Does Information Blocking on Internet Matter Corporate Over-Investment: Evidence from a Quasi-Natural Experiment in China

ABSTRACT

Focusing on Google’s withdrawal from mainland China as a unique setting, this study investigates the impact of information blocking on corporate over-investment. Using a sample of Chinese firms during the period of 2007 to 2014 and employing a difference-in-difference (DID) approach, our findings reveal that corporate over-investment is significantly higher for the treatment firms (firms with high international sales) in the post-period of Google’s withdrawal than that in the pre-period of Google’s withdrawal, suggesting that overseas information blocking boosts corporate over-investment for firms with more global business. Moreover, the positive relation between Google’s exit and firm-level over-investment is less pronounced for cross-listed firms than for non-cross-listed firms, implying that cross listing plays a crucial role in breaking overseas information blockade on internet search for domestic investors. Above findings are still valid after using alternative measures to capture over-investment and the treatment group. Mechanism analysis and heterogeneity tests show that Google’s exit may exacerbate agency conflicts and thus leads to an increase in corporate over-investment. Our study provides an important reference for the understanding of the roles of internet search in information transmission and corporate governance.

KEYWORDS:

• Google’s exit
• information blocking
• cross listing
• over-investment
• China

JEL:

• D83
• G11
• G34

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1. As Xu, Xuan, and Zheng (2021) indicate, Google’ s withdrawal was a surprise to the market because top management team (TMT) of Google always declared to unceasingly defend its rights and protect employees in China before the exit.

2. Please refer to the website: http://tech.sina.com.cn/i//1055707942.shtml.

3. As a report issued by CNNIC in 2009 revealed, about 91% of internet users claimed that they employed search engines to find information at least once a month and about 60% of them used search services every day. There was an increasing tendency in the proportion of search engine users on the Internet in mainland China. According to this report, in China, Baidu held the largest market share, about 69.9%, and Google was the second largest search engine providers, owned about 19.8%. Please refer to http://tech.sina.com.cn/i//15003616472.shtml.

4. Internet users in mainland China are able to visit several blocked foreign websites and google services through virtual private networks (VPNs). However, this approach is unavailable for most individual users. First, duo to tight regulations on the Internet, it is very difficult for individuals to contact with Internet services providers who are permitted to offer VPNs for bypassing the censorship and GFW. Second, unauthorized VPN services are illegal in mainland China (see https://www.scmp.com/news/china/policies-politics/article/2064587/chinas-move-clean-vpns-and-strengthen-great-firewall), and thus it is unstable to visit blocked foreign websites through unauthorized VPNs. That is, it is infeasible for users in mainland China to purchase unauthorized VPNs without any legal risk (see https://www.bbc.com/news/technology-41160383). Third, there are technical barriers for most individual users to visit blocked foreign websites through unauthorized VPNs effectively, and thus the usefulness of unauthorized VPNs is greatly restricted by the technical limitations. Fourth, unauthorized VPNs are more likely to bring about several cybersecurity threats and vulnerabilities in its operation, such as cyberattack, internet worms, privacy breach, and DNS Hijacking. As a result, individual users are more likely to abandon unauthorized VPNs out of regard for cybersecurity. Finally, the Chinese government claims to provide authorized VPN services for users who need to visit blocked foreign websites, but authorized VPNs do not open to the public and are only authorized for a few agencies (e.g., universities, government agencies, research institutions and international companies). Moreover, the payment of authorized VPNs is very expensive for individual users. We contacted with an authorized VPN provider in mainland China and knew that the fee of authorized VPN services is about 80,000 RMB per month in 2021. Notably, according to the Chinese National Bureau of Statistics, annual income per capita in China is about 32,189 RMB in 2020. Accordingly, authorized VPNs are substantially unavailable for the vast majority of individual users in mainland China.

5. According to the 2009 shanghai stock exchange statistics annual, the proportion of individual investors was about 99.76%, and the value of shares held by individual investors was about 13,486.39 million RMB, which accounted for 42.33% of total market value. As a comparison, the proportion of institutional investors was only about 0.08%, but the value of shares held by institutional investors was 8380.21 million RMB and accounted for about 26.24% of total market value. Above statistics suggest that individual investors are an important force in Chinese markets to play a vital role in corporate governance (see: http://www.sse.com.cn/aboutus/publication/yearly/documents/c/tjnj_2009.pdf).

6. Of course, companies are able to increase their disclosures to reduce information asymmetry among shareholders. And the prerequisite is that there are no misguidances in disclosures. Notably, Wang, Yu, and Zhang (2022) reveal that Chinese firms report their foreign transactions more optimistically after Google’s withdrawal. In this case, the disclosures of Chinese companies may generate more biases and misguidances after Google’s exit and therefore is more likely to increase information asymmetry between managers (large shareholders) and owners (minority shareholders).

7. Enikolopov, Petrova, and Sonin (2018) find a negative relation between negative blog posts and stock returns and show that within a day, the CARs of focal firms gradually decrease in the next seven hours after negative postings appear on social media and the average daily CARs after blog posts reduce 0.33% point. Moreover, Wu (2016) finds that investor sentiment on social media is significantly associated with future stock returns in Chinese capital markets. Furthermore, Wu (2016) shows that the relationship between negative investor sentiment and future stock returns is stronger than that between positive investor sentiment and future stock returns, suggesting that negative postings on social media usually attract more public attention comparing with positive postings.

8. In 2008, a great earthquake broke out in Sichuan province. Immediately, the chairman of Vanke (stock code: 000002), a famous listed company in mainland China (the market value of equity was about 280 billion RMB in 2007), claimed that Vanke donated two million RMB to reconstruct disaster area. However, many people believed that the amount of the philanthropic giving of Vanke did not fit well with its market value and social responsibilities and thus expressed their displeasures through social media. Under strong public pressures, Vanke, subsequently, had to donate 100 million RMB additionally. This anecdotal evidence in China supports the view that negative postings issued by social media users can affect managerial decisions by exerting strong external pressures.

9. It seems that the impacts of Google’s exit on corporate information environment are parallel and homologous. However, firms with more global businesses require and generate more oversea information than those with less global businesses. Clearly, getting oversea information relies more on Google searching. In this case, it is natural for scholars to expect that the impacts of Google’s exit on information environment are heterogeneous between above two kinds of firms.

10. Following Xu, Xuan, and Zheng (2021), we employ INVTA_RES and INVRV_RES into the propensity score model. Moreover, we only use corporate fundamentals variable in the first stage of PSM and obtain consistent results.

11. We conduct placebo tests using estimated oversea investments, calculated by the model: INV=α₀+α₁SALE_t+α₂ASSETS_t−1+ε, where INV stands for capital expenditures. SALE is sales revenue. ASSETS is total assets. The non-tabular results show that the likelihood of significant coefficients on TREAT×POST is less than 5%. The distributions of t-values reveal that the biases in identification in the original manuscript may be driven by incorporating domestic investments into DID design. Moreover, we sort the placebo coefficients on TREAT×POST from low to high and find that the rank of the coefficients on TREAT×POST (0.0184, 0.0346) in the main tests is 9997 and 9564, respectively, suggesting that the likelihood of generating similar coefficients in a stochastic process is less than 5%.

Additional information

Funding

We appreciate constructive comments from Shaojuan Lai, Jianying Weng, Ying Zhang and participants of our presentations at Xiamen University. Prof. Du acknowledges financial support from the National Social Science Fund of China [20&ZD111], and Prof. Chang acknowledges financial support from the Fujian Province Social Science Planning Project of China [FJ2022C053].