Red Capitalism: Cadre Parents and Entrepreneurial Children in China

Transcription

1 Red Capitalism: Cadre Parents and Entrepreneurial Children in China Ruixue Jia UCSD IR/PS Xiaohuan Lan Cheung Kong GSB October 22, 2013 Abstract Using a national sample in China, we nd that in provinces with a big government, people whose parents work in government are more likely to become entrepreneurs. Their business revenue is also higher than those entrepreneurs without a cadre parent. The self-evaluation of the determinants of success reveals the key role of social networks and connections to powerful ocials. We provide evidence that cadre parents help their children attract more friends or marry into other cadre families, while big governments create more business opportunities for these children. The eect of cadre parents does not vary with other macro-variables except for government size, nor does the eect of government size vary with other types of parents. Key Words: government size, entrepreneurship, political connection, China Comments are welcome. Jia: rxjia@ucsd.edu. Lan: xhlan@ckgsb.edu.cn. We thank Leora Friedberg, Roger Gordon, Ben Li, Wei Li, Brian Viard, and Michelle White for comments. We thank Shiyao Liu for his excellent research assistance.

2 1 Introduction Entrepreneurship is a key element for economic development and growth. Who becomes an entrepreneur? Social scientists have documented that entrepreneurship can be aected by both nature and nurture. 1 This paper contributes to the literature by examining how the interaction of parental characteristics and institutional environment aects entrepreneurship. As a fast-growing economy where the state plays a dominant role, China is a good research ground to investigate how the interplay of institutional environment and parental traits aects entrepreneurship. The state directly controls state-owned enterprises and indirectly aects the fates of private rms via credit allocations and various regulations, and political connections have been shown to be inuential in the performance of rms (Fan, Wong, and Zhang, 2007; Li et.al., 2008). In particular, plenty of anecdotes suggest that children with parents who have political power have an advantage in the entry into and the management of business, a phenomenon described as Red Capitalism. Among convicted corrupted ocials, numerous cases involve their children who take advantage of the ocial's power and run business. Despite many real-world cases, there has been no systematic research on this phenomenon. In this paper, we examine the role of having cadre parents (who work in government and government-controlled public organizations) on entrepreneurship, along with government size in local economy across Chinese provinces. Using both cross-sectional and retrospective information based on the nation-wide Chinese General Social Survey in 2006, we rst provide evidence that in provinces with a bigger government in the economy, measured by the share of scal spending in GDP, people with cadre parents are more likely to become entrepreneurs. On average, a cadre parent has little eect on entrepreneurship. However, a cadre parent in a province with a government size of one standard deviation (four percentage points) above the mean raises the probability of being an entrepreneur by three percentage points. Naturally, one can imagine that people with cadre parents may be dierent in other aspects. For example, they may be better educated and hence have more human capital to become entrepreneurs. We control for a broad set of demographic variables, including education, age, gender, marital status and ethnicity, as well as their interactions with government size. Moreover, we conduct several falsication tests from both the provincial side and the parental side. On the provincial side, the interaction of having cadre parents and GDP per 1 With regard to the eects of nurture, existing literature has shown that entrepreneurship can be aected by capital constraints (Blanchower and Oswald 1998), regional characteristics such as growth, urbanization and regulations (Reynolds et al. 1994; Djankov et al. 2002) and individual social networks (Djankov et al. 2006). Regarding to the eects of nature, heritable traits, such as risk aversion (Cesarini et al. 2009a) and overcondence (Cesarini et al. 2009b) are shown to be related to entrepreneurship. 1

3 capita or economic openness has no eect on entrepreneurship. On the parental side, the interaction of having entrepreneurial parents and government size likewise has no eect on entrepreneurship. We further investigate how cadre parents, together with government size, aect entrepreneurship. We rst show that for those entrepreneurs with cadre parents, their revenues from business increase with government size. This may explain why they choose to become an entrepreneur. We then study the entrepreneurs' evaluations of the determinants of success in order to understand what matters most for their businesses. We nd that entrepreneurs with cadre parents appreciate social networks and connections to powerful ocials more than other entrepreneurs. Consistent with the views of entrepreneurs, we provide evidence that people with cadre parents do have more friends, measured by the number of non-relative visitors during the Spring Festival, the most important holiday in China. Moreover, cadre families tend to marry each other, which further consolidates their political networks. In addition to these network resources, we show evidence that high government spending creates more rent-seeking opportunities for children with cadre parents in the service sector. As a result, in regions with big governments, entrepreneurial children with cadre parents are more likely to enter the service sector. In contrast, the political careers of children with cadre parents is not aected by government size, suggesting that doing business rather than participating in politics is a more ecient way of exploiting opportunities related to government expansion, particularly for those who already have access to government resources. This study contributes to several lines of literature. Among the large literature on entrepreneurship, a closely related paper studying China's entrepreneurship is Djankov et al. (2006). They use data from seven cities in 2004 and nd that entrepreneurs are more likely to have family members or childhood friends who are entrepreneurs, Similarly, Peng (2004) studies village networks in rural China using cross-sectional data and nds that kinship networks have large positive eects on the size of private rural enterprise and insignicant eects on collective enterprise. Dierent from existing ndings, we nd that the role of networks measured by parental background vary signicantly with institutions. 2 Compared with existing studies, one advantage of our data is that it covers a national sample of urban households (across 28 out of 31 provinces). The data also contain retrospective information so that we can exploit the timing of entrepreneurship and variations in government size within the same province. More importantly, the rich information in our data allows us to 2 It is also interesting to compare our results with Li, Meng, and Zhang (2006). They show that a private entrepreneur is more likely to participate in politics where the market is less developed and the local government imposes more regulations (both factors are related to a bigger government). On the other hand, we show that for those who already possess political capital, a big government promotes entrepreneurial rather than political careers. 2

