proxies (M/B and
AC
RKRV), and three long-horizon return measures (BHAR, CTPR, and subsequent earnings announcement returns), Table 5 generates a total of 18 (3x2x3) return differentials between the high and low valuation groups, among which 17 are negative, with 12 significantly so, and the 15
Unreported for brevity, we find six of the nine BHAR, CTPR, and earnings announcement return differentials between the high and low valuation groups are negative when subsamples are formed on sorts using a valuation level proxy based on a market-to-intrinsic value ratio, where intrinsic value is estimated using a residual income model following Edwards and Bell (1961) and Ohlson (1995), and more specifically, Dong et al. (2006) and Ma et al. (2011). However, none of the return differentials is statistically significant. Page 30
ACCEPTED MANUSCRIPT
remaining positive one is not statistically significant. The evidence based on M/B and RKRV in general is consistent with conventional valuation theory in that overvalued acquirers earn lower returns in the long term. To explore the relation between long-horizon acquirer returns and acquirer RPR (along
CR IP T
with conventional valuation-level proxies) further and account for the impact of other deal and firm characteristics, we also estimate OLS regressions of BHARs on acquirer RPR and the list of control variables used in the baseline regressions of acquirer CAR in Table 3. The results are presented in Panel D of Table 5. For the sample with all targets, the first column of Panel D
AN US
shows a positive and significant coefficient on acquirer RPR and a negative, marginally significant coefficient on acquirer M/B. This pattern carries over to the regressions for the private, public, and CRSP subsamples as well, shown in the last three columns of Panel D.
M
Consistent with Panel A, acquirer RPR and our conventional valuation-level measure, M/B, have exactly opposite influences on long-term returns, albeit without a statistically significant
ED
coefficient on M/B in the public and CRSP subsamples. Specifically, in the full sample and private subsample, acquirers with higher M/B earn lower long-term returns than those with lower
PT
M/B, consistent with conventional valuation theory and the market-timing theory of mergers
CE
(Shleifer and Vishny, 2003). At the same time, however, acquirers with higher RPRs earn higher long-term returns than those with lower RPRs in the full sample and all subsamples, consistent
AC
with the perceived valuation hypothesis.16 The sharp contrast between the results based on RPR and those based on M/B and RKRV
suggests that RPR is different from conventional valuation proxies. The evidence supports the 16
Similar results are obtained when we exclude the potentially endogenous variables, stock, cash, tender offer, and hostility. Likewise, the coefficients on RPR remain positive and significant with similar magnitude when we replace M/B with RKRV or our market-to-intrinsic value estimate. Finally, we also find similar results when we measure M/B by using stock price on day t-6 and book value equity reported in the most recent quarter prior to the announcement. Page 31
ACCEPTED MANUSCRIPT
perceived valuation hypothesis that investors irrationally view acquirers’ 52-week highs as a gauge of “fair” stock value. Consequently, their reactions to acquisition announcements are influenced by the 52-week highs in the short-term, but the influence is eventually corrected in the
5. Reference prices and deal characteristics
CR IP T
long-term.
The analysis to this point indicates that the acquirer’s RPR influences acquirer stock prices at the announcement, particularly when there is less information or greater uncertainty about the impact of the acquisition on firm value. In this section, we examine the impact of
AN US
acquirer RPR on deal characteristics, including bid premia and target announcement returns in Section 5.1 and the method of payment and deal combativeness in Section 5.2. When analyzing the influence of acquirer RPR on deal characteristics, we also consider
M
the influence of acquirer M/B as well as the RPR and M/B of the target wherever possible. Accordingly, we interpret our results along with those reported in Dong et al. (2006) and,
ED
wherever relevant, those in Baker et al. (2012). Specifically, the evidence in Dong et al. (2006) indicates that higher acquirer valuation, as measured by M/B, is related to a greater likelihood of
PT
using stock as the method of payment, lower likelihood of using cash, and a lower likelihood of
CE
acquiring through tender offers. Higher target valuation is associated with a greater likelihood of using stock, lower likelihood of using cash, lower likelihood of using tender offers and deal
AC
hostility, and a higher probability of deal success. 5.1. Bid premia and target announcement-period abnormal returns Table 6 presents five OLS regressions of bid premium on acquirer RPR, acquirer M/B,
target RPR, target M/B, and the other control variables used in Table 3 after excluding the potential endogenous variables, stock, cash, tender offer, and hostility. The deals in this analysis
Page 32
ACCEPTED MANUSCRIPT
are limited to those with target information available on CRSP and with available information on bid premia. The first column shows the regression results for all deals with available data. The second and third columns show results for acquisitions sorted on the acquirers’ sigma to allow for the possibility that uncertainty about acquirer valuation may influence the impact of valuation
CR IP T
levels on bid premia.17 Similarly, although the method of payment is potentially endogenous, the fourth and fifth columns present results for the cash and non-cash subsamples to allow for the possibility that acquirer valuation levels may not influence bid premia when cash is the only method of payment.
AN US
The first regression in Table 6 indicates that the coefficient on acquirer RPR is positive but not statistically significant in the whole CRSP sample. In the subsample of deals with high uncertainty about acquirer valuation levels (High sigma) and the subsample of deals involving
M
non-cash forms of payment (Non-cash), however, the coefficients on acquirer RPR are positive and at least marginally significant. This suggests that when the valuation level of the acquirer’s
ED
stock is difficult to determine or the valuation level of the acquirer is important to the bid premium negotiation process, the acquirer’s 52-week high influences bid premium negotiations.
PT
When the valuation level of the acquirer is easier to determine or is not directly relevant to the
AC
CE
negotiation process, however, the acquirer’s RPR is less likely to influence bid premiums: The
17
Bid premia are determined by managers, consultants, and other relatively sophisticated decision agents, and these decision agents are likely to have access to information not available to acquirers’ investors. To the extent these parties are influenced by the bidders’ 52-week high prices, it is likely to depend more directly on the degree of uncertainty they face in estimating and negotiating the ‘fair’ bid premium. The most direct measure of uncertainty for these decision makers is the acquirers’ sigma. Thus, for the purposes of analyzing the impact of the acquirers’ RPR on bid premia, we focus on the acquirers’ sigma as our proxy for the information environment. The other information environment measures (number of analysts following the acquirer, relative size of the acquisition, and individual investor ownership) reflect the information environment faced by acquirers’ investors but not necessarily the information environment faced by the decision agents involved in negotiating an acquisition bid. Not surprisingly, we do not find any significant differential results on acquirer RPR in subsamples formed on these other uncertainty measures. Unreported for brevity, these results are available upon request. Page 33
ACCEPTED MANUSCRIPT
acquirer RPR coefficient is positive but not significant in the Low sigma subsample and is negative and marginally significant in the Cash subsample.18 Notably, target RPR is negatively related to bid premia in the whole sample as well as the subsamples of cash deals and in transactions with low acquirer sigma. This is consistent with the
CR IP T
perceived valuation hypothesis in that targets with pre-announcement prices close to their 52week highs may be perceived as having relatively high valuation levels. Accordingly, they receive a lower premium. This result is also consistent with the argument put forward by Baker et al. (2012). That is, if the decision makers in merger transactions implicitly use targets’ past
AN US
peak prices as reference points, bid premia will be negatively related to target RPRs. When there is more uncertainty about the value of acquirer shares (High sigma) and when the acquirer’s valuation level is directly relevant to bid premium negotiations (Non-cash), however, the
M
coefficients on target RPR are not statistically significant.
Bid premia are higher when acquirer M/B is higher in the whole CRSP sample. Acquirer
ED
M/B is not significant in the Cash subsample, indicating that conventional acquirer valuation levels play no role in explaining bid premia for cash acquisitions, as expected. In non-cash deals,
PT
however, bid premia are positively related to acquirer M/B, consistent with Dong et al. (2006).
CE
Further, acquirer M/B has at least a marginally significant influence on bid premia in both the Low sigma and High sigma samples. Target M/B has a negative coefficient in the whole CRSP
AC
sample and each subsample, but it is not statistically significant in the Low sigma sample. These results add new evidence beyond Baker et al. (2012). A potential interpretation is
that, in negotiating “fair” bid premia, target managers focus on their own firms’ 52-week highs when the value of the payment is known or has relatively low uncertainty associated with it. In 18
One possible interpretation of the negative coefficient on acquirer RPR in the cash subsample is that, to avoid being paid with overvalued stock, target managers are willing to accept a lower premium in exchange for cash payment when acquirers’ perceived valuation levels are high. Page 34
ACCEPTED MANUSCRIPT
acquisitions involving non-cash payment or when there is high uncertainty regarding the value of the acquirer’s shares, on the other hand, determining the fairness of a bid depends more critically on the valuation of the acquirer’s stock rather than how close the offer price is to the target’s past peak price and, therefore, target managers focus on acquirer RPRs (in addition to conventional
CR IP T
valuation levels) in these bid premium negotiations. With more uncertainty about the value of acquirer shares, target managers may demand higher bid premia from acquirers with higher RPRs or be willing to accept lower bid premia from acquirers with lower RPRs. [Insert Table 6 here]
AN US
Table 7 presents five OLS regressions of target CAR on acquirer RPR, acquirer M/B, target RPR, target M/B, and the other control variables used in Table 3, after excluding the potential endogenous variables, stock, cash, tender offer, and hostility. Similar to the analysis of
M
bid premia, the first regression is based on all transactions with available data on targets from CRSP. The second and third columns show results for acquisitions sorted on acquirers’ sigma
ED
and the last two columns present results for the Cash and Non-cash subsamples.19 Acquirer RPR has positive and significant coefficients in the whole sample, in deals of higher uncertainty about
PT
acquirer valuation, and the non-cash deals. To the extent that acquirer RPR positively affects bid
CE
premia, as shown in Table 6, this positive impact is reflected in target CAR. That is, when the valuation level of the acquirer is directly relevant to the bid premium negotiation process (Non-
AC
cash) or the valuation level of the acquirer is more uncertain (High sigma), the acquirer’s RPR positively influences the bid premium and this carries over to a positive influence on the target’s CAR.
