WHY MEDIA SLANT? Media coverage of mergers and acquisitions in China

Abstract

Using a text analysis of Chinese newspaper articles covering 1321 proposed mergers during 2008–2018, this study proposes attribution theory to examine how the media is susceptible to stereotype bias. Evidence reveals that the media pays considerable attention to and exhibits favourable sentiments toward overseas-experienced acquirers, which is found only in non-state-owned enterprises. Further analyses on the amplifying effect of stereotype bias show that the media slant more positively on large-scale overseas experiences. Results indicate that the media is biased, referring to impression migration from merger and acquisition experience.

Keywords:

• Media bias
• mergers and acquisitions
• overseas experience
• corporate governance

JEL classification:

• G34
• G14
• G41

1. Introduction

The media plays an essential role by providing information or monitoring firm behaviour for stakeholders (Bednar, 2012; Deephouse, 2000; Dyck et al., 2010; Miller, 2006; Pollock & Rindova, 2003; Westphal et al., 2012). Specifically, the media is proved to affect the outcome of merger and acquisition (M&A) transactions (Borochin & Cu, 2018). Social psychology perspective emphasizes how media bias influences public opinion and views (Baron, 2006; Cohen et al., 2017; Pollock & Rindova, 2003) and provides a useful framework for understanding media tendency in democratic societies (Baron, 2006; Chiang & Knight, 2011; Gentzkow & Shapiro, 2006, 2010; Hopmann et al., 2010; Reuter & Zitzewitz, 2006). However, the inherent media bias in socialist countries has not yet to be clearly examined. This study aims to investigate the forces that shape media bias in M&As in China, which is the largest socialist country and newspaper market in the world.

Attribution theory provides a theoretical framework for answering the question on how the media deviates from objectivity. Limitations in information processing capacity allow evaluators to simplify the complex problem by applying an event to a given experience (Jones & Davis, 1965). For example, one of the causes of stereotypes is that evaluators attribute an event to a character regardless of situational factors (Myers & Spencer, 2006). Addressed concretely to media bias in M&As, political goals and profitable incentives have revealed the manifestation of visible biases (Gurun & Butler, 2012; Lott & Hassett, 2014). However, invisible media bias, such as stereotype bias, is ubiquitous but rarely discussed. The development of text analysis tool makes it possible to analyse the sentiment of newspaper articles and provides methods to visualize invisible bias (Bednar, 2012; Borochin & Cu, 2018; Tetlock et al., 2008; You, Zhang, et al., 2018). We investigate stereotype bias by examining how the impression of overseas experience migrates in media coverage on M&As in China. By outlining invisible bias of media response, this article brings a behavioural perspective to the understanding of the media in M&As and ultimately raises questions about how effectively the media can function as an infomediary or social arbitrator.

2. Literature review

2.1. The role of media coverage

The media is believed to be an integral part of corporate governance, especially in asset pricing (Bhattacharya et al., 2009) and capital allocation (Bednar et al., 2013). The social arbiter view takes media coverage as a proxy for corporate reputation, which affects performance and market returns (Bednar, 2012; Deephouse, 2000; Dyck et al., 2008; Joe, 2003; Miller, 2006; Pfarrer et al., 2010; Wartick, 1992). The information intermediary view emphasizes how media coverage legitimate firms by influencing stakeholders’ perceptions (Barber & Odean, 2008; Bushee et al., 2010; Frankel & Li, 2004; Joe et al., 2009; Pollock et al., 2008; Pollock & Rindova, 2003).

Along with large-scale financial budgeting and long event periods, M&A announcements constantly catch the attention of investors and the media (Zaremba & Płotnicki, 2016). Both short-term and long-term performance at the announcement date are widely discussed in the value creation of M&As (Farinós et al., 2020; Latorre et al., 2014). Efficient contracting theory suggests that CEO reputation is positively associated with stock market responses to announcements of capital investments (Jian & Lee, 2011). Liu and McConnell (2013) illustrated that media sentiments can affect managers’ sensitivity to stock price reactions in deciding whether to abandon an acquisition attempt. Chen et al. (2017) demonstrated that negative coverage affects the termination of M&A. Twitter can likewise play a key role in reducing information asymmetry in market reactions to acquisition announcements (Mazboudi & Khalil, 2017). Yang et al. (2019) showed that positive media coverage before M&A predicts stock returns in both short and long run. Gamache and Mcnamara (2019) verified that negative media reaction to M&As will influence the subsequent acquisition activity.

2.2. The form of media bias

Bias deviates from objectivity, accuracy and realism (McQuail, 1992). While Takens et al. (2010) divided bias into issue-based and actor-based biases, Eberl et al. (2017) identified three bias subtypes (visible bias, tonality bias, and agenda bias). Fiske and Taylor (2013) and Swim et al. (2003) addressed blatant bias and subtle bias. Aronson et al. (2015) expounded on the cognitive side of bias, namely, stereotypes, conformity, in-group preference and out-group prejudice.

Political bias and economic benefits are most widely discussed in social media (Gentzkow & Shapiro, 2010; Reuter & Zitzewitz, 2006). Groseclose and Milyo (2005) firstly examined the liberal bias and found that media outlets would slant the coverage to the left. Chiang and Knight (2011) demonstrated that media endorsement influences voters’ behavior. Lott and Hassett (2014) suggested that American newspapers show an obviously positive tendency on Democrats compared with Republicans. Baloria and Heese (2018) noted that firms located in districts with slanted media coverage exposure can suppress negative coverage before the election.