4 investigate the specic channels through which cadre parents and institutional environments operate. This study also contributes to a large literature on inter-generational welfare dependency. Existing studies have shown both theoretically and empirically that parental wealth and income aect direct nancial transfers to their children as well as human capital and earnings among their children (Rosenzweig and Wolpin 1994, Cameron and Heckman 2001). In particular, the empirical literature of entrepreneurship suggests that entrepreneurs are more likely to be from an entrepreneurial family (Lindquist et al., 2012). Some non-cognitive skills that aect entrepreneurship, such as patience and risk tolerance, could be hereditary (Bowles and Gintis 2002). Entrepreneurial parents may also invest more in such characteristics as a part of human capital of their children (Doepke and Zilibotti 2013). Our study investigates children's entrepreneurship from a new perspective: in regions with big government, the political background of parents matters more than the entrepreneurial background of parents. Our results have important implications for understanding the regional inequality in economic development in China. In the three decades since the Chinese economic reform, the average annual growth rate in real GDP was a spectacular 8%. The growth rate, however, was highly imbalanced across provinces, ranging from 6% to 11%. The GDP per capita in the richest province (Shanghai) was typically ten times higher than in the poorest province (Guizhou), and the ratio also varied from 7 to 12 over the three decades. Some researchers have tried to explain the divergence in economic development across Chinese provinces from the angle of scal resources of local governments (Zhang and Zou, 1998; Lin and Liu, 2000). However, both the empirical ndings and the mechanisms remain ambivalent. To our knowledge, this is the rst paper that estimates the eects of local government, combined with family background, on the allocation of entrepreneurship, a key element for economic growth and development. The interplay between institutions, entrepreneurship, and economic development has been widely discussed since Baumol (1990) and Murphy, Shleifer, and Vishny (1991). We apply this line of inquiry to Chinese regional economies, and we further investigate specic channels involved in the interplay of institutions and entrepreneurship, such as social networks. Our results also help understand the severe income inequality in China. According to the World Development Indicators from the World Bank, the Gini coecient in China rose from 29.1 in 1981 to 42.6 in The income inequality also varies greatly in dierent provinces (Han, Liu, and Zhang, 2012). Entrepreneurship plays an important role in shaping income inequality and social mobility. Entrepreneurs earn more than the average, tend to concentrate in the top wealth class, and experience greater upward wealth mobility than others (Quadrini, 2000; Cagetti and De Nardi, 2006). Our results show that when governments be- 3

5 come powerful, entrepreneurial activities tend to shift away from ordinary families towards cadre families who are already resourceful. This may exacerbate income inequality across households and reduce inter-generational mobility. The rest of the paper is organized as follows. Section 2 briey discusses the institutional background. Section 3 describes the data. Section 4 presents the main results and several falsication tests, while section 5 investigates the mechanism. Section 6 concludes. 2 Institutional Background 2.1 Chinese Local Governments China is organized as a unitary state with four levels of subnational governments: 31 provincial level divisions, 374 prefectures or cities, 2,789 counties or urban districts, and 33,368 townships. 3 The provincial level administrative divisions are the geographic units used in this paper. These divisions include 22 provinces, 4 municipalities that report directly to the central government, and 5 autonomous regions. We refer them simply as provinces. Many provinces are larger and more populated than most countries in the world. According to the 2010 census, the median of the provincial population is 37.3 million (Shaanxi). The most populated province, Guangdong, has over 100 million people, while the least populated Tibet has only 3 million people. Provinces also vary wildly in their wealth, topography, dialect, and even culture. For example, the GDP per capita in the richest Shanghai was 73,263 RMB (or $11,067 using the 2010 exchange rate) in 2010, but only 13,205 RMB (or $1,994) in the poorest Guizhou. Unlike states and counties in the American system, provincial and sub-provincial governments in China do not have powers that inherently belong to them by law. In principle, the powers exercised by these local governments are all delegated to them from the central governmnet. 4 In practice, however, Chinese local governments enjoy autonomy to a large extent. For example, they control land lease rights and can aect the entry of rms. 5 In many aspects, local governments play a much more important role in the local economy than does the central government. In 2005, local governments spent 76% of the whole national budget, a much larger share than the central government (Lou, 2008). This share was also much larger than that of the world's largest federal countries such as the U.S. (46%), Germany (40%), and Russia (38%) (Xu, 2011). 3 The information is current as of February 12, 2013, when we accessed the People's Daily Online: 4 Liberthal (2004) provides an excellent introduction of Chinese political system. 5 Xu (2011) describes the details of the decentralized economic governance in China. 4