19
For the sake of brevity, we include only the results of subsamples sorted according to the acquirers’ sigma. In unreported results, we find a positive and significant coefficient on the acquirers’ RPR in all subsamples sorted on other information environment proxies, similar to the results in the first model of Table 7. Page 35
ACCEPTED MANUSCRIPT
Target RPR and target M/B have negative and significant coefficients in all five models. The results concerning the influence of target M/B on target CAR are consistent with Dong et al. (2006). The magnitudes of the target RPR coefficients appear larger in the Low sigma subsample than in the High sigma subsample, and larger in the Cash subsample than in the Non-cash
CR IP T
subsample. As discussed in Dong et al. (2006), target valuation can affect target CAR via two possible mechanisms. First, as shown in Table 6, targets with higher RPR (or M/B) receive lower premia, particularly among Cash and Low sigma subsamples. Thus, targets with higher RPR (or M/B) earn lower announcement abnormal returns, and this relation is stronger among the Cash
AN US
and Low sigma subsamples. Second, the acquisition announcement triggers a reassessment of valuation levels for acquirer and target shares, and targets with relatively high valuation levels tend to experience a negative reassessment. The coefficients on target RPR in Table 7 are
M
consistent with both mechanisms.
[Insert Table 7 here]
ED
5.2.Method of payment and deal combativeness
Table 8 presents Probit regression models on the method of payment (Stock and Cash) in
PT
Panel A and deal combativeness (Tender offers, Hostile, and Success) in Panel B for the
CE
subsamples of deals involving private targets, public targets, and targets covered in CRSP, respectively.20 We exclude method of payment and deal combativeness variables from the list of
AC
independent variables to avoid potential endogeneity. Our results in Panel A on the acquirer and target M/B ratios are in general consistent with Dong et al. (2006). The first three columns show that the acquirer’s M/B is positively related to the use of stock as the method of payment, consistent with Dong et al. (2006) and valuation theory. Further, in the CRSP subsample, the 20
The samples in Dong et al. (2006) and Baker et al. (2012) are limited to acquisitions of target firms with sufficient data available in CRSP. Therefore, direct comparisons to their results are limited to our CRSP sample results. In addition, (unreported) results from a Logit analysis lead to the same conclusions. Page 36
ACCEPTED MANUSCRIPT
target’s M/B also carries a positive and significant coefficient. The coefficient on the acquirer’s RPR is not statistically significant except within the CRSP subsample that also includes the target’s RPR. In this model, the coefficient on the acquirer’s RPR is positive and marginally significant while the coefficient on the target’s RPR is negative and significant. Thus, there is
CR IP T
some evidence that the acquirer’s RPR positively influences the decision to use stock as the method of payment, consistent with the perceived valuation hypothesis, but the negative coefficient associated with target RPR seems inconsistent with RPR acting as an indicator of valuation levels.
AN US
The last three columns of Panel A in Table 8 present Probit regressions on the use of cash. The pattern of the coefficients on acquirer and target M/B is generally consistent (in the opposite direction) with the coefficients in the first three columns. The coefficient on acquirer
M
RPR is positive in the private subsample and negative in the public and CRSP subsamples but none of the coefficients is statistically significant. The coefficient on the target RPR is positive
ED
and significant. Overall, the results in Panel A offer mixed evidence concerning whether the RPRs for the bidder and target influence the method of payment.
PT
Panel B of Table 8 reports Probit regression results on indicators of deal combativeness,
CE
specifically whether the mode of acquisition is a tender offer, whether the bid is met with hostility, and whether the deal is ultimately successful. Consistent with Dong et al. (2006),
AC
targets with higher M/B are less likely to be hostile and more likely to complete the deal. To the extent that tender offers are mostly paid in cash, bidders with low perceived
valuation should be more likely to make tender offers than bidders with high perceived valuation. However, we find that acquirer RPR is insignificant in the Tender offer regressions. In addition, the perceived valuation hypothesis does not have a specific prediction on how acquirer
Page 37
ACCEPTED MANUSCRIPT
misvaluation affects deal hostility. On the one hand, an overvalued acquirer is expected to offer a high premium, reducing the likelihood of deal combativeness. On the other hand, perceived acquirer overvaluation increases the probability of target hostility. We observe a significant negative effect of acquirer RPR but a positive effect of acquirer M/B on deal hostility in the
significance in the Hostile regression for the private subsample.
CR IP T
CRSP subsample, yet acquirer RPR is insignificant and acquirer M/B is negative with marginal
The coefficients on acquirer RPR in the deal success regressions are positive and significant in the private and public subsamples, and positive but not significant in the CRSP
AN US
subsample. Again, this provides weak evidence that the 52-week high influences perceived valuation levels. Interestingly, the regression analysis in Dong et al. (2006) does not find a statistically significant role for acquirer valuation levels in explaining deal success, but in the
M
public and CRSP subsamples we find a negative and statistically significant impact of acquirer M/B on deal success when acquirer RPR is also included in the regression. In the CRSP
ED
subsample, target M/B is positively related to deal success, consistent with Dong et al. (2006);
PT
the coefficient of target RPR is insignificant. [Insert Table 8 here]
CE
In summary, although acquirer and target RPRs influence acquirer and target CARs and bid premia in a manner consistent with the perceived valuation hypothesis, reference prices have
AC
little consistent impact on the method of payment and deal combativeness. We suggest that because stock market reactions depend on investor perceptions of acquirer and target valuations, reference prices affect acquirer and target CARs. Reference prices also affect bid premia because bid premia are the result of negotiations between target and acquirer managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and
Page 38
ACCEPTED MANUSCRIPT
deal combativeness, on the other hand, are primarily managerial decisions by either the target or the acquirer, and if managers are more informed of their own firms' true valuations, these decisions will be less affected by reference prices. 6. Robustness checks
CR IP T
We examine the robustness of the relation between acquirer RPR and acquirer CAR by considering alternative measures of CAR, alternative measures of RPR, and in subsamples of time periods. Following the analysis of Table 3, we estimate regressions using these alternatives for the overall sample with all targets and the subsamples involving only private and public
AN US
targets.21 The same general pattern emerges throughout this section. That is, the reference price effect is strong in the overall sample and the private subsample but weaker in the public subsample, a result implied by the perceived valuation hypothesis. The full regression
M
specifications follow those of Table 3. For brevity, we only show the coefficients and t-statistics of the acquirer RPR.
ED
6.1.Alternative methods of estimating CAR
In our main analysis, we use the market model approach to estimate abnormal returns
PT
over the event window of [-5, +1]. To check robustness, we estimate three models following,
CE
respectively, Moeller et al. (2004) for CAR 1, a simple market-adjusted abnormal return for CAR 2, and Lang et al. (1991) for CAR 3.22
AC
We use each of these alternative CARs to estimate the baseline regressions, with results reported in Panel A of Table 9. In all regressions, the RPR coefficients are negative and
21
Further analysis indicates that results based on the subsample of transactions involving targets covered in CRSP are robust to the variations in the measurement of acquirer CAR, acquirer RPR, and in time subsamples, though with weaker statistical significance. Unreported for brevity, these results are available upon request. 22 Using additional alternative models (Fuller et al., 2002; Faccio et al., 2006; Officer et al., 2009), as adopted in recent studies, yields similar results. Page 39
ACCEPTED MANUSCRIPT
significant at the 1% level in the overall sample and the subsample involving private targets. As expected, the reference price effect is weaker in acquisitions of public targets. It is possible that the reference price effect exists even without announcements of mergers and acquisitions. That is, there could be a general pattern between RPR and stock
CR IP T
returns. To test this possibility, we re-estimate the acquirers’ CAR using a benchmark formed on reference price ratios. Specifically, for each acquirer in an acquisition deal, we identify a set of control firms whose reference price ratios at day t-6 are similar (i.e., same percentile) to the acquirer and who do not make acquisitions. We then use their average returns over the event
AN US
window [-5, +1] of the acquirer as the benchmark. Thus, the abnormal returns to the acquirer constructed using this approach have already filtered out the “general” return pattern associated with reference price ratios. Using this RPR-controlled cumulative abnormal return as the
M
dependent variable in our baseline regressions yields similarly strong results. Thus, the reference price effect we find is not a general return pattern. Rather, it is associated with the
ED
announcements of mergers and acquisitions.
[Insert Table 9 here]
PT
6.2.Alternative measures of RPR
CE
In the main analysis, we measure the acquirer RPR as of the sixth day prior to the acquisition announcement. To test whether our results are sensitive to this choice, we estimate
AC
three alternative measures of RPR: RPR on the 11th day prior to the announcement, RPR at the month-end prior to the announcement month, and RPR quintiles (0 to 4). We then estimate the baseline regression models by replacing the RPR with each of the three alternative RPR measures. Results are presented in Panel B of Table 9. All RPR coefficients are negative and
Page 40
ACCEPTED MANUSCRIPT
significant for regressions based on the overall sample and the subsample involving private targets. Throughout the paper, we only consider the 52-week high. To the extent that the 52-week low is also salient, it might play a role. To test this possibility, we first add a variable P/L, which
CR IP T
is the ratio of stock price at day t-6 to L, the 52-week low, in the baseline regressions. We find that the coefficients of P/L are not significant and the RPR coefficients are -5.495 (t=-8.53), 6.294 (t=-8.61), and -2.651 (t=-2.01) in the regressions based on the whole sample, the subsample of private targets, and the subsample of public targets, respectively. Further, we
AN US
construct an alternative reference price ratio (P-L)/(H-L), where P is the stock price on day t-6, and H and L are the 52-week high and low prices, respectively (see Lee and Yerramilli, 2016). We find that this new measure has a correlation coefficient of 0.843 with our RPR. When we use
M
this alternative variable in place of the RPR in the baseline regressions, the coefficients on this variable are negative and significant. When both RPR and (P-L)/(H-L) are included in the
ED
regressions, the coefficients on RPR are -8.468 (t=-7.89), -9.949 (t=-8.04), and -3.115 (t=-1.51) in the regressions based on the whole sample, the subsample of private targets, and the
PT
subsample of public targets, respectively. By contrast, the coefficients on (P-L)/(H-L) become
CE
positive.