For economic benefits, Reuter and Zitzewitz (2006) verified the positive correlation between mutual fund recommendations and advertising in personal newspapers compared with national newspapers. Gurun and Butler (2012) indicated that the local media uses fewer negative words when reporting on local firms compared with nonlocal firms. Gentzkow and Shapiro (2006, 2010) indicated that publications slant reports toward the beliefs of their audience, and that bias can be alleviated when audiences receive independent evidence.

2.3. Causal attribution of media bias

Despite evidence that independence and accuracy are important determinants of media coverage (Bednar, 2012; Frankel & Li, 2004; Joe et al., 2009), causal attribution of the media is generally underemphasized. Bednar et al. (2015) demonstrated that managers are likely to suffer a severe reputational penalty when evaluators make internal attributions regarding the adoption of poison pill. Jones and Harris (1967) identified antecedent factors that affect evaluators’ attribution of a specific behaviour to one factor over another. Heider (1958) argued the covariation of antecedent factors in causal attribution and explained the formation of bias. For example, stereotypes generate fundamental attribution errors by over-reliance on existing information, which form biases (Kelley & Michela, 1980).

Cultural evidence showed that East Asians make less fundamental attribution errors than Westerners (Choi et al., 2003; Ishii et al., 2003; Miyamoto & Kitayama, 2002; Morris & Peng, 1994). In the Chinese media, the politico-economic trade-off generates market segmentation under the context of fierce competition and strict supervision (Qin et al., 2018). Borochin and Cu (2018) verified that the Chinese media employs few negative words for local and overseas M&A deals. You, Chen, et al. (2018) proved that the Chinese media report more positively on companies in region of prosperous economies, developed systems, and high levels of social trust. You, Zhang, et al. (2018) find that articles from market-oriented media are more critical and accurate than those from state-controlled media.

3. Theory and hypotheses

3.1. Media bias in M&As

Despite the abundant research on media coverage and M&As in efficient markets, the predictive effect of media coverage on capital allocation for inefficient markets as China is relatively vague. Zhang and Su (2015) empirically identified that a strong media governance environment restricts Chinese firms’ overinvestment behaviour. Borochin and Cu (2018) indicated that the predictive power of media coverage on M&A outcomes is found only in non-SOEs.

In developing economies such as China, media coverage is produced under the context of government censorship and vibrant competition (Qin et al., 2018; You, Zhang, et al., 2018). Borochin and Cu (2018) have shown that media coverage of overseas deals is more favourable, which are encouraged by the Chinese government. We extend the concept of stereotype bias and expect that perception of a previous transaction should lead to increased favourable evaluations of the media, especially those encouraged by the government. Thus, one would expect the favourability of an M&A deal’s coverage to increase if the bidder has overseas experiences. Our hypothesis can be summarized as follows:

H1: If the newspaper coverage of Chinese M&As shows stereotype bias, then the media coverage of overseas-experienced deals should be more favourable.

3.2. Media bias under political sensitivity

Previous literature has shown that political connections exert an important influence on firm decisions and performance in China (Claessens et al., 2000; Fan et al., 2007; Gul et al., 2010; Li & Zhou, 2005). The difference between SOEs and non-SOEs lies not only in ownership but also in external governance mechanisms, such as monitoring and takeovers (Li et al., 2011; Liao et al., 2014). Evidence from China also suggests that negative coverage can influence non-SOEs to abandon acquisition attempts by reducing the asymmetry information, which is not found in SOEs (Borochin & Cu, 2018). As SOEs are naturally politically controlled (Aharony et al., 2000; Allen et al., 2005; Sun & Tong, 2003), the political information provided by overseas experience is more effective for the media response regarding to M&As of non-SOEs. Therefore, we expect to observe stereotype bias in non-SOEs compared with SOEs. Our hypothesis can be summarized as follows:

H2: The media coverage of overseas-experienced deals should be more favourable for non-SOEs than for SOEs.

4. Sample and methods

4.1. Sample description

We obtain all M&As announced by Chinese listed companies between January 1, 2008 and December 31, 2018 from the Zephyr Database. The final sample is selected as follows: (1) the acquirer should initially own less than 50% of the target firm’s shares before the acquisition and seek to own more than 50% of the target firm’s shares from the acquisition; (2) the deal value of a transaction should be at least 100 million CNY; and (3) acquirers in financial industry are excluded (two-digit standard industrial classification codes between 60 and 69). Financial records data are obtained from the China Stock Market and Accounting Research (CSMAR) Database. The ultimate controller data are identified from the Sinofin Economic and Financial Database. Our final sample contains a set of 1321 transactions. Table 1 presents the variable definitions.

Table 1. Variable definitions.