6 Local ocials also face very dierent incentives from their counterparts in the U.S. (Gordon and Li, 2011). There are no local elections above the village level, so ocials do not have to worry about being voted out of oce. Meanwhile, foot voting is also restricted. The Hukou system, a registration system that ties people to their birth place, restrict people from moving freely across regions. As a result, the incentives of local ocials and the form of regional competition are fundamentally dierent from the well-known Tiebout competition across jurisdictions in the U.S. Chinese ocials are very much motivated by their career concerns and internal promotion in governments, while in Tiebout competition local ocials are driven by scal incentives or by winning votes (Tiebout 1956; Xu 2011). Another salient feature of Chinese local governments is the huge disproportion between local scal spending and local scal revenue. After the tax reform in 1994, the majority of tax revenue has been collected by the central government but the majority of scal spending remains in local governments. 6 In 2005, while local governments spent 76% of the whole national budget, their revenue only accounted for 48% of the total government revenue (Lou, 2008). The gap between the local spending and revenue is fullled by transfers from the central government. Thus, this paper uses local scal spending instead of revenue to measure the size of local governments in the local economy, which is a more accurate measure of the amount of resources controlled and allocated by the local governments Firms and Cadre Families Given the dominant role of Chinese governments in the economy, it is not surprising that rms benets from connections to governments (Li, Meng, and Zhang, 2006; Li et.al, 2008). 8 One of the strongest types of connection is family ties. Close relatives, particularly children, of powerful government ocials are frequently approached by rms for their access to more resources and business opportunities. Even multinational rms in China exploit these valuable family ties. In August 2013, the Securities and Exchange Commission in the U.S. initiated an investigation into J.P. Morgan for its hiring practices in China. A New York Times article reports that the bank hired the son of the now-chairman of China Everbright Group and the daughter of a Chinese railway ocial, and it went on to win business from 6 Lou (2008) introduces the process of the scal reform since the 1980s. 7 Both government spending and government revenue have been used to measure government size in the literature, though government spending seems more popular. See, for example, Rodrik (1998), Gldfajn and Valdes (1999), Fatas and Mihov (2003), and Ferreira and Gyourko (2009). 8 Proting from political connections isn't unique to China. For a related large literature based on experiences from other countries, see, for example, Fisman (2001), Khwaja and Mian (2005), Faccio (2006), and Ferguson and Voth (2008). 5

7 Everbright and state-controlled China Railway Group. 9 After about three weeks, the same railway ocial was indicted on charges of accepting graft payments of nearly $8 million. 10 The same NYT article also points out that global companies routinely hire the sons and daughters of Chinese politicians. The list of these companies is long, including many big investment banks such as Goldman Sachs and Citigroup, and manufactuerers such as the pharmaceutical company Novartis AG. 11 With the strong and valuable government connections, many children of government ocials start their own business. Among convicted corrupted ocials, numerous cases involve their children who take advantage of the ocial's power and run business. For example, a son of Jiating Li, the convicted former governor of Yunnan province, made a huge amount of illegal prots in real estate development. 12 A son of Weigao Cheng, the convicted former secretary of the Communist Party of China (CPC) of Hebei province, set up 32 companies in ten years. 13 The rise of entrepreneurship among cadre families is presumably prevalent enough across China so that a specic term has been coined to describe it: zheng shang jia ting (a family of ocials and business people). As a reection of the severity of the problem, the Disciplinary Committee of the CPC has tried to ban a range of business activities for the spouse and the children of top ocials. 14 Of course, only very few individuals are related to top government ocials. In general, the family ties of the children of government ocials would be valuable if they choose to run business themselves. 3 Data We use the individual-level data from the Chinese General Social Survey in 2006 (CGSS06), a Chinese version of the General Social Survey in the U.S. that is conducted by the National Opinion Research Center. The CGSS is also a part of the International Social Survey Programme that covers 48 countries including the U.S. Started in 2003, the CGSS is an annual 9 Silver-Greenberg, Jessica, Ben Protess, and David Barboza, Hiring in China by JPMorgan Under Scrutiny, New York Times (2013, August 18). 10 Gough, Neil, Bribery Charges in China for Ocial Whose Child Worked for JPMorgan, New York Times (2013, September 4). 11 Koons, Cynthia, Dinny McMahon, and Jeremy Page, J.P. Morgan Under Investigation for Hiring Practices in China, Wall Street Journal (2013, August 18). 12 The story is from the Xinhua News Agency (the largest news agency in China): accessed on September 23, The provincial party secretary is head of the Party in a province, while the governor is the chief executive. The story is from the Guangming Daily (one of the most inuential newspapers in China): accessed on September 23, For more details of this banning order in 2001, see the ocial website of the Communist Party of China: accessed on September 23,