As another possibility, investors might react differently to announcements of mergers and
AC
acquisitions because the overall market is near or far below its 52-week high. To examine this possibility, we construct the reference price ratio for the Standard & Poor’s (S&P) 500 index, or RPR of SP. We find that firm-level RPR is positively correlated with RPR of SP, with a correlation coefficient of 0.287. When we replace the firm-level RPR with the RPR of SP in our baseline regressions, however, the coefficients are largely positive with weak statistical
Page 41
ACCEPTED MANUSCRIPT
significance. When both firm- and market-level RPRs are in the regressions, the coefficients on the firm-level RPR are -5.802 (t=-8.85), -6.495 (t=-8.80), and -3.236 (t=-2.39), and those on the market-level RPR are 4.943 (t=3.00), 3.490 (t=1.83), and 8.169 (t=2.66) in the regressions based on the whole sample, the subsample of private targets, and the subsample of public targets,
CR IP T
respectively. Likewise, considering the 52-week low of the S&P 500 level in the market-level RPR yields virtually identical results. Thus, the reference price effect we document in this paper is not due to the contemporaneous overall market index relative to its 52-week high or low. 6.3.The reference price effect over time
AN US
We examine whether the reference price effect holds over time. If, for example, investors learn about this bias and correct it over time, we expect the reference price effect to decay over time. To test this possibility, we form three subsamples by the year of announcements: 1981–
M
1990 (1980s), 1991–2000 (1990s), and 2001–2014 (2000s) and estimate the baseline regression models for each of the subsamples. The regression results are listed in Panel C of Table 9. The
ED
general pattern of a stronger reference price effect among acquisitions of private targets than acquisitions of public targets remains. This pattern also holds for each of the three subperiods.
PT
The effect appears stronger over the 1990s in the subsample of private targets.
CE
Further (unreported) analyses are conducted to determine whether investor inattention interacts with the reference price effect. Louis and Sun (2010) document that acquisitions
AC
announced on Fridays receive less investor attention. In subsamples of deals on Fridays and deals on other days we find that the reference price effect exists in both subsamples with similar coefficients as in the baseline regressions.
Page 42
ACCEPTED MANUSCRIPT
7. Conclusion We examine whether acquirer 52-week high prices influence perceived valuation levels, market responses to acquisition announcements, and deal characteristics. In a comprehensive sample of mergers and acquisitions, we document a significant acquirer reference price effect:
CR IP T
Acquirers earn higher (lower) announcement-period returns when their pre-announcement stock prices are well below (near) their 52-week highs. The acquirer reference price effect is stronger in acquisitions of private targets, deals involving greater uncertainty (more volatile acquirers, acquirers followed by fewer analysts, greater relative size, and non-cash payment), and acquirers
AN US
with greater individual investor ownership. Further, the announcement-period abnormal return pattern is reversed in the subsequent year, suggesting that the influence of 52-week high prices on perceived valuation levels is the result of an irrational behavioral bias.
M
We further report that acquirer reference prices affect bid premia and target announcement returns when the uncertainty of acquirer valuation is high and when non-cash
ED
payment is made for the acquisition. By contrast, we find little consistent evidence regarding whether reference prices (of either the acquirer or the target) affect method of payment or deal
PT
combativeness. A possible interpretation is that stock market reactions depend on investor
CE
perceptions of acquirer and target valuations and that investors have less information than managers. Therefore, investors are more likely to rely on heuristics when assessing the valuation
AC
implications of acquisitions. Reference prices also affect bid premia because bid premia are the result of negotiations between target and acquirer managers, and both the acquirer and target reference prices can influence these negotiations. Method of payment and deal combativeness, on the other hand, are primarily decisions by the acquirer and target managers, and if acquirer
Page 43
ACCEPTED MANUSCRIPT
and target managers are more informed of their own firms’ true valuations, these decisions are less likely affected by reference prices. While we focus on M&A events, the anchoring bias and its influence on perceived valuations might also be at work in other corporate events. Such a general pattern would suggest
CR IP T
a theory on how past peak prices affect investors’ reactions to news. The extent to which this conjecture squares with data is left for future research. Regardless of the general pattern, however, our paper provides strong evidence of anchoring bias influencing perceived valuation
AC
CE
PT
ED
M
AN US
levels and investors’ reactions to M&A events.
Page 44
ACCEPTED MANUSCRIPT
Appendix A. Variable definitions
AC
CE
PT
ED
M
AN US
CR IP T
BHAR is calculated using the reference portfolio approach. Each month, we sort all NYSE common stocks into size quintiles based on their month end market capitalization. These quintiles are further sorted into quintiles using book-to-market ratios, which are further sorted into quintiles based on returns over the past 12 months. We then place Amex and Nasdaq firms into one of these 125 portfolios based on their month-end size, book-tomarket ratios, and past 12-month returns. After excluding firms that made significant acquisitions in the past two-year period, we calculate the monthly return for each of the 125 reference portfolios by averaging the monthly returns across all stocks in each portfolio. These portfolio returns are then used as benchmarks to calculate buy-and-hold abnormal returns for our sample firms. The BHAR is measured over the one-year period following the month of announcement. Bid premium is the premium the target firm receives relative to the price four weeks prior to the acquisition announcement, as reported in the SDC. CAR is the cumulative abnormal returns to the acquirer, measured over the event window [–5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. Cash is equal to one if the total consideration is paid in cash. Cross border is equal to one if the acquirer and target come from different countries. Dormant > 1 yr is a dummy variable equal to one if the current transaction is at least one year apart from the previous transaction made by peers in the same CRSP four-digit Standard Industrial Classification code (SIC) industry. The definition follows Cai et al. (2011). Few analysts is binary equal to one if the number of analysts following the acquirer (over the past six months prior to the acquisition announcement month) is below the sample median (by year). Hostile is equal to one if the SDC classifies the deal as hostile or unsolicited. Indiv. own. is one minus the institutional ownership measure, which is the total number of shares owned by institutional investors divided by the number of shares outstanding at the end of the most recent reporting quarter. A missing value for institutional ownership is assigned as zero and the Indiv. own. is assigned as one. Leverage is total assets minus book value of equity, normalized by total assets, all measured at the year-end before announcement. Ln(M/B) is the natural logarithm of the acquirer’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Non-cash is equal to one if at least part of the payment is not in cash. Past return is the raw return over the 12-month period prior to the announcement month. Private is equal to one if the SDC classifies the target public status as subsidiary or private firm. Rel. size is the ratio of the transaction value excluding assumed liability, divided by market capitalization of the acquirer, measured at the beginning of the calendar year in which the announcement occurs. RKRV is defined in Appendix B. RPR is the acquirer’s stock price as of t – 6, the sixth day prior to the announcement, divided by the highest closing price over the past 252 days relative to t – 6. Same industry (or Same ind.) is a dummy variable equal to one if the acquirer and acquired assets share the same two-digit Standard Industrial Classification code (SIC). Page 45
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Sigma is the weekly return standard deviation of the acquirer over the past 52 weeks prior to the acquisition announcement month, as in Zhang (2006). Size is the natural logarithm of market capitalization as of the calendar year-end prior to the announcement in constant $million as of year 2014. Stock is equal to one if the total consideration is paid in stock. Success is equal to one if the transaction is eventually completed, as recorded in the SDC. Target CAR is the cumulative abnormal returns to the target, measured over the event window [– 5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. Target Ln(M/B) is the natural logarithm of the target firm’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Target RPR is the RPR ratio of the target firm (if covered in CRSP) as of the sixth day prior to the announcement date. RPR is the ratio of stock price to the 52-week high price level. Tender offer is equal to one if the deal is classified as a tender offer. Toehold is equal to one if the acquirer owns 5% or more of the target before the announcement.
Page 46
ACCEPTED MANUSCRIPT
Appendix B. RKRV definition We follow Rhodes-Kropf, Robinson, and Viswanathan (2005) and Hertzel and Li (2010) to construct the valuation measure RKRV. All firms covered in the CRSP/Compustat merged database from fiscal years 1977 to 2014 are used in the estimation.
M
AN US
CR IP T
Step 1: For each fiscal year t (from 1977 to 2014) and industry j (of the total 12 Fama and French industries), the following cross-sectional regression is estimated, ( ) , where is the natural logarithm of market value of equity, or log(shrout*prc) from CRSP, measured three months following the fiscal year-end month; is a vector of firm-specific accounting variables at fiscal year t, [ ]; is a vector of conditional accounting multiples [ ]; is the natural logarithm of book equity (Compustat annual item CEQ); is the natural logarithm of net income if it is positive (Compustat annual item NI); is the natural logarithm of the absolute value of net income if it is negative; and is the leverage, equal to 1 – MVE/MVA, where MVE is CRSP shrout*prc, and MVA is MVE plus total assets (Compustat AT) minus deferred taxes (Compustat TXDB) minus book equity (Compustat CEQ). The regressions generate a time series of estimated accounting multiples ̂ .
ED
Step 2: For each fiscal year t (from 1978 to 2014) and industry j, the trailing 20-year moving ∑ averages of the accounting multiples are calculated as ̅ ̂ , where is the total number of accounting multiples available in the trailing 20-year period. ( ̂ ); ( ̂ ) ( ̅ )
CE
PT
Step 3: The firm-specific error is The time series sector error is The long-run value to book is Thus, the RKRV valuation measure is
(
̅ ); and . (
̅ )
AC
Step 4: Each M&A transaction is matched with the acquirer’s RKRV variable such that the announcement month is at least four months, but at most 15 months (inclusive) later than the fiscal year-end month.
Page 47
ACCEPTED MANUSCRIPT
References Akbulut, M., 2013. Do overvaluation-driven stock acquisitions really benefit acquirer shareholders? Journal of Financial and Quantitative Analysis 48 (4), 1025–1055. Amihud, Y., Lev, B., 1981. Risk reduction as a managerial motive for conglomerate mergers. Bell Journal of Economics 12 (2), 605–617.