Variables	Definition
Media Attention	Number of news articles with bidder and target companies reported by Chinese newspapers from the announcement date to a minimum of +60 days (date withdrawn). News articles are obtained from the China Core Newspaper Full-text Database.
Positive Tone	The proportion of positive words to total word count for each deal analyzed by ROST EA text analysis software.
Positive Stock	Product of Media Attention and Positive Tone.
State Media Attention	Number of deal-specific news articles reported by the China Securities Journal, Securities Daily, Securities Times, and the Shanghai Securities Journal.
State Positive Tone	Proportion of positive words to total words in state-controlled articles analyzed by ROST EA.
State Positive Stock	Product of State Media Attention and State Positive Tone.
Market Media Attention	Number of deal-specific news articles reported by the China Business Journal, First Financial Daily, The Economic Observer, and the 21st Century Business Herald.
Market Positive Tone	Proportion of positive words to total words in market-controlled articles analyzed by ROST EA.
Market Positive Stock	Product of Market Media Attention and Market Positive Tone.
Prior Experience Dummy	A dummy variable that equals 1 if an acquirer has M&A experience before a deal, and 0 otherwise. We manually identify the experience data from the Zephyr database.
Gap Days	Difference between a prior M&A deal and objective M&A deal.
Overseas Experience Dummy	A dummy variable that equals 1 if an acquirer’s prior M&A deal was an overseas transaction, and 0 otherwise.
Prior Deal Size	Natural logarithms of the deal value of a prior M&A deal.
Deal Size	Natural logarithms of the deal value of a transaction.
Completed Dummy	A dummy variable that equals 1 if a transaction is successfully completed, and 0 if the transaction is withdrawn.
Overseas Deal Dummy	A dummy variable that equals 1 if a target firm is a foreign firm, and 0 otherwise.
Diversifying Dummy	A dummy variable that equals 1 if the target and bidder firms are not in the same industry, and 0 otherwise.
SOE	A dummy variable that equals 1 for SOE acquirer status at the end of the fiscal year prior to an acquisition announcement, and 0 otherwise. Ownership data are obtained from the Sinofin Economic and Financial Database.
ROA	Net return on total assets at the end of the fiscal year prior to an acquisition announcement. Financial data are obtained from the CSMAR database.
Board Independence	Ratio of independent board members to total board members at the end of the fiscal year prior to an acquisition announcement.
Management Ownership	Proportion of bidder firm owned by managers and board members at the end of the fiscal year prior to an acquisition announcement.
Leverage	Book value of debt over book value of total assets at the end of the fiscal year prior to an acquisition announcement.
Administrative Expense/Sales	Ratio of administrative expense to sales revenue at the end of the fiscal year prior to an acquisition announcement.
Sales/Assets	Ratio of sales revenue to total assets at the end of the fiscal year prior to an acquisition announcement.
Liquidity	Ratio of current assets to current liabilities at the end of the fiscal year prior to an acquisition announcement.
Sales Expense Growth	Growth rate of selling expense at the end of the fiscal year prior to an acquisition announcement.

Source: Authors formation.

Figure 1 describes the deal value of the M&A transactions in our sample. In 2008, only 65 transactions occurred with an accumulative total deal value of 64.085 billion. The transaction deal value peaked in 2015, with an accumulative total deal value of 340.513 billion for 207 transactions. The average deal value peaked in 2009, indicating that the occurrence of increased block transactions before 2010 and the subsequent rise in the number of transactions.

Figure 1. Sample of evolution of M&A transactions during 2008–2018.

Source: Authors formation.

4.2. Variable construction

Media Coverage. Following Borochin and Cu (2018), Lott and Hassett (2014), and Tetlock (2007), we count the proportion of positive words in a news text to measure the tendency of the media tone. Newspaper articles are collected by searching for target and bidder names in the China Core Newspaper Full-text Database. The sample window is from the date of an acquisition announcement to 60 days after or to the end of the negotiations, whichever comes first (Borochin & Cu, 2018). The number of newspapers constructs the media attention variable Media Attention. We then use the text analysis tool employed by Borochin and Cu (2018) to obtain the proportion of positive words and get the Positive Tone variable,¹ which is bounded between 0 and 1. Then we generate the Positive Stock variable as the product of Media Attention and Positive Tone.

Additionally, we identify eight large financial newspapers and categorize them into state-controlled newspapers (the China Securities Journal, Securities Daily, Securities Times, and the Shanghai Securities Journal) and market-oriented newspapers (the China Business Journal, First Financial Daily, The Economic Observer, and the 21st Century Business Herald) as You, Zhang, et al. (2018) did. The number of articles in the state-controlled newspapers is defined as the State Media Attention, and the proportion of positive words in the state-controlled newspapers is defined as the State Positive Tone. We create the State Positive Stock as the product of State Media Attention and State Positive Tone. Cosh x, we create the Market Positive Stock from market-oriented newspapers.

Overseas Experience Dummy. We identify whether a bidder’s previous M&A transaction is an overseas deal as the proxy index for stereotype bias since 2000.

4.3. Descriptive statistics

Table 2 presents the descriptive statistics of the variables. In the 1321 observations, 678 M&A transactions receive at least one news article. While 81.42% of the transactions attract the attention of state-controlled media, only 28.61% of the transactions capture the attention of market-controlled media. The average positive tone is 44%, with a standard deviation of 15.6%. State-controlled media shows higher positive tone than market-controlled media. As for M&A experience, 46.33% of the M&A deals have an M&A experience while only 7% have overseas M&A experience.

Table 2. Summary statistics.