8 repeated cross-sectional survey, jointly conducted by the Renmin University of China and the Hong Kong Science and Technology University. The survey uses multi-stage stratied sampling design based on administrative units, from provinces to communities. 15 Designed by sociologists, the CGSS records very rich information of family backgrounds and social networks. For example, besides individuals' education and occupations, the CGSS also records education and occupations of the spouse, parents, and the father-in-law. A typical wave of the CGSS includes about 10,000 urban or rural households. The CGSS data from 2003, 2005, 2006, and 2008 are available upon application. The CGSS06 is the only wave that includes all information necessary for this research. We focus on urban residents whose most recent job started after 1980, the rst year after the Chinese economic reform from a strict central-planning economy. Before the reform, people typically accepted whatever jobs assigned by governments. The CGSS06 covers 28, out of 31, provinces in China mainland. The three western provinces that are not covered are Ningxia, Qinghai, and Tibet. Cross-Province Data The cross-province sample includes people who worked in By residential provinces, we match this sample with provincial level characteristics in 2005 from the Chinese Statistical Yearbook. Panel A of Table 1 summarizes the sample. [Table 1 about here] We dene entrepreneurs as business owners or self-employed, about 21% of the sample. A cadre parent refers to a parent who works in government or in a public organization aliated with governments (known as shi ye dan wei or public institutions). Parents' occupation is the occupation at the time when the surveyee was 18 years old. 24% of the individuals have a cadre parent. We measure government size by the scal spending of the residential province divided by the provincial GDP. The provincial scal spending also includes the spending of subprovincial governments such as city and county governments. As discussed in Section 2.1, scal spending is a better measure of the resources of local governments than scal revenue. Divided by the provincial GDP, this measure of government size captures the amount of resources controlled and allocated by all the local governments within a province. The size varies signicantly from 0.08 (Shandong) to 0.26 (Guizhou), with a mean of GDP per capita varies from RMB 5,052 in Guizhou to 52,060 in Shanghai, or from 616 U.S. dollars to 6,349 dollars, based on the average exchange rate in Figure 1 plots the spatial 15 Bian (2012) describes the survey design of the CGSS in details. 7

9 distribution of provincial government size, which is negatively correlated with economic development. Small governments concentrate in the developed coastal provinces in the east, while large governments concentrate in the underdeveloped western area. In 2005, the correlation coecient between provincial government size and the log of GDP per capita was Table 1 also shows large variations in provincial economic openness that is the total trade ow (imports plus exports) over GDP: from 0.06 in Guizhou to 1.63 in Shanghai. Again, economic openness is negatively related to provincial government size though the correlation coecient is small as [Figure 1 about here] Panel A of Table 1 also reports some demographic variables that could aect the choice of entrepreneurship. The CGSS only surveys people who are older than 18 but younger than 70. Since our sample includes urban residents only, the probability of college education (four-year or two-year colleges) is high as In Chinese elementary and middle school system, there are some magnet schools that are typically characterized by high-quality peers, rigorous curricula, and better teachers. These schools are highly selective, and only 18% of people attend such a school. Retrospective Data In the cross-province and over-time analysis, we use the retrospective questions on the start year of the current job, or the last job if currently unemployed or retired. This sample includes all people who have ever worked, 57% larger than the cross-province sample that only includes current workers. By job years and residential provinces, we match the sample with provincial government size from according Statistical Yearbooks. Panel B of Table 1 shows the summary statistics of three key variables. 16% are entrepreneurs, ve percentage points lower than in the cross-province sample in 2005, which reects the expansion of the private economy since the economic reform. 22% have a cadre parent. Government size over the 26 years varied from 0.05 to 0.27, a wider range than that in the cross-province sample. 4 Cadre Parents, Government Size and Entrepreneurial Children 4.1 Cross-Province Analysis Our main hypothesis is that when the local government expands, people with a cadre parent are more likely to become an entrepreneur, compared to people without a cadre parent. The 8