CR IP T
Andrade, G., Mitchell, M., Stafford, E., 2001. New evidence and perspectives on mergers. Journal of Economic Perspectives 15 (2), 103–120. Baker, M., 2009. Capital market-driven corporate finance. Annual Review of Financial Economics 1, 181–205. Baker, M., Mendel, B., Wurgler, J., 2016. Dividends as reference points: A behavioral signaling approach. Review of Financial Studies 29 (3), 697–738.
AN US
Baker, M., Pan, X., Wurgler, J., 2012. The effect of reference point prices on mergers and acquisitions. Journal of Financial Economics 106 (1), 49–71. Baker, M., Wurgler, J., 2012. Behavioral corporate finance: An updated survey. In George Constantinides, Milton Harris, and Rene Stulz (Ed.), The Handbook of the Economics of Finance: Empirical Corporate Finance, vol. 2. Elsevier, North Holland., pp. 357–424.
M
Baker, M., Xuan, Y., 2016. Under new management: Equity issues and the attribution of past returns. Journal of Financial Economics 121 (1), 66–78.
ED
Barber, B., Odean, T., 2008. All that glitters: The effect of attention and news on the buying behavior of individual and institutional investors. Review of Financial Studies 21 (2), 785–818.
PT
Barberis, N., Shleifer, A., Vishny, R., 1998. A model of investor sentiment. Journal of Financial Economics 49 (3), 307–343. Bartov, E., Radhakrishnan, S., Krinsky, I., 2000. Investor sophistication and patterns in stock returns after earnings announcements. Accounting Review 75 (1), 43–63.
CE
Bereskin, F., Byun, S., Officer, M.S., Oh, J., 2017. The effect of cultural similarity on mergers and acquisitions. Journal of Financial and Quantitative Analysis (forthcoming).
AC
Betton, S., Eckbo, B.E., Thorburn, K., 2008. Corporate takeovers. In: Eckbo, B.E. (Ed.), The Handbook of Corporate Finance: Empirical Corporate Finance, vol. 2., Elsevier, North Holland, pp. 291–421. Bouwman, C., Fuller, K., Nain, A., 2009. Market valuation and acquisition quality: Empirical evidence. Review of Financial Studies 22 (2), 633–679. Cai, J., Song, M.H., Walkling, R.A., 2011. Anticipation, acquisitions and bidder returns: industry shocks and the transfer of information across rivals. Review of Financial Studies 24 (7), 2242–2285. Chang, S., 1998. Takeovers of privately held targets, methods of payment and bidder returns. Journal of Finance 53 (2), 773–784. Page 48
ACCEPTED MANUSCRIPT
Chapman, G., Johnson, E., 1994. The limits of anchoring. Journal of Behavioral Decision Making 7 (4), 223–242. Chopra, N., Lakonishok, J., Ritter, J., 1992. Measuring abnormal returns: Do stocks overreact? Journal of Financial Economics 31 (2), 235–268. Danbolt, J., Siganos, A., Vagenas-Nanos, E., 2015. Investor sentiment and bidder announcement abnormal returns. Journal of Corporate Finance 33, 164–179.
CR IP T
DeGeorge, F., Patel, J., Zeckhauser, R.,1999. Earnings management to exceed thresholds. The Journal of Business 72 (1), 1–33. Dong, M., Hirshleifer, D., Richardson, S., Teoh, S.H., 2006. Does investor (mis)valuation drive the takeover market? Journal of Finance 61 (2), 725–762.
AN US
Driessen, J., Lin, T., Van Hemert, O., 2013. How the 52-week high and low affect option-implied volatilities and stock return moments. Review of Finance 17, 369–401. Duchin, R., Schmidt, B., 2013. Riding the merger wave: Uncertainty, reduced monitoring, and bad acquisitions. Journal of Financial Economics 107 (1), 69–88. Edwards, E.O., Bell, P.W., 1961. The Theory and Measurement of Business Income. University of California Press, Berkeley, CA.
M
El-Khatib, R., Fogel, K., Jandik, T., 2015. CEO network centrality and merger performance. Journal of Financial Economics 116 (2), 349–382. Engelberg, J., McLean, D., Pontiff, J., 2017. Anomalies and news. Journal of Finance (forthcoming).
ED
Faccio, M., McConnell, J., Stolin, D., 2006. Returns to acquirers of listed and unlisted targets. Journal of Financial and Quantitative Analysis 41 (1), 197–220.
PT
Fama, E.F., 1998. Market efficiency, long-term returns, and behavioral finance. Journal of Financial Economics 49 (3), 283–306.
CE
Fama, E.F., French, K.R.,1997. Industry costs of equity. Journal of Financial Economics 43 (2), 153–194. Fama, E.F., French, K.R., 2015. A five-factor asset pricing model. Journal of Financial Economics 116 (1), 1–22.
AC
Fu, F., Lin, L., Officer, M.S., 2013. Acquisitions driven by stock overvaluation: Are they good deals? Journal of Financial Economics 109 (1), 24–39. Fuller, K., Netter, J., Stegemoller, M., 2002. What do returns to acquiring firms tell us? Evidence from firms that make many acquisitions. Journal of Finance 57 (4), 1763–1793. Genesove, D., Mayer, C., 2001. Nominal loss aversion and seller behavior: Evidence from the housing market. Quarterly Journal of Economics 116 (4), 1233–1260. George, T.J., Hwang, C., 2004. The 52-week high and momentum investing. Journal of Finance 59 (5), 2145–2176. Page 49
ACCEPTED MANUSCRIPT
George, T.J., Hwang, C., Li, Y., 2014. Anchoring, the 52-week high and post earnings announcement drift. Unpublished working paper. University of Houston, Nanyang Technological University, and University of Cambridge. George, T.J., Hwang, C., Li, Y., 2018. The 52-week high, q-theory, and the cross section of stock returns. Journal of Financial Economics 128 (1), 148–163.
CR IP T
Golubov, A., Petmezas, D., Travlos, N.G., 2012. When it pays to pay your investment banker: new evidence on the role of financial advisors in M&As. Journal of Finance 67 (1), 271–312. Golubov, A., Yawson, A., Zhang, H., 2015. Extraordinary acquirers. Journal of Financial Economics 116 (2), 314–330. Gompers, P.A., Metrick, A., 2001. Institutional investors and equity prices. Quarterly Journal of Economics 116 (1), 229–259.
AN US
Hansen, R., 1987. A theory for the choice of exchange medium in mergers and acquisitions. Journal of Business 60 (1), 75–96. Harford, J., 1999. Corporate cash reserves and acquisitions. Journal of Finance 54 (6), 1969–1997. Harford, J., Humphery-Jenner, M., Powell, R., 2012. The sources of value destruction in acquisitions by entrenched managers. Journal of Financial Economics 106 (2), 247–261.
M
Hastie, R., Schkade, D., Payne, J., 1999. Juror judgment in civil cases: Effects of plaintiff’s requests and plaintiff’s identity on punitive damage awards. Law and Human Behavior 23 (4), 445–470.
ED
Heath, C., Huddart, S., Lang, M., 1999. Psychological factors and stock option exercise. Quarterly Journal of Economics 114 (2), 601–627.
PT
Hertzel, M.G., Li, Z., 2010. Behavioral and rational explanations of stock price performance around SEOs: Evidence from a decomposition of market-to-book ratios. Journal of Financial and Quantitative Analysis 45 (4), 935–958.
CE
Huddart, S., Lang, M., Yetman, M., 2009. Volume and price patterns around a stock’s 52-week highs and lows: theory and evidence. Management Science 55 (1), 16–31.
AC
Jacowitz, K.E., Kahneman, D., 1995. Measures of anchoring in estimation tasks. Personality and Social Psychology Bulletin 21 (11), 1161–1166. Jarrell, G.A., Brickley, J.A., Netter, J.M., 1988. The market for corporate control: The empirical evidence since 1980. Journal of Economic Perspectives 2 (1), 49–68. Jegadeesh, N., Titman, S., 1993. Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance 48 (1), 65–91. Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76 (2), 323–329.
Page 50
ACCEPTED MANUSCRIPT
Jensen, M., Ruback, R., 1983. The market for corporate control: The scientific evidence. Journal of Financial Economics 11 (1-4), 5–50. Jin, L., Scherbina, A., 2011. Inheriting losers. Review of Financial Studies 24 (3), 786–820. Kahneman, D., 2011. Thinking, Fast and Slow. Farrar, Straus and Giroux, New York.
CR IP T
La Porta, R., Lakonishok, J., Shleifer, A., Vishny, R., 1997. Good news for value stocks: further evidence on market efficiency. Journal of Finance 52 (2), 859–874. Lang, L., Stulz, R., Walkling, R.A., 1989. Managerial performance, Tobin’s q, and the gains from successful tender offers. Journal of Financial Economics 24 (1), 137–154. Lang, L., Stulz, R., Walkling, R.A., 1991. A test of the free cash flow hypothesis: The case of bidder returns. Journal of Financial Economics 29 (2), 315–335.
AN US
Lee, S., Yerramilli, V., 2016. Relative values, announcement timing, and shareholder returns in mergers and acquisitions. Unpublished working paper. University of Houston. Li, J., Yu, J., 2012. Investor attention, psychological anchors, and stock return predictability. Journal of Financial Economics 104 (2), 401–419. Loughran, T., Ritter, J., 2002. Why don't issuers get upset about leaving money on the table in IPOs? Review of Financial Studies 15 (2), 413–444.
M
Louis, H., Sun, A., 2010. Investor inattention and the market reaction to merger announcements. Management Science 56 (10), 1781–1793.
ED
Lyon, J.D., Barber, B.M., Tsai, C., 1999. Improved methods for tests of long-run abnormal stock returns. Journal of Finance 54 (1), 165–201.
PT
Ma, Q., Whidbee, D.A., Zhang, W., 2011. Value, valuation, and the long-run performance of merged firms. Journal of Corporate Finance 17 (1), 1–17.