Variables	N	Mean	Std. Dev.	Min	Max
Media Attention	1321	1.739	4.026	0.000	67.000
Positive Tone	678	0.440	0.156	0.000	0.870
Positive Stock	678	1.535	2.336	0.000	33.172
State Media Attention	1321	0.939	1.804	0.000	20.000
State Positive Tone	552	0.426	0.154	0.000	0.828
State Positive Stock	552	0.972	0.979	0.000	7.608
Market Media Attention	1321	0.214	0.705	0.000	13.000
Market Positive Tone	194	0.378	0.145	0.000	0.800
Market Positive Stock	194	0.554	0.515	0.000	5.287
Prior Experience Dummy	1321	0.463	0.499	0.000	1.000
Gap Days	612	1086.838	953.287	1.000	5726.000
Overseas Experience Dummy	612	0.070	0.256	0.000	1.000
Prior Deal Size	612	6.009	1.170	4.605	11.584
Deal Size	1321	6.162	1.179	4.605	11.584
Completed Dummy	1321	0.918	0.274	0.000	1.000
Overseas Deal Dummy	1321	0.045	0.207	0.000	1.000
Diversifying Dummy	1321	0.108	0.311	0.000	1.000
SOE	1321	0.438	0.496	0.000	1.000
ROA	1321	0.046	0.057	–0.362	0.863
Board Independence	1321	0.369	0.052	0.091	0.667
Management Ownership	1321	0.126	0.197	0.000	0.822
Leverage	1321	0.443	0.240	0.021	5.080
Administrative Expense/Sales	1321	0.103	0.083	0.002	1.068
Sales/Assets	1321	0.671	0.530	0.062	4.928
Liquidity	1321	2.362	3.480	0.103	50.462
Sales Expense Growth	1321	3.531	76.985	–1.000	1975.772

Source: Authors formation.

The deal characteristics show that the average Deal Size is 6.162 million, with a standard deviation of 1.179. Only 4.5% of our sample are cross-border transactions, whereas 10.8% are cross-industry transactions. Only 8.2% of our M&A transactions are withdrawn, and 43.8% of the acquirers’ ultimate controller are SOEs. Prior to an announcement, the average Management Ownership of the acquirer is 12.6%, with a maximum is 82.2%. Moreover, the mean of the leverage debt ratio is 44.3%. The average total asset turnover and selling expense growth are 0.671 and 3.53, respectively.

5. Empirical results

5.1. Univariate analysis of key variables

The univariate test in Table 3 presents that the possibility of prior experience and Gap Days are both higher for the SOE acquirers than for the non-SOE acquirers on average. The deal size of the SOE acquirers are significantly higher than those of the non-SOE acquirers. However, the non-SOE acquirers have a significantly higher return on assets, board independence, and management shareholding ratio. As for operating abilities, the non-SOE acquirers perform better than the SOE acquirers in terms of liquidity. However, the SOE acquirers have a higher total asset turnover ratio compared with the non-SOE acquirers.

Table 3. Univariate test of differences between SOE and non-SOE acquirers.

Variables	Non-SOE acquirers		SOE acquirers		Homogeneity across means
	N	Mean	N	Mean
Positive Stock	369	1.431	309	1.659	–0.227
State Positive Stock	296	0.980	256	0.963	0.017
Market Positive Stock	108	0.556	86	0.551	0.006
Prior Experience Dummy	743	0.380	578	0.571	–0.191***
Gap Days	282	865.106	330	1276.318	–411.212***
Overseas Experience Dummy	282	0.067	330	0.073	–0.005
Deal Size	743	6.053	578	6.303	–0.250***
Completed Dummy	743	0.890	578	0.955	–0.065***
Overseas Deal Dummy	743	0.047	578	0.042	0.006
Diversifying Dummy	743	0.094	578	0.126	–0.032*
ROA	743	0.056	578	0.034	0.021***
Board Independence	743	0.372	578	0.365	0.008***
Management Ownership	743	0.219	578	0.007	0.212***
Leverage	743	0.373	578	0.532	–0.160***
Administrative Expense/Sales	743	0.120	578	0.082	0.037***
Sales/Assets	743	0.600	578	0.761	–0.161***
Liquidity	743	2.959	578	1.595	1.365***
Sales Expense Growth	743	6.094	578	0.237	5.856

Source: Authors formation.

5.2. Regressions of positive media coverage

We evaluate our hypotheses in multivariate linear regression analyses with other factors controlled.

Column 1 of Table 4 reports that the media prefers larger-scale M&A transactions and overseas deals. The bidder characteristics indicate that financial performance (ROA) and sales expense growth are positively associated with positive coverage, whereas the management shareholding ratio is negatively related to positive coverage.

Table 4. Regression of overseas M&A experience on positive coverage.