10 simplest test is to run the following simple regression in each province: entrepreneur i = β 1 cadreparent i + ε i, (1) where entrepreneur i is a binary variable with 1 for individual i being an entrepreneur in cadreparent i is an indicator that either the father or the mother of i worked in governments at the time when i was 18 years old. Then for each province, we collect the estimated ˆβ 1, the dierence in the probability of entrepreneurship with or without a cadre parent. Figure 2 shows that these ˆβ 1 s increase with government size. A simple regression of ˆβ 1 of each province on government size suggests that an one percentage-point increase in government size is related to 0.7 percentage-point increase in the dierence in the probability of entrepreneurship between a cadre family and a common family. [Figure 2 about here] More formally, we combine all province-specic regressions in equation (1) and use the following dierence-in-dierences specication to test our hypothesis: entrepreneur ip = β 2 cadreparent ip dgovsize p + β c cadre parent ip + provincef E p + +X ip + X ip dgovsize p + ε ip, (2) where dgovsize p is the share of scal spending in GDP in province p minus the sample mean: govsize p govsize p. The demeaning transformation does not aect β 2, while β c can be interpretated as the eect of cadre parents on entrepreneurship at the mean value of government size. A set of province dummies provincef E p controls for local characteristics, such as economic environment and culture of entrepreneurship etc. We cannot include government size separately in the equation since it would be absorbed by the province dummies. Besides political connection, children with cadre parents could be dierent in other dimensions. A vector of individual characteristics X ip controls for college education, having attended a magnet elementary or middle school, age, gender, marital status, and minority status. Furthermore, their interactions with government size, X ip dgovsize p, allow the eects of these observables vary with government size as well. Columns 1-2 in Table 2 show the average eect of cadre parents on entrepreneurship. The eect is close to 0 once the demographic information of the children is considered. Columns 3-5 report the interaction eect of cadre parents and government size. The standard errors are clustered at the province level to allow for spatial correlation among individuals within the same province. The clustered errors also reect the fact that the variation in government 9

11 size is at the province level. The number of clusters is Consistent with Figure 2, ˆβ 2 is signicantly positive, with or without controlling for demographic variables. A cadre parent with a government size of one standard deviation (four percentage points) above the mean raises the probability of being an entrepreneur by about three percentage points. [Table 2 about here] Across specications, the probability of entrepreneurship is 15 percentage points lower among college graduates. Beofre 1996, jobs of college graduates in China were assigned upon graduation by governments, typically a position in governments or state-owend enterprises. This arrangement was terminated in 2000, after three years of transition. 40% of college-educated people in our sample graduated before 1996 and 53% graduated before 2000, and most of them chose a government job. On average, the probability of working in governments is about 30% higher among college graduates in our sample (see column 6 of Table 10). Married people are more likely to become an entrepreneur. Because of the underdeveloped capital market in developing countries, poor people are less likely to start a business (Banerjee and Newman, 1993). Marriage and pooled resources may help overcome these budget contraints. We will discuss more about the role of marriage in Section 5.3. One concern with identication is migration across provinces. If people with entrepreneurial skills move across provinces in response to local scal expenditure, and if the moving pattern diers across types of parents, our estimates might be biased. However, the estimation uses only urban residents among whom cross-province migration is rare, only 3% in our sample. 17 Column 6 in Table 2 shows that our estimate of equation (2) in the sample of only natives is similar to other columns. Another concern with identication is the endogenous choice of parents' cadre status. Parents could choose a government job in order to help their children to run business, particularly when parents expect that their government connection would becomemore important with the expansion of governments. ˆβ2 would be overestimated due to this selection bias. However, this is unlikely in our case. cadreparent ip is the cadre status at the time when i was 18 years old, which had been predetermined before i started a job. Chinese governments are a very closed system and the job mobility between governments and other sectors 16 As shown in the Monte Carlo exercises with cross-sectional data in Cameron, Gembach, and Miller (2008), the cluster-robust standard errors work ne with 25 or 30 clusters. Considering the small number of clusters, we also follow their procedure of wild cluster bootstrap and bootstrap t-statistics 500 and 999 times. The resulted p-values are and In a national survey of domestic migrants in China in 2010 (NPFPCC, 2011), cross-province urban migrants only accounted for 8% of all migrants. The number is higher than the number of 3% in our sample, the percentage of urban migrants among urban residents. The dierence is the result of the larger population of urban residents (460 million in 2010) than the population of migrants (221 million in 2010). 10