CE
Malmendier, U., Tate, G., 2008. Who makes acquisitions? CEO overconfidence and the market's reaction. Journal of Financial Economics 89 (1), 20–43. Masulis, R.W., Wang, C., Xie, F., 2007. Corporate governance and acquirer returns. Journal of Finance 62 (4), 1851–1889.
AC
Matsusaka, J. G., 1993. Takeover motives during the conglomerate merger wave. Rand Journal of Economics 24 (3), 357–379. Mitchell, M., Stafford, E., 2000. Managerial decisions and long-term stock price performance. Journal of Business 73 (3), 287–329. Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2004. Firm size and the gains from acquisitions. Journal of Financial Economics 73 (2), 201–228. Moeller, S.B., Schlingemann, F.P., Stulz, R.M., 2005. Wealth destruction on a massive scale? A study of acquiring-firm returns in the recent merger wave. Journal of Finance 60 (2), 757–782. Page 51
ACCEPTED MANUSCRIPT
Morck, R., Shleifer, A., Vishny, R.W., 1990. Do managerial objectives drive bad acquisitions? Journal of Finance 45 (1), 31–48. Mussweiler, T., Strack, F., 2000. Numeric judgments under uncertainty: The role of knowledge in anchoring. Journal of Experimental Social Psychology 36 (5), 495–518.
CR IP T
Nofsinger, J. R., Sias, R.W., 1999. Herding and feedback trading by institutional and individual investors. Journal of Finance 54 (6), 2263–2295. Northcraft, G.B., Neale, M.A., 1987. Experts, amateurs, and real estate: An anchoring-and-adjustment perspective on property pricing decisions. Organizational Behavior and Human Decision Processes 39(1), 84–97. Odean, T., 1998. Are investors reluctant to realize their losses? Journal of Finance 53 (5), 1775–1798.
AN US
Officer, M., 2007. The price of corporate liquidity: Acquisition discount for unlisted targets. Journal of Financial Economics 83 (3), 571–598. Officer, M., Poulsen, A., Stegemoller, M., 2009. Target–firm information asymmetry and acquirer returns. Review of Finance 13 (3), 467–493. Ohlson, J.A., 1995. Earnings, book values, and dividends in equity valuation. Contemporary Accounting Research 11 (2), 661–687.
M
Oppenheimer, D., LeBoeuf, R., Brewer, N., 2008. Anchors aweigh: A demonstration of cross-modality anchoring and magnitude priming. Cognition 106 (1), 13–26.
ED
Rau, P. R., Vermaelen, T., 1998. Glamour, value and the post-acquisition performance of acquiring firms. Journal of Financial Economics 49 (2), 223–253.
PT
Rhodes-Kropf, M., Viswanathan, S., 2004. Market valuation and merger waves. Journal of Finance 59 (6), 2685–2718.
CE
Rhodes-Kropf, M., Robinson, D.T., Viswanathan, S., 2005. Valuation waves and merger activity: The empirical evidence. Journal of Financial Economics 77 (3), 561–603. Roll, R., 1986. The hubris hypothesis of corporate takeovers. Journal of Business 59 (2), 197–216.
AC
Shleifer, A., 2000. Inefficient Markets: An Introduction to Behavioral Finance. Oxford: Oxford University Press, New York. Shleifer, A., Vishny, R.W., 2003. Stock market driven acquisitions. Journal of Financial Economics 70 (3), 295–311. Shue, K., Townsend, R., 2017. Growth through rigidity: An explanation for the rise in CEO pay. Journal of Financial Economics 123 (1), 1–21. Sias, R., Whidbee, D.A., 2010. Insider trades and demand by institutional and individual investors. Review of Financial Studies 23 (4), 1544–1595.
Page 52
ACCEPTED MANUSCRIPT
Slovin, M.B., Sushka, M.E., Polonchek, J.A., 2005. Methods of payment in asset sales: Contracting with equity versus cash. Journal of Finance 60 (5), 2385–2407. Strack, F., Mussweiler, T., 1997. Explaining the enigmatic anchoring effect: Mechanisms of selective accessibility. Journal of Personality and Social Psychology 73 (3), 437–446. Titman, S., Wei, K.C.J., Xie, F., 2004. Capital investments and stock returns. Journal of Financial and Quantitative Analysis 39 (4), 677–700.
CR IP T
Tversky, A., Kahneman, D., 1974. Judgement under uncertainty: heuristics and biases. Science 185 (4157), 1124–1131.
AC
CE
PT
ED
M
AN US
Zhang, X.F., 2006. Information uncertainty and stock returns. Journal of Finance 61(1), 105–136.
Page 53
ACCEPTED MANUSCRIPT
PT
Stock 0.19
Cash 0.23
0.01 0.00 0.00 0.02 0.19 0.23 0.25 0.15 0.24 0.20 0.28 0.32 0.28 0.31 0.34 0.31 0.27 0.27 0.28 0.29 0.22 0.14 0.14 0.10 0.10 0.07 0.08 0.08
0.02 0.00 0.01 0.04 0.36 0.24 0.22 0.21 0.25 0.22 0.19 0.15 0.19 0.19 0.19 0.14 0.17 0.18 0.19 0.21 0.22 0.25 0.27 0.28 0.32 0.34 0.35 0.33
Tender Hostile Success RPR 0.04 0.02 0.88 0.81
AN US
0.68 0.73 0.83 0.72 0.55 0.63 0.59 0.66 0.70 0.79 0.74 0.76 0.80 0.75 0.73 0.74 0.74 0.76 0.72 0.70 0.69 0.78 0.75 0.78 0.79 0.78 0.75 0.75
ED
312.46 140.71 89.95 194.20 277.84 218.29 181.04 180.99 188.92 166.49 112.06 112.00 185.00 152.21 232.11 325.64 293.67 595.03 559.41 1027.34 658.65 275.51 256.64 366.00 599.38 590.15 609.35 688.86
CE
221 236 341 370 278 339 276 335 394 322 387 452 603 766 910 985 1,242 1,348 972 759 623 570 590 694 657 681 591 416
AC
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Private 0.75
M
Panel A: Sample description Transaction Year Nobs value ($m) All 19,119 453.31
CR IP T
Table 1 Sample description and summary statistics. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. Panel A lists the overall and year-by-year averages of sample characteristics. Transaction value ($million) is as reported in SDC; Private is the percentage of transactions involving Private targets (private firms or subsidiaries); Stock (Cash) is the percentage of deals paid for purely with stock (cash); Tender is the percentage of deals that are tender offers; Hostile is the percentage of deals involving target resistance; Success is the percentage of deals that are eventually completed; RPR is the average RPR ratio of the acquirers. RPR ratio is the acquiring firm’s closing stock price on the sixth day prior to the announcement date divided by the stock’s 52-week high. Panel B presents the main variables’ summary statistics. The variables listed in Panel B are defined in Appendices A and B.
Page 54
0.13 0.12 0.07 0.08 0.14 0.12 0.08 0.13 0.08 0.04 0.03 0.02 0.03 0.03 0.04 0.03 0.04 0.03 0.05 0.05 0.05 0.04 0.04 0.03 0.02 0.02 0.03 0.05
0.08 0.06 0.04 0.02 0.04 0.04 0.08 0.07 0.05 0.04 0.03 0.02 0.01 0.02 0.03 0.02 0.02 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.02 0.00 0.01 0.03
0.79 0.83 0.90 0.84 0.71 0.86 0.81 0.78 0.76 0.83 0.84 0.87 0.88 0.87 0.86 0.87 0.90 0.89 0.90 0.89 0.90 0.91 0.93 0.91 0.92 0.90 0.89 0.83
0.83 0.79 0.89 0.79 0.89 0.86 0.81 0.74 0.87 0.75 0.81 0.81 0.83 0.79 0.84 0.84 0.83 0.78 0.73 0.73 0.71 0.74 0.83 0.85 0.83 0.84 0.83 0.69
ACCEPTED MANUSCRIPT
2009 2010 2011 2012 2013 2014
331 418 442 489 508 573
866.73 560.83 776.13 556.45 584.73 1533.99
0.74 0.78 0.84 0.82 0.80 0.80
0.15 0.07 0.06 0.05 0.07 0.11
0.28 0.39 0.36 0.37 0.37 0.33
0.06 0.03 0.02 0.03 0.02 0.02
0.02 0.02 0.03 0.01 0.01 0.02
0.90 0.90 0.90 0.93 0.92 0.89
0.71 0.84 0.83 0.83 0.90 0.86
Mean 1.50 0.81 -3.62 6.30 0.73 0.32 0.30 0.32 0.58 0.79 44.11 21.23
Median 0.57 0.86 -7.74 6.30 0.65 0.18 0.31 0.18 0.52 0.87 35.74 18.17
Std. dev. 9.40 0.18 50.41 2.01 0.73 0.70 0.63 0.35 0.74 0.21 41.95 21.02
Min Q1 -22.93 -3.46 0.01 0.72 -113.35 -32.71 1.84 4.84 -0.80 0.25 -0.72 -0.06 -1.51 -0.06 0.05 0.09 -1.12 0.12 0.13 0.69 -42.95 19.01 -22.14 7.30
AN US
N 19,119 19,119 17,218 19,119 19,119 19,119 18,996 19,119 3,435 3,887 3,624 3,678
AC
CE
PT
ED
M
Variables CAR RPR BHAR Size Ln(M/B) Past return RKRV Rel. size Target Ln(M/B) Target RPR Bid premium Target CAR
CR IP T
Panel B: Summary statistics
Page 55
Q3 5.37 0.95 17.96 7.69 1.12 0.49 0.65 0.40 0.98 0.96 60.74 31.70
Max 36.57 1.00 211.62 11.12 3.19 3.81 2.16 1.72 3.03 1.00 231.72 94.91
CR IP T
ACCEPTED MANUSCRIPT
AN US