Independent variables	Total sample		Non-SOE acquirers	SOE acquirers
	(1)	(2)	(3)	(4)
Overseas Experience Dummy		1.155** (2.05)	2.401** (2.62)	0.034 (0.05)
SOE	–0.011 (–0.05)	–0.176 (–0.40)
Deal Size	0.623*** (8.32)	0.630*** (4.89)	0.932*** (3.76)	0.561*** (3.66)
Completed Dummy	–0.382 (–1.27)	–0.886 (–1.65)	–0.596 (–0.93)	–2.169** (–2.13)
Overseas Deal Dummy	0.920** (2.16)	2.224*** (2.73)	9.745*** (5.67)	–0.291 (–0.27)
Diversifying Dummy	–0.449 (–1.12)	–0.019 (–0.03)	–1.138 (–0.87)	–0.447 (–0.44)
ROA	4.029*** (2.71)	9.110** (2.37)	10.040 (1.33)	8.460* (1.75)
Board Independence	3.160** (1.99)	4.571* (1.67)	3.626 (0.83)	6.675* (1.89)
Management Ownership	–0.938* (–1.65)	–1.066 (–0.83)	–2.010 (–1.37)	1.553 (0.31)
Leverage	0.267 (0.71)	2.265* (1.81)	2.397 (1.10)	1.705 (1.03)
Administrative Expense/Sales	–1.323 (–1.06)	0.296 (0.11)	2.329 (0.68)	–4.233 (–0.81)
Sales/Assets	–0.036 (–0.18)	–0.416 (–1.24)	–1.020* (–1.69)	–0.030 (–0.07)
Liquidity	0.025 (0.99)	–0.007 (–0.07)	–0.052 (–0.44)	0.104 (0.37)
Sales Expense Growth	0.002** (2.01)	0.003** (2.53)	0.008*** (5.54)	–0.202 (–0.63)
Year and industry fixed effects	Yes	Yes	Yes	Yes
Constant	–2.135 (–0.90)	–8.685*** (–2.69)	–10.220** (–2.52)	–3.816 (–1.03)
Observations	678	303	129	174
Pseudo-R square	0.226	0.384	0.687	0.366

Note: Regression results of determinants of media positive coverage of Chinese M&A announcements during 2008–2018. Robust standard errors are reported in parentheses; ∗∗∗, ∗∗, and ∗ denote significance at 1%, 5%, and 10% levels respectively.

Source: Authors formation.

The results of this estimation of H1 are given in column 2 of Table 4. The coefficient of the Overseas Experience Dummy is positive and statistically significant (p-value < 0.05). H1 is supported that the media demonstrates stereotype bias by slanting politically supported deal experiences, which is consistent with the conclusion in Borochin and Cu (2018). The significantly positive coefficient of the Overseas Deal Dummy (p-value < 0.01) suggests that the relationship between overseas experience and positive coverage is not based on the transfer of learning.

To assess H2, we estimate the regression by dividing the sample into SOE acquirers and non-SOE acquirers. According to Table 4, the Overseas Experience Dummy is positively associated with Positive Stock (p-value < 0.05) for the non-SOE acquirers other than the SOE acquirers. H2 is verified that the politically supported deal experience adds more information in non-SOEs. The association between deal size and positive coverage are significant in both SOEs and non-SOEs. In the non-SOEs, overseas transactions and high sales expense growth acquirers are positively associated with positive coverage, whereas asset turnover is negatively related to positive coverage. In the SOEs, positive coverage is negatively associated with the possibility whether a transaction is completed, whereas financial performance (ROA) and board independence are positively related to positive coverage.

5.3. Regressions of media attention

We extend our tests by examining the association between overseas M&A experience and media attention.

Table 5 reports that the coefficient of the Overseas Experience Dummy is statistically significant, which exits only for the non-SOE acquirers. The results of media attention also support H1 and H2 and provide further evidence that the media shows stereotype bias in the attribution process.

Table 5. Regression of overseas M&A experience on media attention.

Independent variables	Total sample		Non-SOE acquirers	SOE acquirers
	(1)	(2)	(3)	(4)
Overseas Experience Dummy		1.729** (2.56)	3.736*** (3.62)	0.144 (0.16)
SOE	0.126 (0.47)	0.156 (0.33)
Deal Size	1.234*** (13.45)	1.158*** (7.86)	0.961*** (3.83)	1.211*** (6.44)
Completed Dummy	–1.211*** (–3.07)	–1.676** (–2.42)	–1.634** (–2.00)	–3.677** (–2.44)
Overseas Deal Dummy	0.279 (0.55)	0.917 (1.06)	3.845*** (2.87)	–1.717 (–1.40)
Diversifying Dummy	–0.687* (–1.88)	–0.637 (–0.97)	–0.333 (–0.30)	–1.079 (–1.23)
ROA	6.490*** (3.24)	11.130*** (2.70)	10.400 (1.55)	7.352 (1.29)
Board Independence	5.312*** (2.67)	4.923 (1.47)	1.783 (0.34)	10.370** (2.18)
Management Ownership	–0.892 (–1.34)	–1.864 (–1.40)	–2.368 (–1.57)	6.041 (0.83)
Leverage	0.698 (1.32)	2.476* (1.82)	4.879** (2.02)	–0.476 (–0.25)
Administrative Expense/Sales	0.672 (0.42)	3.766 (1.19)	5.838 (1.45)	–4.123 (–0.71)
Sales/Assets	0.040 (0.17)	–0.002 (–0.00)	–0.511 (–0.62)	0.354 (0.68)
Liquidity	0.009 (0.26)	0.031 (0.24)	0.016 (0.10)	–0.298 (–0.82)
Sales Expense Growth	0.002 (1.62)	0.003* (1.90)	0.006*** (3.06)	–0.241 (–0.78)
Year and industry fixed effects	Yes	Yes	Yes	Yes
Constant	–7.720*** (–2.74)	3.186 (0.63)	–13.480** (–2.53)	–9.742* (–1.92)
Observations	1321	612	282	330
Pseudo-R square	0.230	0.301	0.447	0.290

Note: Regression models of the determinants of media attention of Chinese M&A announcements during 2008–2018. Robust standard errors are reported in parentheses; ∗∗∗, ∗∗, and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.

Source: Authors formation.

