Institutional distance, geographic distance, and Chinese venture capital investment: do networks and trust matter?

In our context, network strength is defined as a VC firm’s centrality in the Chinese VC syndication network, following El-Khatib et al. (2015). Trust is defined as the subjective belief about the extent that a target investee will perform as planned, similar to Bottazzi et al. (2016). Please see Section 3 for detailed definitions.

China’s Sixth Population Census in 2010 showed the Han Chinese are 91.60% of the population (www.​stats.​gov.​cn). According to the Ministry of Education of the People’s Republic of China, Mandarin use was nearly 80% nationwide in September 2019 (www.​moe.​gov.​cn).

In China, the difference between venture capital and private equity is ambiguous, and the terms VC and PE are often used interchangeably. Throughout this paper, we refer to such investments as VC.

For example, the provincial administrative region of Guangdong has an area of 179,800 km², a population of 126 million in 2020 and a GDP of 1.7 trillion USD in 2020. In comparison, Italy has an area of 301,230 km², a population of 59 million in 2020 and a GDP of 1.9 trillion USD in 2020. Due to data limitations, Hong Kong, Macau, and Taiwan are not in the scope of our analysis. The 31 provincial administrative regions in mainland China are Anhui, Beijing, Chongqing, Fujian, Gansu, Guangdong, Guangxi, Guizhou, Hainan, Hebei, Heilongjiang, Henan, Hubei, Hunan, Inner Mongolia, Jiangsu, Jiangxi, Jilin, Liaoning, Ningxia, Qinghai, Shaanxi, Shandong, Shanghai, Shanxi, Sichuan, Sinkiang, Tianjin, Tibet, Yunnan, and Zhejiang.

For example, see a recent survey by Servaes and Tamayo (2017).

See this report from http://​startupsusa.​org/​global-startup-cities/​.

For example, the Proposal of Speedily Developing Chinese Venture Capital in 1998, the Interim Measures for the Management of Venture Capital Firms in 2005, the Guidance on the Normative Establishment and Operation of Venture Capital Guidance Funds in 2008, and the Interim Measures for the Supervision and Administration of Private Equity Investment Funds in 2014.

Grilli et al. (2019) provide a comprehensive review of institutional and related determinants of VC activity. The authors consider both formal and informal institutions which have been found to affect VC activity in the literature. Formal institutions include regulatory institutions, government quality, and financial market conditions, while informal institutions include entrepreneurialism, other cultural attitudes, and social capital. As well as formal and informal institutions, other determinants of VC activity include technological activities and macroeconomic conditions. In addition, VC firm characteristics such as VC firm age and investment experience affect their activity (e.g., Croce et al., 2019; Cumming and Dai, 2010). In the context of China, the types of VC firms also play a role in VC activity (e.g., Humphery-Jenner and Suchard, 2013; Suchard et al., 2021).

In the process of transitioning from a planned economy toward a market economy over the past decades, the pace of marketization across Chinese provinces exhibits great disparity (Wang et al., 2019). The central government allowed experiments of new economic policies in certain provinces. For example, the government created the special economic zones (SEZs) of Shenzhen, Zhuhai, and Shantou in Guangdong province and Xiamen in Fujian province in the early 1980s to attract foreign capital by exempting them from taxes and regulations. The experiment was later expanded to cover other coastal provinces, and the success of the reform encouraged many other peer provinces to subsequently implement similar economic policies. Gradually, the coastal provincial regions in China (e.g., Zhejiang, Shanghai, and Guangdong) have experienced tremendous economic growth and developed relatively advanced legal, financial, technological, government, and market institutions. By contrast, economic growth in the inland provincial regions of China is much slower and their institutional environments remain relatively weak and underdeveloped.

This consideration is similar in spirit to the argument proposed by Shenkar (2001) that cultural distance is not a symmetric construct, because home culture is embedded in a firm, while host culture is a national environment. The same argument applies to Chinese venture capital investments.

They do so by serving as board members in the portfolio companies and getting actively involved in the professionalization of these companies, such as replacing the founder with an outside CEO, recruiting managers and outside directors, and so on (e.g., Baker and Gompers, 2003; Hellmann and Puri, 2002; Lerner, 1995).