Table 2 Acquirer cumulative abnormal returns (CAR) by RPR groups. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. The sample is two-way sorted yearly on the target listing status (private vs. public) and the acquirer RPR ratio. The Low and High RPR subsamples are formed by the sample medians of acquirer RPR ratios within each listing-status group. Panel A shows the mean RPR values and the number of observations for the whole sample and the subsamples of private and public targets, respectively. In Panel B, the sample is three-way sorted on target listing status, acquirer RPR, and acquirer M/B ratios. For the whole sample and the subsamples, Panel B reports the mean values of the acquirer cumulative abnormal returns (CAR). Row “H – L” presents the differences in CAR between the High RPR and Low RPR groups, with t-statistics in square brackets. Acquirer CAR is the cumulative abnormal returns to the acquirer, measured over the event window [–5, +1], adjusted by the market model expected return estimated over the [-370, -253] window. RPR is the acquirer’s stock price as of t – 6, the sixth day prior to the announcement, divided by the highest closing price over the past 252 days relative to t – 6. Acquirer M/B is the acquirer’s market value of equity divided by book value of equity, all measured at the fiscal year-end before the announcement. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
All targets 19,119 9,543 9,576
Panel B: Acquirer CAR (%) by target listing status, acquirer RPR, and acquirer M/B Groups All targets Private targets All Low High All Low M/B M/B M/B M/B M/B All RPR 1.501a 1.624a 1.378a 2.320a 2.364a Low RPR (L) 2.021a 2.272a 1.787a 2.916a 3.124a a a a a High RPR (H) 0.982 1.017 0.945 1.726 1.648a
High M/B 2.276a 2.719a 1.808a
All M/B -0.911a -0.618a -1.201a
H-L t(H - L)
-0.911 [-3.89]a
-0.582 [-2.36]b
-1.255 [-6.89]a
AC
CE
-1.040 [-7.66]a
PT
ED
M
Panel A: Acquirer RPR levels RPR groups Average acquirer RPR level All targets Private targets Public targets All RPR 0.807 0.800 0.829 Low RPR (L) 0.677 0.665 0.713 High RPR (H) 0.937 0.934 0.944
-0.842 [-4.18]a
-1.190 [-7.48]a
Page 56
-1.476 [-6.86]a
No. of obs. Private targets 14,270 7,128 7,142
Public targets 4,849 2,415 2,434
Public targets Low M/B -0.564a -0.278 -0.827a -0.549 [-1.72]c
High M/B -1.255a -0.932a -1.600a -0.668 [-1.78]c
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Table 3 Baseline acquirer CAR regressions. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. This table presents the coefficients and t-statistics (in parentheses) for four OLS regressions based on the whole sample (All), the subsample of deals involving only private targets (Private), the subsample of deals involving only public targets (Public), and the subsample of deals involving targets that are covered in CRSP (CRSP). Estimated t-statistics are based on robust standard errors clustered by firm. The dependent variable is the acquirer CAR as measured over the seven-day event window [-5, +1]. All regressions include Fama and French (1997) industry and year fixed effects. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 57
ACCEPTED MANUSCRIPT
Stock Cash Private Stock*Private Rel. size Size Leverage Dormant > 1 yr
M
Same industry Tender offer
ED
Hostile
CE
Past return
PT
Toehold Cross border
Private -6.286 (-8.66)a -0.168 (-1.12) 0.709 (2.05)b -0.138 (-0.74)
3.873 (11.33)a -0.249 (-3.97)a -0.670 (-1.51) 0.512 (2.04)b -0.079 (-0.43) 2.751 (0.82) -2.919 (-1.53) 0.566 (0.56) -0.352 (-1.32) 0.792 (4.24)a
Public -2.581 (-1.97)b -0.460 (-1.63) -0.402 (-1.31) 1.187 (3.74)a
CRSP -3.100 (-1.92)c -0.228 (-0.67) 0.078 (0.21) 1.429 (3.64)a
-0.486 (-1.34) -0.579 (-6.54)a 1.252 (1.49) 1.091 (2.59)a 0.637 (2.24)b 0.651 (1.77)c -0.148 (-0.36) -0.479 (-0.88) 0.413 (0.77) 0.774 (2.13)b
-0.900 (-2.19)b -0.466 (-4.19)a 0.961 (1.02) 1.276 (2.66)a 0.589 (1.75)c 0.801 (1.88)c 0.371 (0.79) -0.207 (-0.33) -1.876 (-1.56) 0.713 (1.68)c 4.607 (4.28)a -0.524 (-1.70)c 3.969 (2.21)b 3,336 0.048
CR IP T
Ln(M/B)
All -5.487 (-8.56)a -0.285 (-2.14)b -1.208 (-4.06)a 0.145 (0.91) 2.321 (10.60)a 2.125 (5.04)a 2.216 (8.80)a -0.398 (-7.69)a -0.081 (-0.21) 0.628 (2.88)a 0.107 (0.67) 1.058 (2.22)b -0.759 (-1.89)c 0.047 (0.10) -0.187 (-0.77) 0.777 (4.70)a
AN US
Variables RPR
Target RPR
AC
Target Ln(M/B) Intercept N Adj. R2
7.284 (9.46)a 19,119 0.064
8.711 (10.45)a 14,270 0.054
Page 58
7.776 (5.03)a 4,849 0.049
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 4 Acquirer CAR regressions with interaction terms. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period. This table presents the coefficients and t-statistics (in parentheses) for ten OLS regressions, with the first five based on the subsample of acquisitions involving private targets and the last five based on the subsample of acquisitions involving public targets. Estimated t-statistics are based on robust standard errors clustered by firm. The dependent variable is the acquirer CAR as measured over the seven-day event window [-5, +1]. The continuous interaction variables Sigma, Rel. size, and Indiv. own. are demeaned. The full regression specifications follow those of Table 3. For brevity, coefficients and t-statistics of all other variables are not shown. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Variables Private targets Public targets Model 1 2 3 4 5 Model 6 7 8 9 10 RPR -4.697 -3.480 -6.290 -3.564 -5.226 -2.927 -0.356 -1.829 -3.684 -2.187 (-5.73)a (-3.75)a (-8.66)a (-2.80)a (-7.36)a (-1.97)b (-0.20) (-1.38) (-1.83)c (-1.69)c RPR*Sigma -0.455 -0.534 (-2.90)a (-1.94)c RPR*Few analysts -4.822 -3.815 (-4.04)a (-1.87)c RPR*Rel. size -6.623 -5.461 (-3.42)a (-2.10)b RPR*Non-cash -3.364 1.324 (-2.40)b (0.61) RPR*Indiv. own. -0.112 -0.103 (-5.40)a (-2.82)a Sigma 0.364 0.241 (3.01)a (1.08) Few analysts 3.807 2.939 (3.69)a (1.63) Indiv. own. 0.094 0.092 (5.21)a (2.87)a Rel. size 3.857 3.844 8.969 3.859 3.841 -0.447 -0.518 4.023 -0.493 -0.492 (11.31)a (11.27)a (5.53)a (11.29)a (11.27)a (-1.23) (-1.42) (1.77)c (-1.36) (-1.36) N 14,270 14,270 14,270 14,270 14,270 4,849 4,849 4,849 4,849 4,849 Adj. R2 0.056 0.056 0.057 0.055 0.058 0.052 0.050 0.051 0.049 0.053 Page 59
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 5 Long-term acquirer returns. The sample includes a total of 17,218 acquisitions made by U.S. acquirers over the 1981–2014 period, for which the BHAR data are available. In the whole sample and the two subsamples formed on target listing status, the observations are further sorted yearly into two equal groups based on acquirer RPR, acquirer M/B, or acquirer RKRV. For each of the subsamples, Panel A presents the average BHAR, Panel B presents the Fama and French (2015) five-factor monthly alphas of calendar-time portfolio returns (CTPR), and Panel C presents the average abnormal returns of the four earnings announcements over the subsequent one-year period. Both BHAR and CTPR returns are measured over the one-year period starting from the month after the announcement. For each of the subsequent quarterly earnings announcements, the abnormal returns are measured over the [-1,+1] window, adjusted by the CRSP equal-weighted market returns. Rows “H – L” show the return differences of the two groups formed on acquirer RPR, acquirer M/B, or acquirer RKRV, with t-statistics in square brackets. Panel D presents regressions of BHAR based on the whole sample (All), the private and public subsamples, and the subsample of deals involving targets covered in CRSP, respectively. Estimated t-statistics (in parentheses) are based on robust standard errors clustered by firm. All regressions include Fama and French (1997) industry and year fixed effects. All variables are defined in Appendices A and B. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 60
CR IP T
ACCEPTED MANUSCRIPT