Interestingly, the large-scale deals and those that are highly likely to be withdrawn receive more media attention in both subsamples. As for the acquirer characteristics, the overseas deals of the non-SOE acquirers attract more media attention. The leverage debt ratio and selling expense growth are positively associated with media attention for the non-SOE acquirers, whereas the board independence of the SOE acquirers is positively related to media attention.

5.4. Interaction effect of deal size

We further explore the interaction of overseas experience and deal size of the previous transaction on both positive coverage and media attention.

Table 6 indicates that the interaction term of overseas experience and previous deal size is positive and statistically significant (p-value < 0.01). Paradoxically, the deal size of a previous transaction is negatively associated with positive coverage. The result of the interaction term also indicates that large-scale overseas M&A experience receives more favourable coverage, which is consistent with our assumption. Subsample results show that the coefficient of the interaction term between overseas experience and prior deal size is significantly positive in the non-SOE acquirers compared with the SOE acquirers. This finding suggests that the amplify effect of Prior Deal Size exits only for the non-SOE acquirers.

Table 6. Interaction effect of deal size and overseas experience on positive coverage.

Independent variables	Total sample			Non-SOE acquirers	SOE acquirers
	(1)	(2)	(3)	(4)	(5)
Prior Deal Size	–0.142 (–1.03)	–0.123 (–0.90)	–0.253* (–1.79)	–0.028 (–0.12)	–0.360* (–1.98)
Overseas Experience Dummy		1.121** (1.99)	–7.843*** (–2.67)	–13.220** (–2.12)	–2.027 (–0.54)
Overseas Experience Dummy× Prior Deal Size			1.505*** (3.10)	2.639** (2.53)	0.319 (0.53)
SOE	–0.158 (–0.36)	–0.150 (–0.34)	–0.136 (–0.32)
Deal Size	0.670*** (4.98)	0.662*** (4.95)	0.672*** (5.11)	0.738*** (2.89)	0.659*** (4.11)
Completed Dummy	–0.884 (–1.63)	–0.890* (–1.65)	–0.923* (–1.75)	–0.650 (–1.04)	–2.187** (–2.16)
Overseas Deal Dummy	2.331*** (2.85)	2.236*** (2.75)	2.295*** (2.87)	7.958*** (4.40)	–0.121 (–0.11)
Diversifying Dummy	0.005 (0.01)	–0.038 (–0.05)	–0.033 (–0.04)	–0.770 (–0.60)	–0.376 (–0.38)
ROA	10.210*** (2.66)	9.270** (2.41)	7.481* (1.95)	8.733 (1.18)	8.876* (1.82)
Board Independence	5.396* (1.96)	4.766* (1.73)	4.721* (1.75)	3.218 (0.75)	7.138** (2.03)
Management Ownership	–1.188 (–0.92)	–1.006 (–0.78)	–1.235 (–0.98)	–2.106 (–1.48)	1.540 (0.31)
Leverage	2.161* (1.72)	2.194* (1.75)	1.617 (1.30)	1.834 (0.85)	1.377 (0.83)
Administrative Expense/Sales	0.270 (0.10)	0.328 (0.12)	–0.153 (–0.06)	2.419 (0.72)	–4.834 (–0.91)
Sales/Assets	–0.385 (–1.14)	–0.417 (–1.24)	–0.383 (–1.16)	–0.924 (–1.57)	–0.036 (–0.08)
Liquidity	–0.031 (–0.31)	–0.013 (–0.13)	–0.030 (–0.30)	–0.054 (–0.47)	0.033 (0.11)
Sales Expense Growth	0.002** (2.29)	0.003** (2.44)	0.002** (2.46)	0.006*** (3.93)	–0.107 (–0.33)
Year and industry fixed effects	Yes	Yes	Yes	Yes	Yes
Constant	–3.294 (–0.96)	–3.068 (–0.90)	–1.451 (–0.43)	–7.749* (–1.80)	–0.750 (–0.19)
Observations	303	303	303	129	174
Pseudo-R square	0.376	0.386	0.409	0.711	0.386

Note: Regression models of the interaction effect of prior deal size and overseas experience on positive coverage related to Chinese M&A announcements during 2008–2018. Robust standard errors are reported in parentheses; ∗∗∗, ∗∗, and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.

Source: Authors formation.

According to Table 7, the interaction effect of prior deal size and overseas experience is positively correlated with media attention (p-value < 0.01) and the amplify effect is valid only for non-SOEs. For non-SOE acquirers, media would show stereotype bias toward overseas-experienced bidders and the effect would be considerably impressive for large-scale transactions.

Table 7. Interaction effect of deal size and overseas experience on media attention.