By analyzing individuals’ angel investment decisions across 25 countries, Ding et al. (2015) show that investors from countries with a high level of trust tend to have a higher perception of entrepreneurial skills and therefore are more likely to make angel investments. Similarly, Bottazzi et al. (2016) examine the effect of trust on European venture capital deal formation. They argue that a higher degree of trust can encourage VC firms to invest and increase investors’ valuation to outbid their competitors.

For example, the trust score in our sample for the VC investment analysis has a mean of 0.064, while its standard deviation is relatively large at 0.142 (panel A of Table 1).

We obtain VC data from https://​www.​pedata.​cn, a product developed by Zero2IPO.

Since our empirical design requires that VC firms are located in mainland China, we remove the transactions where VC firms’ headquarters are overseas.

In this study, the CPMI for 2008–2016 is calculated with the year 2008 as the base year. In 2008, each component of the CPMI among provinces ranges from 0 to 10 according to their relative level of marketization. To enable the cross-year comparability of marketization in the period of 2008–2016, each component of subsequent years is based on the year 2008 and is allowed to exceed 10 or lower than 0. Therefore, from 2008 to 2016, the CPMI can reflect the rise or decline of the degree of marketization in each province. More details about the CPMI can be found in Wang et al. (2019).

The data on CPMI for 2008–2016 are retrieved from Wang et al. (2019).

We obtain longitudes and latitudes of Chinese cities from http://​www.​gpsspg.​com/​maps.​htm.

In this questionnaire, the question related to trust is “According to your experience, which five provinces have the most trustworthy enterprises? Please list them in order.” The province which ranks the first is assigned a score of 5; the province which ranks the second is assigned 4, and so on. The generalized trust of province i towards province j (note that the case of i = j is possible in the trust data) is the weighted average of the scores, where the weights are the fraction of entrepreneurs in province i who regard province j as first-ranking, second-ranking trustworthy, and so on. For example, Beijing is ranked number one by 1.8%, number two by 2.2%, number three by 0.9%, number four by 0%, and number five by 0.5% of the responding entrepreneurs in Tianjin. Therefore, the trustworthiness of Beijing is 21% (1.8% × 5 + 2.2% × 4 + 0.9% × 3 + 0% × 2 + 0.5% × 1) in the view of Tianjin’s entrepreneurs.

Nevertheless, for robustness checks, we also use two alternative measures as proxies for trust. The description of these alternative measures and the associated results are presented in Section 5.3.

Although we focus on VC firms headquartered in mainland China, some VC firms do obtain their funding from overseas sources. To control for the potential effect of foreign-backed VC firms, following Humphery-Jenner and Suchard (2013), we include Capital Source Fixed Effects.

Since the cases where a VC firm invests in a province during a year (i.e., the cases where the variable VC investment equals 1) constitute only 2.97% of the VC firm–destination province–year sample, as a robustness check, we also use the rare event logit model (King and Zeng, 1999) in the VC investment analysis. The results, reported in Appendix Table A5, remain qualitatively similar to our main findings.

For robustness, we alternatively cluster standard errors at VC firm–destination province level. Such clustering controls for the potential time series correlation among a VC firm’s investments in a certain target province. The results are still statistically significant as presented in Appendix Table A6.

We apply the discrete-time survival model rather than the widely used Cox (1972)’s proportional hazard model (hereafter Cox PH model) for several reasons. First, the Cox PH model is for continuous time data where the time for event occurrence can take on any nonnegative values. However, in the VC context, the exit time is often measured discretely, in days, months, or years. Second, the Cox PH model assumes that the effect of an explanatory variable on the chance of an event occurrence is unchanged over time, which is often unrealistic (Singer and Willett, 1993). We test the proportional hazard assumption behind the Cox model using our sample on successful VC exits, and the results indicate non-proportional hazards.

For example, if a VC–portfolio company pair experiences a successful exit event at year 5, then there will be 5 VC–portfolio company–year observations for that VC–portfolio company pair in the sample. For the fifth observation, the dependent variable, Success, equals 1. For the other four observations, Success equals 0. For those pairs that experience an unsuccessful exit (e.g., liquidation) or never have an exit event, Success equals 0 from the investment year to the exit year or to year 2018.

Results are almost the same if we use a conditional logit model, whose results are reported in Appendix Table A9.