Panel A: Acquirer buy-and-hold abnormal returns (BHAR) (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -3.625a -3.388a -4.278a -3.625a -3.388a Low (L) -5.889a -5.589a -6.720a -2.792a -2.470a a b b a High (H) -1.369 -1.194 -1.854 -4.455 -4.305a
Public -4.278a -3.683a -4.868a
H–L t(H - L)
-1.835 [-1.95]c
-1.185 [-0.93]
Panel B: Acquirer Fama and French (2015) five-factor monthly alphas (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -0.057 -0.034 -0.194b -0.057 -0.034 Low (L) -0.165 -0.124 -0.342b 0.061 0.086 High (H) 0.017 0.011 -0.052 -0.171 -0.157
Public -0.194b -0.199c -0.195
4.395 [4.68]a
4.866 [3.83]a
0.003 [0.02]
Panel C: Acquirer average earnings announcement-period abnormal returns (%) Sort variable Sort by RPR Sort by M/B All targets Private Public All targets Private All -0.006 0.010 -0.050 -0.006 0.010 Low (L) -0.108b -0.080 -0.185b 0.139a 0.186a High (H) 0.096b 0.100b 0.085 -0.150a -0.165a
Public -0.050 0.009 -0.109
H–L t(H – L)
-0.118 [-1.13]
0.135 [0.87]
-0.231b [-2.30]b
0.180 [2.27]b
0.270 [2.58]a
-0.289 [-4.47]a
AC
CE
0.204 [3.16]a
PT
ED
0.181 [1.29]
M
-0.242b [-2.02]b
H-L t(H – L)
0.290c [1.76]c
-1.663 [-2.16]b
AN US
4.520 [5.89]a
Page 61
-0.351 [-4.42]a
Sort by RKRV All targets Private Public -3.628a -3.400a -4.256a -2.776a -2.684a -3.028a a a -4.477 -4.113 -5.476a -1.701 [-2.21]b
-1.429 [-1.52]
-2.448 [-1.92]c
Sort by RKRV All targets Private Public -0.063 -0.041 -0.198b 0.146 0.192 -0.065 -0.272a -0.271b -0.333a -0.419a [-4.05]a
-0.464a [-4.05]a
-0.268c [-1.95]c
Sort by RKRV All targets Private Public -0.005 0.012 -0.050 0.056 0.067 0.028 -0.066 -0.043 -0.127c -0.122 [-1.89]c
-0.110 [-1.39]
-0.154 [-1.47]
ACCEPTED MANUSCRIPT
ED
PT
CE
CRSP 20.677 (2.59)a -1.098 (-0.63) -2.069 (-1.07) 5.389 (2.48)b
0.329 (0.20) 0.501 (1.17) -4.092 (-0.94) 0.200 (0.09) 4.511 (2.96)a -1.198 (-0.60) 1.107 (0.45) 0.177 (0.05) 0.937 (0.34) -4.882 (-3.04)a
0.973 (0.50) 0.439 (0.81) -4.295 (-0.83) -1.391 (-0.56) 5.014 (2.77)a -0.579 (-0.25) -0.528 (-0.19) -3.127 (-0.81) -5.270 (-1.06) -3.700 (-1.78)c 3.465 (0.65) -1.831 (-1.25) -30.506 (-3.39)a 3,257 0.030
AN US
-1.167 (-0.67) -0.769 (-1.92)c 5.381 (1.86)c 2.033 (1.41) 3.945 (3.78)a 3.970 (0.50) -8.702 (-1.12) -1.723 (-0.39) -1.627 (-1.06) -2.830 (-2.50)b
Public 24.813 (4.01)a -1.139 (-0.81) -2.827 (-1.80)c 4.008 (2.28)b
CR IP T
Private 20.752 (5.03)a -1.674 (-1.75)c -4.075 (-2.11)b 0.861 (0.77)
M
Panel D: Regressions of acquirer BHAR Variables All RPR 20.803 (5.86)a Ln(M/B) -1.447 (-1.77)c Stock -3.214 (-2.14)b Cash 1.324 (1.42) Private 0.747 (0.66) Stock*Private -0.845 (-0.38) Rel. size -1.019 (-0.80) Size -0.421 (-1.37) Leverage 2.905 (1.17) Dormant > 1 yr 1.628 (1.36) Same industry 4.039 (4.58)a Tender offer 0.694 (0.36) Hostile 0.529 (0.22) Toehold -0.400 (-0.14) Cross border -1.181 (-0.88) Past return -3.134 (-3.18)a Target RPR
AC
Target Ln(M/B) Intercept N Adj. R2
-20.965 (-4.65)a 17,218 0.016
-18.531 (-3.56)a 12,637 0.015
Page 62
-27.530 (-3.92)a 4,581 0.030
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
CR IP T
Table 6 Bid premium. The sample includes a total of 2,757 acquisitions made by U.S. acquirers over the 1981–2014 period involving targets covered in CRSP that have data on bid premium and target announcement returns. This table presents the coefficients and t-statistics (in parentheses) for five OLS regressions on bid premium, one for the whole sample and four on the subsamples. The Low (High) sigma subsample includes deals with acquirer sigma below (above) the sample median; the Cash subsample includes all deals with cash as the sole consideration of payment; the Non-cash subsample includes all deals with at least some payment not in cash. All regressions include Fama and French (1997) industry and year fixed effects. Estimated tstatistics are based on robust standard errors clustered by firm. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Whole sample Low sigma High sigma Cash Non-cash RPR 9.814 0.454 17.782 -25.359 15.649 b c (1.42) (0.04) (2.06) (-1.82) (1.96)c Ln(M/B) 4.207 3.499 3.961 2.996 4.726 (2.83)a (1.66)c (1.84)c (1.12) (2.61)a Target RPR -10.132 -16.713 -2.620 -27.785 -8.614 (-1.96)c (-1.88)c (-0.40) (-3.04)a (-1.33) Target Ln(M/B) -6.130 -2.886 -8.808 -5.025 -6.492 a a c (-3.98) (-1.44) (-3.90) (-1.94) (-3.41)a Rel. size -3.830 -5.670 -2.643 -2.102 -3.306 b b (-2.21) (-2.43) (-1.02) (-0.52) (-1.64) Size 0.013 0.012 -0.142 1.472 -0.260 (0.02) (0.02) (-0.16) (1.26) (-0.42) Leverage -4.088 -4.235 -1.966 9.348 -7.336 (-0.89) (-0.58) (-0.31) (1.17) (-1.31) Dormant > 1 yr 2.846 4.752 -0.071 -1.127 4.870 (1.14) (1.40) (-0.02) (-0.25) (1.54) Same industry -2.379 -5.043 2.093 -4.337 -0.926 (-1.44) (-2.41)b (0.82) (-1.49) (-0.45) Toehold -3.346 -1.114 -3.948 -5.257 -1.805 (-0.90) (-0.21) (-0.73) (-1.02) (-0.36) Cross border 3.456 2.447 10.138 4.381 4.965 (0.57) (0.32) (1.13) (0.61) (0.69) Past return 5.220 -1.689 4.652 -1.822 6.191 a b (3.07) (-0.41) (2.53) (-0.51) (3.29)a Intercept 41.395 60.131 26.776 74.611 34.827 (5.18)a (4.16)a (2.31)b (5.07)a (3.61)a N 2,757 1,371 1,386 637 2,120 Adj. R2 0.090 0.128 0.068 0.096 0.094
Page 63
ACCEPTED MANUSCRIPT
Rel. size Size Leverage Dormant > 1 yr Same industry
CE
Toehold Cross border
AC
Past return Intercept N Adj. R2
High sigma 21.564 (4.81)a 0.129 (0.13) -17.449 (-4.83)a -7.428 (-7.50)a -6.449 (-5.01)a 0.876 (1.73)c -3.174 (-0.97) -3.000 (-1.50) 3.189 (2.30)b -1.458 (-0.58) 2.086 (0.45) 0.899 (1.09) 27.363 (4.43)a 1,386 0.143
AN US
Target Ln(M/B)
Low sigma 2.060 (0.27) 1.628 (1.46) -33.898 (-6.97)a -3.445 (-3.39)a -8.836 (-7.09)a 0.346 (0.88) 0.793 (0.20) 2.714 (1.47) -1.410 (-1.15) -5.811 (-3.00)a 0.449 (0.08) 1.445 (0.55) 49.850 (5.91)a 1,371 0.230
M
Target RPR
ED
Ln(M/B)
Whole sample 17.556 (4.75)a 0.951 (1.26) -21.890 (-7.75)a -5.707 (-8.12)a -7.363 (-8.15)a 0.697 (2.29)b -1.622 (-0.68) 0.004 (0.00) 0.128 (0.14) -3.483 (-2.25)b -0.362 (-0.11) 1.444 (1.71)c 29.738 (6.74)a 2,757 0.153
PT
Variables RPR
CR IP T
Table 7 Target CAR. The sample includes a total of 2,757 acquisitions made by U.S. acquirers over the 1981–2014 period involving targets covered in CRSP that have data on bid premium and target CAR. This table presents the coefficients and t-statistics (in parentheses) for five OLS regressions on target CAR, one for the whole sample and four on the subsamples. The Low (High) sigma subsample includes deals with acquirer sigma below (above) the sample median; the Cash subsample includes all deals with cash as the sole consideration of payment; the Non-cash subsample includes all deals with at least some payment not in cash. All regressions include Fama and French (1997) industry and year fixed effects. Estimated t-statistics are based on robust standard errors clustered by firm. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Page 64
Cash -0.165 (-0.02) -1.058 (-0.75) -46.359 (-7.93)a -3.571 (-2.34)b -5.764 (-2.36)b 1.233 (1.80)c 0.373 (0.08) -0.099 (-0.04) -1.559 (-0.91) -1.556 (-0.54) 4.150 (0.59) -2.225 (-1.02) 63.423 (6.75)a 637 0.235
Non-cash 19.132 (4.56)a 2.050 (2.35)b -19.402 (-5.98)a -5.924 (-7.52)a -6.290 (-6.21)a 0.670 (2.01)b -1.300 (-0.46) 0.023 (0.01) 1.388 (1.31) -4.687 (-2.62)a -0.787 (-0.21) 2.189 (2.39)b 22.744 (4.54)a 2,120 0.136
ACCEPTED MANUSCRIPT
Table 8 Probabilistic regressions of deal characteristics. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981 –2014 period, with 14,270 involving private targets, 4,849 involving public targets, and 3,336 involving targets covered in CRSP. Panels A and B present 15 Probit regression models with the dependent variables shown in the top rows and samples in the second rows. Year fixed effects are included but not reported. All variables are defined in Appendix A. Estimated t-statistics are in parentheses. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively.