12 is very low, particularly among older generations. Most cadre parents started to work for governments long time before their children reach the work age or even before their children were born. For example, among government workers who had an adult child in 2005, 83% started their whole job career in governments and 12% started in state-owned enterprises. 4.2 Cross-Province and Over-time Analysis Cross-province analysis is not able to capture the timing of starting a business. Fiscal spending uctuates over time, even within a province. If high spending generates extra business opportunities for cadre families, it should aect both the probability and the timing of becoming an entrepreneur. Based on the retrospective questions on the job start year, this section extends the cross-province analysis to over time. We still use those who started their current job or the most recent job between 1980 and Figure 3 shows the national pattern of entrepreneurship and government size from 1980 to 2005, divided into three decades. Immediately after the economic reform from a centrally planned economy in 1979, the share of the scal spending in GDP declined, from 0.22 in the 1980s to 0.13 in the 1990s. The private sector started to boom, and entrepreneurship rose fast among both cadre families and non-cadre families, slightly faster for non-cadre families. From the 1990s to the 2000s, government size rose again. While entrepreneurship among non-cadre families declined, it kept rising among cadre families. In the 2000s, children of cadre families were more likely to become entrepreneurs than children of non-cadre families. [Figure 3 about here] We extend the cross-province regression in equation (2) into the following regression entrepreneur ipt = β 3 cadreparent ipt dgovsize pt + β c cadre parent ipt + provincef E p + yearf E t +(provincef E p yearf E t ) + X ipt + X ipt dgovsize pt + ε ipt, (3) where entrepreneur ipt is 1 for individual i in province p who becomes an entrepreneur in year interval t. The period of 26 years (from 1980 to 2005) is divided into ve ve-year intervals. 18 All the results are robust if we use three 10-year intervals instead, , , and These results are presented in Appendix Table A1. govsize pt is the average share of scal spending in GDP in province p during the time interval t, and dgovsize pt is its deviation from the sample mean: govsize pt govsize pt. 18 The 5-year intervals are , , , , and The last interval includes six years. 11

13 In addition to province xed eects provincef E p, a set of time dummies yearf E t captures general uctuation in macroeconomy in China over time. We also include all interaction terms between these time dummies and province dummies. This exible specication captures all province-year characteristics of a local market. X ipt is the same set of demographic variables as in equation 2, and the interactions with dgovsize p allow their eects vary with government size as well. 19 Despite a larger sample and a dierent specication, the estimates of equation (3) in Table 3 are essentially the same as in Table 2. A cadre parent with a government size of one standard deviation above the mean raises entrepreneurship by about three percentage points. As in Table 2, the probability of entrepreneurship is lower among college graduates while higher among married people. The probability is lower among women, while there is no such a dierence in Table 2 that uses the data of 2005 only. This might reect the improved labor market conditions and more job opportunites for women over 26 years from 1980 to The probability of entrepreneurship is also ve percentage points lower among those who are ten years older, while there is no such a dierence in Table 2. This might also reect more entrepreneurial opportunites among younger generations after the economic reform in China. [Table 3 about here] 4.3 Falsication Tests Cadre Parents and Other Provincial Dimensions The eect of cadre parents on entrepreneurship varies with the role of governments in the economy. As a falsication exercise, we test whether the eect of cadre parents also varies with other provincial characteristics. The general eects of these characteristics are absorbed in the province xed eects in the cross-province analysis (equation 2) and their interactions with year xed eects in the crossprovince and over-time analysis (equation 3). Here we focus on their interaction eect with cadre parents. We choose those provincial characteristics that are both related to government size and entrepreneurship. The level of provincial income, measured by GDP per capita, could aect entrepreneurship, by changing local factor prices and the demand for products and services. GDP per capita could be related to government size as well. The well-known Wagner's Law states that government size tends to grow as a society becomes wealthier. The large empirical literature, 19 In equation (2), the marital status refers to the status at the time of survey in Here, it refers to the marrital status at the time of starting the job. 90% of people report that they had nished their highest educatoin before they started their current job or the most recent job. Thus, we use the same educational attainment and quality variables as in equation (2). 12

14 however, produces ambiguous evidence on this stipulation (see, for example, Shelton 2007; Durevall and Henrekson 2011). Across Chinese provinces, a larger provincial government size is related to lower GDP per capita, as shown in Figure 1. We replace demeaned provincial government size govsize p in equation (2) by demeaned ln(gdp per captia) p in Column 1 of Table 4 shows that the eect of the interaction term between a cadre parent and GDP per capita is neither economically nor statistically signicant. After cotrolling for this interaction, column 2 shows that the eect of cadreparent ip dgovsize p is still signicantly positive, while the magnitude is slightly larger than the baseline estimates in Table 2. [Table 4 about here] Another relevant macro-variable is economic openness, measured by the total foreign trade volume (exports plus imports) over GDP. Across countries, some inuential works have documented that economic openness is related to government size (Rodrik 1996; Ram 2009). A more open local economy could attract more foreign business partners and competitors, which changes the environment of local entrepreneurs. Across Chinese provinces, economic openness is negatively related to provincial government size, though the correlation coecient is small as We replace demeaned govsize p in equation (2) by demeaned (economic openness) p. Column 3 shows that the interaction between economic openness and a cadre parent is neither economically nor statistically signicant. After cotrolling for this interaction, column 4 shows that the eect of cadreparent ip dgovsize p is close to the baseline estimates. It is impossible to exhaust all relevant macro-variables in our falsication tests. Since we have controlled for all xed eects of the local market and most macro-variables are related to at least one of the three variables (income, openness, and government size), we nish this section by including all the interactive terms between cadre parents and the four macro-variables in one regression. We also control for the impact of demographic variables as in the main specications. The results are displayed in columns 5 and 6 of Table 4. The estimated coecient of (cadreparent ip dgovsize p ) is somewhat larger than the estimates of equation 2: a cadre parent with a government size of one standard deviation above the mean increase in the size of the local government raises the probability of being an entrepreneur by about four percentage points. Government Size and Entrepreneurial Parents The empirical literature of entrepreneurship suggests that entrepreneurs are more likely to be from an entrepreneurial family. 20 Although government expansion naturally enhances the role of a cadre parent, it seems unclear 20 For recent empirical evidence and a literature summary, see Lindquist et al. (2012). 13