Independent variables	Total sample			Non-SOE acquirers	SOE acquirers
	(1)	(2)	(3)	(4)	(5)
Prior Deal Size	–0.011 (–0.07)	0.005 (0.03)	–0.142 (–0.91)	0.075 (0.30)	–0.227 (–1.07)
Overseas Experience Dummy		1.730** (2.56)	–11.860*** (–3.29)	–20.800*** (–3.57)	–1.868 (–0.37)
Overseas Experience Dummy× Prior Deal Size			2.314*** (3.83)	4.072*** (4.26)	0.326 (0.38)
SOE	0.147 (0.31)	0.155 (0.33)	0.254 (0.55)
Deal Size	1.175*** (7.80)	1.157*** (7.71)	1.154*** (7.79)	0.883*** (3.61)	1.254*** (6.51)
Completed Dummy	–1.692** (–2.43)	–1.677** (–2.42)	–1.779*** (–2.60)	–1.985** (–2.51)	–3.684** (–2.44)
Overseas Deal Dummy	0.926 (1.06)	0.919 (1.06)	0.966 (1.13)	3.534*** (2.73)	–1.674 (–1.36)
Diversifying Dummy	–0.685 (–1.04)	–0.637 (–0.97)	–0.678 (–1.04)	–0.373 (–0.34)	–0.994 (–1.13)
ROA	11.790*** (2.85)	11.130*** (2.70)	9.751** (2.38)	9.468 (1.46)	7.573 (1.31)
Board Independence	5.294 (1.57)	4.922 (1.47)	4.518 (1.36)	1.854 (0.37)	10.140** (2.12)
Management Ownership	–1.974 (–1.47)	–1.865 (–1.40)	–2.012 (–1.53)	–2.697* (–1.86)	5.489 (0.75)
Leverage	2.389* (1.74)	2.476* (1.81)	1.740 (1.28)	3.190 (1.35)	–0.602 (–0.31)
Administrative Expense/Sales	3.615 (1.13)	3.768 (1.19)	2.981 (0.95)	5.398 (1.40)	–4.545 (–0.78)
Sales/Assets	0.0616 (0.15)	–0.002 (–0.00)	0.030 (0.07)	–0.601 (–0.76)	0.350 (0.67)
Liquidity	0.011 (0.09)	0.031 (0.24)	0.018 (0.14)	0.003 (0.02)	–0.316 (–0.87)
Sales Expense Growth	0.003* (1.72)	0.003* (1.90)	0.003* (1.91)	0.005** (2.39)	–0.223 (–0.71)
Year and industry fixed effects	Yes	Yes	Yes	Yes	Yes
Constant	3.083 (0.60)	3.160 (0.61)	4.878 (0.96)	–11.570** (–2.15)	–8.249 (–1.57)
Observations	612	612	612	282	330
Pseudo-R square	0.292	0.301	0.319	0.493	0.293

Note: Regression models of the interaction effect of prior deal size and overseas experience on media attention. Robust standard errors are reported in parentheses; ∗∗∗, ∗∗, and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.

Source: Authors formation.

5.5. Comparison between state-controlled media and market-oriented media

Table 8 illustrates that the market-oriented media pays more attention to the overseas-experienced deals than the state-controlled media. Evidence also reveals that the coefficient of overseas experience is positive and statistically significant for the non-SOE acquirers compared with the SOE acquirers in both subsamples, as stated in H2. The results show that the market-oriented media exhibits more stereotype bias than the state-controlled media. Considering You et al. (2018b) who find that market-oriented media have significant corporate governance impact, the empirical finding of invisible bias in market-oriented media provides new insights in understanding the information efficiency in the Chinese market.

Table 8. Comparison between state-controlled media and market-oriented Media.

Independent variables	State-controlled media attention			Market-oriented media attention
	Total sample	Non-SOE acquirers	SOE acquirers	Total sample	Non-SOE acquirers	SOE acquirers
	(1)	(2)	(3)	(4)	(5)	(6)
Overseas Experience Dummy	0.303 (1.14)	0.603* (1.79)	–0.058 (–0.14)	0.354*** (2.96)	0.842*** (4.08)	–0.015 (–0.11)
SOE	0.219 (1.19)			–0.033 (–0.40)
Deal Size	0.483*** (8.33)	0.394*** (4.81)	0.530*** (6.23)	0.158*** (6.07)	0.142*** (2.83)	0.155*** (5.58)
Completed Dummy	–1.094*** (–4.01)	–0.910*** (–3.42)	–2.579*** (–3.78)	–0.153 (–1.24)	–0.145 (–0.89)	–0.298 (–1.34)
Overseas Deal Dummy	0.011 (0.03)	0.523 (1.20)	–0.493 (–0.89)	0.150 (0.98)	0.853*** (3.19)	–0.433** (–2.39)
Diversifying Dummy	–0.422 (–1.63)	–0.279 (–0.77)	–0.627 (–1.58)	–0.016 (–0.14)	0.095 (0.42)	–0.196 (–1.52)
ROA	3.427** (2.12)	4.109* (1.88)	2.004 (0.78)	1.277* (1.75)	0.838 (0.62)	0.825 (0.98)
Board Independence	–0.075 (–0.06)	–0.769 (–0.45)	1.965 (0.91)	0.915 (1.54)	0.621 (0.59)	1.891*** (2.69)
Management Ownership	–0.437 (–0.83)	–0.522 (–1.06)	2.586 (0.79)	–0.301 (–1.28)	–0.399 (–1.32)	0.308 (0.29)
Leverage	1.005* (1.87)	2.119*** (2.69)	0.152 (0.17)	0.407* (1.69)	0.786 (1.63)	–0.128 (–0.45)
Administrative Expense/Sales	2.225* (1.78)	3.419*** (2.61)	–1.308 (–0.50)	0.221 (0.39)	0.338 (0.42)	–0.556 (–0.65)
Sales/Assets	–0.0520 (–0.32)	–0.299 (–1.11)	0.125 (0.53)	0.017 (0.23)	–0.039 (–0.24)	0.051 (0.67)
Liquidity	0.011 (0.22)	0.023 (0.46)	–0.050 (–0.30)	0.024 (1.06)	0.020 (0.65)	–0.028 (–0.53)
Sales Expense Growth	0.002*** (2.89)	0.002*** (3.50)	–0.174 (–1.24)	0.0004 (1.41)	0.001*** (2.75)	–0.058 (–1.28)
Year and industry fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Constant	–0.312 (–0.16)	–2.987* (–1.72)	–2.549 (–1.11)	–2.041** (–2.26)	–2.846*** (–2.67)	–1.179 (–1.58)
Observations	612	282	330	612	282	330
Pseudo-R square	0.304	0.483	0.273	0.269	0.434	0.246