Private -0.038 (-0.34) 0.198 (9.50)a
Stock Public 0.243 (1.34) 0.163 (4.98)a
-0.041 (-3.60)a 0.136 (3.06)a 0.129 (2.10)b 4.616 (8.38)a -0.306 (-5.87)a 0.180 (5.96)a 0.512 (2.61)a -0.216 (-4.07)a 0.062 (2.73)a -1.574 (-12.64)a 0.148 14,270 12,498 1,772
-0.032 (-2.41)b 0.116 (2.34)b 0.552 (6.29)a 7.909 (8.00)a -0.348 (-4.85)a 0.191 (4.36)a -0.300 (-2.81)a -0.318 (-4.08)a 0.057 (1.34) -0.612 (-3.67)a 0.141 4,849 2,939 1,910
RPR Ln(M/B) Target RPR
Leverage Sigma Dormant > 1 yr Same industry Toehold
CE
Cross border
ED
Rel. size
PT
Size
AC
Past return Intercept
McFadden's LRI N N(dep. var. = 0) N(dep. var. = 1)
M
Target Ln(M/B)
CRSP 0.384 (1.68)c 0.080 (1.87)c -0.582 (-3.99)a 0.152 (3.77)a -0.028 (-1.58) 0.126 (2.12)b 0.743 (6.71)a 9.503 (7.26)a -0.351 (-4.18)a 0.155 (2.95)a -0.369 (-2.97)a 0.115 (0.71) 0.077 (1.43) -0.192 (-0.85) 0.146 3,336 1,982 1,354
Private 0.105 (1.02) -0.067 (-3.47)a
AN US
Variables
CR IP T
Panel A: Method of payment
Page 65
0.065 (6.97)a -0.149 (-3.48)a -0.065 (-1.22) -3.666 (-6.65)a 0.083 (2.29)b 0.090 (3.68)a -0.679 (-3.07)a 0.037 (0.96) -0.049 (-2.02)b -0.868 (-10.15)a 0.077 14,270 10,933 3,337
Cash Public -0.169 (-0.79) -0.108 (-2.87)a
-0.015 (-1.01) -0.561 (-9.09)a -0.803 (-8.23)a -7.302 (-5.98)a 0.197 (2.79)a -0.303 (-6.52)a 0.301 (2.87)a 0.424 (5.64)a -0.079 (-1.43) 0.007 (0.04) 0.138 4,849 3,799 1,050
CRSP -0.356 (-1.32) -0.038 (-0.79) 0.658 (4.01)a -0.108 (-2.37)b -0.056 (-2.77)a -0.720 (-9.57)a -0.918 (-7.43)a -7.088 (-4.42)a 0.176 (2.09)b -0.346 (-6.15)a 0.283 (2.26)b -0.139 (-0.78) -0.047 (-0.70) -0.235 (-0.92) 0.154 3,336 2,599 737
ED
PT
CE
AC
Hostile Public -0.580 (-2.12)b 0.061 (1.28)
CRSP -0.701 (-2.12)b 0.122 (2.04)b 0.141 (0.67) -0.291 (-4.96)a 0.101 (4.08)a 0.576 (7.45)a -0.273 (-1.76)c -4.345 (-2.22)b 0.131 (1.29) -0.043 (-0.60) 0.864 (6.99)a 0.220 (1.08) 0.047 (0.59) -2.161
AN US
Private 0.045 (0.07) -0.222 (-1.71)c
0.175 (3.50)a 0.770 (5.07)a -0.013 (-0.04) -3.788 (-0.98) -0.457 (-1.63) 0.107 (0.78) 1.665 (6.48)a -0.195 (-0.54) -0.039 (-0.22) -8.918
M
Panel B: Tender offer, hostility, and deal success Variables Tender offer Private Public CRSP RPR 0.127 -0.135 -0.251 (0.23) (-0.63) (-0.96) Ln(M/B) -0.298 -0.030 -0.053 (-2.79)a (-0.76) (-1.06) Target RPR 0.117 (0.70) Target Ln(M/B) -0.060 (-1.32) Size 0.115 0.132 0.118 (2.69)a (7.94)a (5.58)a Rel. size 0.535 0.055 -0.009 (3.74)a (0.91) (-0.13) Leverage -0.656 -1.066 -1.065 (-2.52)b (-9.90)a (-8.07)a Sigma -3.141 -3.186 -3.332 (-0.99) (-2.55)b (-2.11)b Dormant > 1 yr -0.356 0.277 0.310 (-1.58) (3.83)a (3.74)a Same industry -0.293 -0.179 -0.210 (-2.44)b (-3.62)a (-3.63)a Toehold 1.414 0.438 0.420 (5.82)a (4.32)a (3.53)a Cross border 0.042 0.943 0.327 (0.15) (12.69)a (1.94)c Past return 0.153 -0.099 -0.032 (1.30) (-1.82)c (-0.51) Intercept -8.126 -1.827 -1.516
CR IP T
ACCEPTED MANUSCRIPT
Page 66
0.106 (5.31)a 0.618 (9.37)a -0.219 (-1.69)c -3.831 (-2.40)b 0.159 (1.82)c -0.092 (-1.50) 0.745 (7.02)a 0.341 (3.53)a -0.028 (-0.40) -2.340
Private 0.191 (1.66)c -0.020 (-0.91)
Success Public 0.416 (2.13)b -0.061 (-1.75)c
0.051 (4.50)a -0.230 (-5.24)a -0.407 (-6.34)a -2.355 (-4.07)a 0.033 (0.71) 0.040 (1.34) -0.144 (-0.80) -0.119 (-2.45)b 0.031 (1.18) 1.190
0.032 (2.19)b -0.396 (-7.68)a 0.168 (1.78)c -1.041 (-0.99) -0.033 (-0.46) 0.132 (2.85)a -0.307 (-3.10)a -0.233 (-2.97)a 0.039 (0.83) 0.395
CRSP 0.292 (1.18) -0.125 (-2.71)a 0.177 (1.13) 0.166 (3.77)a 0.024 (1.25) -0.418 (-6.68)a 0.138 (1.15) -0.485 (-0.35) -0.029 (-0.35) 0.118 (2.10)b -0.418 (-3.59)a -0.293 (-1.75)c 0.029 (0.50) 0.243
(-8.09)a 0.140 4,849 4,077 772
(-5.36)a 0.099 3,336 2,780 556
(-27.30)a 0.348 14,270 14,227 43
(-9.22)a 0.112 4,849 4,480 369
(-6.66)a 0.125 3,336 3,041 295
AC
CE
PT
ED
M
AN US
McFadden's LRI N N(dep. var. = 0) N(dep. var. = 1)
(-30.78)a 0.400 14,270 14,195 75
CR IP T
ACCEPTED MANUSCRIPT
Page 67
(10.62)a 0.043 14,270 1,384 12,886
(2.18)b 0.065 4,849 929 3,920
(0.99) 0.084 3,336 648 2,688
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
M
AN US
Table 9 Robustness checks. The sample includes a total of 19,119 acquisitions made by U.S. acquirers over the 1981–2014 period, with 14,270 (4,849) involving private (public) targets. For the whole sample and subsamples, this table presents regressions of acquirer CAR by varying CARs in Panel A, varying RPRs in Panel B, and subperiods in Panel C. CAR 1 is estimated as in Moeller et al. (2004), CAR 2 is estimated over the [-5, +1] window with return adjusted by market return, and CAR 3 follows Lang et al. (1991). RPR 1 is measured on the 11 th day prior to the announcement, RPR 2 is measured at the month-end prior to the announcement, and RPR 3 is the RPR quintiles (0 to 4). The three subperiods are defined as 1981–1990 (1980s), 1991–2000 (1990s), and 2001–2014 (2000s) based on the year of announcements. The full regression specifications follow those of Table 3. For brevity, coefficients and t-statistics (in parentheses) of all other variables are not shown. All variables are defined in Appendix A. Superscripts a, b, and c denote statistical significance at the 1%, 5%, and 10% levels, respectively. Panel A: Robustness by alternative CARs Variables All targets Private targets Public targets CAR 1 CAR 2 CAR 3 CAR 1 CAR 2 CAR 3 CAR 1 CAR 2 CAR 3 RPR -4.477 -5.989 -4.938 -5.143 -6.661 -5.728 -1.584 -3.590 -2.168 (-8.88)a (-9.50)a (-6.65)a (-9.35)a (-9.25)a (-6.74)a (-1.48) (-2.89)a (-1.46) N 19,119 19,119 19,119 14,270 14,270 14,270 4,849 4,849 4,849 Adj. R2 0.073 0.070 0.057 0.054 0.060 0.047 0.073 0.051 0.054 Panel B: Robustness by alternative RPRs Variables All targets Private targets Public targets RPR 1 RPR 2 RPR 3 RPR 1 RPR 2 RPR 3 RPR 1 RPR 2 RPR 3 RPR -4.245 -3.508 -0.410 -5.027 -4.018 -0.508 -1.319 -2.015 -0.158 (-6.57)a (-5.69)a (-5.66)a (-6.88)a (-5.68)a (-5.95)a (-1.00) (-1.63) (-1.19) N 19,119 19,119 19,119 14,270 14,270 14,270 4,849 4,849 4,849 Adj. R2 0.061 0.060 0.059 0.051 0.049 0.048 0.048 0.049 0.048 Panel C: Robustness by subperiods Variables All targets Private targets Public targets 1980s 1990s 2000s 1980s 1990s 2000s 1980s 1990s 2000s RPR -5.811 -7.725 -3.390 -6.467 -8.452 -3.947 -4.219 -3.881 -2.032 (-4.42)a (-7.12)a (-3.82)a (-4.11)a (-6.88)a (-4.04)a (-1.82)c (-1.89)c (-0.95) N 3,112 8,424 7,583 2,154 6,247 5,869 958 2,177 1,714 Adj. R2 0.070 0.078 0.050 0.056 0.072 0.037 0.074 0.044 0.075
Page 68
ACCEPTED MANUSCRIPT
Panel A: Acquisitions of private targets
3.00 2.50
CR IP T
2.00
1.50
Low RPR High RPR
1.00 0.50
Days relative to announcement date
0.00 -6
-4
-2
0
2
Panel B: Acquisitions of public targets
8
10
12
14
16
18
20
Low RPR High RPR
CE
PT
-0.50
-1.50
6
ED
0.00
-1.00
4
M
0.50
-2.00
Days relative to announcement date
-2.50
AC
Cumulative abnormal return (CAR, in %)
1.00
AN US
Cumulative abnormal return (CAR, in %)
3.50
-6
-4
-2
0
2
4
6
8
10
12
14
16
18
20
Fig. 1. Acquirer cumulative abnormal returns and reference price ratios. The sample includes 14,270 acquisitions of private targets (Panel A) and 4,849 acquisitions of public targets (Panel B) over the 1981–2014 period made by U.S. acquirers. Within each subsample, acquirers are yearly sorted into two equal groups by their reference price ratios as of the sixth day prior to the announcement date. Low RPR (High RPR) represents the group with reference price ratios below (above) the subsample median. The abnormal returns (in %) are based on a market model estimated from days [-370, -253] relative to the announcement date. Page 69