15 that the expansion would also change the role of an entrepreneurial parent. Instead of replacing government size with other macro variables as in the last section, we conduct another falsication test by replacing cadreparent ip with (entrepreneurial parent) ip in equation 2. Without including provincial dummies and local scal spending, column 1 of Table 5 shows that an entrepreneurial parent raises entrepreneurship by about 9 percentage points. This result is consistent with Djankov et al. (2006) who also report a positive correlation between the probability of entrepreneurship and having a entrepreneurial family member in China. However, after including provincial dummies, columns 2 and 3 show that neither an entrepreneurial parent nor its interaction with local government size is signicant. Columns 4 and 5 show that our baseline results do not alter even after controling for entrepreneurial parents. [Table 5 about here] 5 Discussion of the Mechanism In this section, we investigate how cadre parents, together with government size, increase the probability of entrepreneurship. Section 5.1 shows that cadre parents and larger governments increase business earning of entrepreneurs, but they do not aect salary of non-entrepreneurs. In order to explain what contributes to the success of entreprenurs with cadre parents, Section 5.2 shows that the self-evaluation of the determinants of success reveals the importance of social networks and connections to powerful ocials. Guided by these ndings, Section 5.3 documents that cadre parents help attract broad networks and Section 5.4 provides evidence that larger governments create more business opportunities for those with cadre parents. 5.1 Earnings of Entrepreneurs and Non-entrepreneurs The expansion of governments creates some advantage of doing business for those who are connected with governments, which may explain their higher probability of entrepreneurship. Before discussing the source of the advantage, we rst show more evidence of the advantage: higher business earnings. Among all entrepreneurs, the following specication estimates the eect of cadre parents and government size on business earnings: ln(businessrev) ip = β 4 (cadreparent ip dgovsize p ) + cadre parent ip + provincef E p + +X ip + X ip dgovsize p + SectorF E i + F irmsize ip + ε ip, (4) 14

16 where ln(businessrev) ip is the log of after-tax earnings from the business for entrepreneur i in province p. 21 The data on this variable are only available for the year of SectorF E i is three dummies for four broad business sectors: agriculture, mining and manufacturing, utilities and construction, and service. These dummies control for the systematic dierence in earnings across sectors. The rm size, F irmsize ip, is measured by the log of the number of employees, which varies from 0 (self-employed) to 100. The results are presented in columns 1-3 in Table 6. Column 1 shows that a cadre parent per se does not aect the business reveue. The estimates in columns 2 and 3 show that for entrepreneurs from a cadre family, a standard deviation (four percentage-point) increase in the size of the government increases the business prots by about 40%. Naturally, the earnings increase with rm size. These results are broadly consistent with Li et.al (2008). They show that the connection of entrepreneurs to the ruling Communist Party of China, measured by their party membership, increases the protability of private rms, particularly in regions with less liberal markets. Their measure of the Party connection, however, could be endogenous, since those who are interested in running business would be more likely to build political connections, particularly in regions with weaker market institutions (Li, Meng, and Zhang, 2006). [Table 6 about here] In contrast, for salary earners, columns 5 and 6 show that the coecient of the interaction between cadre parents and government size is neither statistically nor economically signicant. However, unlike among entrepreneurs, a cadre parent per se increases salary by about 13%. This result is surprisingly close to the results of Li et.al (2012). Among fresh college graduates in urban China, they nd that a cadre parent increases the wage of their rst job by 14% to 15%. The eects of other demographic variables behave mostly as expected. These results suggest the advantage created by cadre parents combined with larger governments contributes more to the performance of entrepreneurs than to other workers. Now we turn to investigate the source of this advantage. 5.2 Self-evaluation of the Determinents of Success In order to understand the success of entrepreneurs, we resort to the self-evaluation of determinants of success in the CGSS06. The survey lists many factors of success, such as networks or education. For each factor, people evaluate its importance in success and choose one of 21 If the business has multiple partners, the earnings here are the individual share earned by the surveyee, not the total earnings generated by the business. 15