Note: Regression models of the effect of overseas experience on media attention by state-controlled media and market-oriented media. Robust standard errors are reported in parentheses; ∗∗∗, ∗∗, and ∗ denote significance at the 1%, 5%, and 10% levels, respectively.

Source: Authors formation.

5.6. Robustness test

To address the problem of strategic selection, we use propensity score matching models (PSM) to match treated and control samples on their observable. We compare overseas-experienced deals with a propensity score-matched control sample of M&A deals without overseas experience for the standard nearest-neighbour matching estimator (1:1). Results indicate an average treatment effect of future potential of 2.93 (p < 0.5) when the dependent variable is media attention. Table 9 demonstrates that most characteristics are insignificantly different after matching. Media attention is significantly higher for overseas-experienced deals than deals without overseas experience even after controlling for the characteristics in the PSM model. These findings suggest that our main results are robust to self-selection concerns.

Table 9. PSM estimation.

Variables	Panel A: Before matching overseas experience					Panel B: After matching overseas experience
	Without overseas experience		With overseas experience		Homogeneity across means	Without overseas experience		With overseas experience		Homogeneity across means
	N	Mean	N	Mean		N	Mean	N	Mean
Media Attention	569	1.656	43	3.558	−1.903***	40	0.725	40	2.100	−1.375**
Positive Stock	277	1.559	26	2.890	−1.331**	14	0.833	23	1.791	−0.958
State Positive Stock	220	0.973	18	1.411	−0.438*	11	0.792	16	1.128	−0.336
Market Positive Stock	77	0.529	10	0.980	−0.451**	5	0.395	9	0.502	−0.107
Prior Experience Dummy	569	1.000	43	1.000	0.000	40	1.000	40	1.000	0.000
Gap Days	569	1091.476	43	1025.465	66.011	40	907.425	40	998.050	−90.625
Deal Size	569	6.244	43	6.409	−0.165	40	5.816	40	6.282	−0.466*
Completed Dummy	569	0.924	43	0.930	−0.006	40	0.950	40	0.925	0.025
Overseas Deal Dummy	569	0.040	43	0.070	−0.029	40	0.050	40	0.050	0.000
Diversifying Dummy	569	0.102	43	0.070	0.032	40	0.125	40	0.075	0.050
SOE	569	0.538	43	0.558	−0.020	40	0.450	40	0.575	−0.125
ROA	569	0.042	43	0.060	−0.019**	40	0.063	40	0.052	0.011
Board Independence	569	0.367	43	0.372	−0.005	40	0.375	40	0.366	0.009
Management Ownership	569	0.093	43	0.072	0.021	40	0.111	40	0.076	0.035
Leverage	569	0.488	43	0.471	0.017	40	0.418	40	0.469	−0.050
Administrative Expense/Sales	569	0.098	43	0.081	0.017	40	0.094	40	0.084	0.010
Sales/Assets	569	0.649	43	0.914	−0.264***	40	0.708	40	0.808	−0.100
Liquidity	569	1.805	43	1.619	0.186	40	1.779	40	1.626	0.153
Sales Expense Growth	569	7.267	43	0.328	6.939	40	0.236	40	0.324	−0.088

Source: Authors formation.

6. Conclusions

One assumption of media’s monitoring role in M&A transactions is that the media is impartial. However, bias is undoubtedly widespread in media coverage. Based on attribution framework, we test the idea that the media demonstrates stereotype bias and memorizes the impression formed by bidders’ previous transactions.

Using a measure of positive coverage for Chinese M&A announcements, we find that overseas experience is significantly correlated with positive coverage. We interpret this finding to mean that stereotype bias is shaped when information is excessive in the attribution process of the media. Moreover, the correlation between overseas experience and positive coverage exists only in non-SOE acquirers. This finding verifies that political impression only provides excess information to non-SOE acquirers compared with SOE acquirers. Results also verify that market-oriented media significantly pays more attention to overseas-experienced deals than state-controlled media, which reinforces the evidence provided by You, Zhang, et al. (2018). Our results strongly suggest that stereotype bias slant media coverage in M&A transactions. Future studies will consider other types of biases, such as conformity or out-group prejudice.

Disclosure statement

No potential competing interest was reported by the authors.

Additional information

Funding

This work was supported by the [Humanities and Social Science Fund of Ministry of Education of China] under Grant [number 18YJC630195]; and the [National Natural Science Foundation of China] under Grant [number 72002064; 71802041; 71801034].

Notes

1 The Chinese text-mining software ROST EA, which was developed by Professor Yang Shen and his team at Tsinghua University, is widely used for text analysis, webpage crawling, news analysis, online public opinion, micro blogs, and so on.