1Robin Li, founder of China’s dominant internet search engine Baidu, and Hugo Shong, a prestigious venture capitalist and founding partner of IDG Capital Partners, are two examples of numerous successful “sea turtles”.
2See for example “Sea turtles reverse China’s brain drain” by Jaime FlorCruz, CNN World, October 28, 2010, http://edition.cnn.com/2010/WORLD/asiapcf/10/28/florcruz.china.sea.turtles.overseas/index.html?hpt=C1; and “Sea turtles are dead”, NetEase News, December 26, 2011, http://news.163.com/special/reviews/overseasreturnee.html.
3For example, the “1000 Talents” program targets patent holders and high achievers in academia. Upon their return, senior scholars, for instance, can earn several times higher than what local faculty make in addition to other benefits (eg. generous relocation payment, social security benefits, and better education opportunities for their children). For those who are more likely to join the non-state business sector or run their own private business, the state establishes a number of business parks and offers special policies such as tax breaks (Pan 2010).
4See Kortum and Lerner (2000) for a discussion on the contribution of the VC investment to the innovative activities in the US.
5The sum of IPOs and M&As far exceeds the number of VC deals, because most of the Chinese companies acquiring IPO or M&A are not actually VC-backed.
6The Chinese VC industry has a much smaller scale of economy compared to the US. According to the National Venture Capital Association Yearbook 2011, there have been in the US a total of 25,213 VC-backed companies, 73,640 VC deals, 2,984 IPOs and 4,961 M&As between 1985 and 2010.
7Often times, venture capitalists from different VC firms form a syndicate and invest in a company together. This is a good way to disperse risks and share each other’s unique resources.
8Studying the Chinese VC industry, this paper also contributes to the literature that focuses on the determinants of VC performance. Networks (Hochberg et al. 2007) and reputation (Nahata 2008) of the VC firms, human capital characteristics of the venture capitalists (Knockaert et al. 2006; Dimov and Shepherd 2005) and of the entrepreneurs (Wang and Wang 2011) are shown to be important factors.
9It is not clear to me whether there is selectivity among the 1,304 VC deals with educational information on their venture capitalists. In China’s Venture Capital Yearbook 2009, overseas Chinese and internationals constitute 9.4% of the overall population of venture capitalists, however, the comparable percentage in CVSource is 19.5%. There is no information available regarding return migrants in the yearbook, but it is possible that my current sample over-represents the foreign-educated personnel.
10In an earlier draft, I used the four-group categorization for the statistical analysis, which yields qualitatively and quantitatively similar results.
11According to the CVSource criteria, the development stage is defined to be when products and services are being developed, potential customers are found, technology risk has declined, whereas no revenues have yet occurred. The start-up stage before development is when only a concept or business plan has been produced, and expansion and late stage are when market are being expanded and profits have been observed respectively.
12The sale of a company in its early stages to another VC firm or investor. Slightly different from Gompers and Lerner (2000), Hochberg et al. (2007) and Nahata (2008), trade sale is considered another type of success in this paper, in that it also enables VC firms to exit (although with a modest return) and get them a chance to access other promising projects.
13Hochberg et al. (2007) show that most of the profits VC firms obtain are from investments that eventually exit, which are only a subsample of their total invested projects.
14
ExitRatio is likely endogenous, but removing it from the regression does not change the result at all.
15The full dataset includes 2,865 individual observations in total, and 887 of them have non-missing values for all of the variables
.16Although detailed data on venture capitalists’ previous work experience are not present for everyone, I found from the available evidence that many of the domestic venture capitalists formerly worked for government agencies or state-owned companies, while the foreign-educated group did not share this feature.
17See Breznitz and Murphree (2011), p.12, for a discussion on “structured uncertainty” in the Chinese society.
18I also tried interacting Foreign with Offshore and JointVenture to see if foreign educated venture capitalists perform uniformly worse in domestically funded and foreign funded VC firms. This corresponds to the conjecture that they play different roles in different types of VC firms (for example, the main task of foreign educated venture capitalists in domestic VC firms is to cooperate with international partners or clients using their language proficiency and knowledge of the global market, rather than to bring VC-backed companies to the public market.), or that they are misallocated across firms. However, there is no sign showing that these effects are strong.
19In this question, the dataset is expanded by 2, meaning one original observation becomes two identical observations in the new regression. The failure type indicator is set to be 1,0 for the two observations if they originate from a “fair exit” case; 0,1 if from a “good exit” case; and 0,0 if censored in the original regression.
20To the extent that government agencies, research institutions and other employers have the flexibility to make and adjust their talent policies based on their own needs, it is hard to find an obviously enforceable policy change during the sample period that can directly affect returnees’ intention to come back.
21See Donald and Lang (2007) for a recent discussion of statistical power in comparative case studies such as this.
22A venture capitalist’s industry is defined as the industry where he made his first investment, which is also considered the date on which he started operating as a venture capitalist in China. Recall that investments typically take several years to either succeed or fail, and that a venture capitalist supervises several investments over a career. Thus, the lag between my instrument and the dates of most of the outcomes I am examining is considerably greater than one year.
23The data to construct instrumental variables are collected from the Bureau of Labor Statistics.
24Zarutskie (2008) provides evidence that the industry-specific human capital of the venture capitalists has a positive impact on the exit of their portfolio companies.
25For a recent application of the AFT framework in economics, see Gordon B. Dahl: Latent and Behavioral Responses to Extensions in Unemployment Insurance Benefits (2011), in preparation.
26I also tried two other sets of instruments, which are the employment change in the investment & securities industry in the U.S. and Dow Jones Venture Capital Index for U.S. companies. These two instruments reflect the employment prospect and the overall prosperity of the U.S. VC industry, and should therefore affect venture capitalists’ propensity to return. These two instruments only have time variation and therefore have fewer degrees of freedom, but yield qualitatively similar results.
27See Simon and Cao (2009), Breznitz and Murphree (2011), and Wang (2011) for reference.
28Another potential benefit for the “brain drain” countries is documented in the “emigration lottery” literature (Mountford 1997; Stark et al. 1997; 1998). It argues that given a positive probability of migrating to another country, people in the sending country tend to increase their investment in education in order to get a better chance. For those who end up being left behind, they accumulate more human capital than what they would have had, which can eventually turn into a good thing for the economic growth in the sending country.