17 the six answers: a critical determinant, very important, important, not very important, not important at all, and hard to say. We choose eight factors and test whether the evaluation diers between cadre families and common families, among entrepreneurs and among other workers. These eight factors are: connection to powerful ocials, broad networks, ambition, intelligence, talents and appearances, gender, luck, and good education. For each factor, we construct a binary variable that is 1 if its evaluation is a critical determinant of success. Then we regress this binary variable on the indicator of a cadre parent in the sample of entrepreneurs, controlling for province xed eects and the same set of demographic variables as in equation 2. Panel A of Table 7 shows that among entrepreneurs, those from a cadre family are more likely to think connection to powerful ocials, broad networks, and ambition as a critical determinant of success than those from a common family. There is no dierence in the evaluation of intelligence, luck, and good education. Interestingly, among entrepreneurs, those from a common family appreciate more the role of gender and appearances. This suggest that in the risky and competitive world of entrepreneurs, people from a common family may have to play all their cards that the nature delivers. 22 [Table 7 about here] As a comparison, Panel B of Table 7 re-runs the same group of regressions in the sample of other workers. In this sample, connection to powerful ocials and ambition are less important for people from a cadre family, so are intelligence and luck. There is no dierence in the evaluation of other factors. We nally use the whole sample and runs a dierence-in-dierences type of regression, which includes an indicator of cadre parents, an indicator of entrepreneurship, and their interaction. The coecient of the interaction term reects the extra importance of a factor in the success for entrepreneurs with a cadre parent. This coecent is very large and signicant for factors of knowing powerful ocials, broad networks, and ambition, as shown in Panel C of Table 7. The above results suggest that entrepreneurs from a cadre family benet more from connections to powerful ocials and networks; or that people from a cadre family who appreciate the two factors and are willing to take advantage of them are more likely to become an entrepreneur. Either way, entrepreneurs from a cadre family exploit their government connections and network resources. Two questions remain: What are these network resources? How does knowing powerful ocials help? We answer these two questions in the following subsections. 22 We do not report coecients of demographic variables for brevity. Across samples and specications, female tend to appreciate more the role of intelligence, talents and appearances, luck, and good education. 16

18 5.3 Social Networks Related to Cadre Parents Friend Networks The CGSS06 provides an interesting and informative measure of social networks, namely visitors during the Chinese Spring Festival. Like the Christmas or the Thanksgiving in the western world, the Spring Festival is the most important Chinese holiday and it is the time for family reunion and friend visit. The CGSS06 separates visits by friends from visits by relatives, the latter of which are more likely to be driven by family size rather than social networks and hence can serve as a placebo. For an individual i in province p, we examine the correlation between having a cadre parent and the number of visitors during the Spring Festival of 2005: ln(1 + #V isitors) ip = β 5 cadreparent ip + provincef E p + X ip + ε ip, (5) The results are presented in Table 8. Columns 1-2 show that having a cadre parent has no impact on the visits of relatives but increases the number of visiting friends by 9%. Since a larger family tends to have both more relatives and more friends, Column 3 controls for the number of visiting relatives, and the eect of cadre parents on visiting friends becomes a bit larger. Columns 4-6 add having entrepreneurial parents as a control. Dierent from the eect of cadre parents, they show that entrepreneurial parents aect neither type of visitors. Not surprisingly, college graduates tend to have more visiting friends, whereas marriage brings both more relatives and more friends. Minorities tend to have more visiting relatives, which may reect their larger family size since they are not restricted by the one-child policy. [Table 8 about here] Marriage Networks The evidence above may be an underestimate of the eect on networks, as friends are only a part of networks. We further take advantage of the detailed information in the CGSS06 and examine the eect on marriage networks. In particular, we would like to know whether those with cadre parents are more likely to marry a spouse with similar cadre backgrounds. This expanded political connections may further facilitate entrepreneurship when governments become more inuential in the economy. For married people, the CGSS06 documents job information of the spouse and the fatherin-law at the time of wedding. We estimate the eects of a cadre parent on the spouse type by using a regression similar to equation 3: spousetype ipt = β 6 cadre parent ipt + provincef E p + yearf E t +(provincef E p yearf E t ) + X ipt + ε ipt, (6) 17

19 where spousetype ipt is a binary indicator of the spouse type of individual i in province p who is married in year t, which is 1 if either the spouse or the father-in-law works in government or in a public organization aliated with governments. Similar to equation (3), we include those who were married between 1980 and 2005, and we divide these 26 years into ve 5-year intervals. The full set of province dummies and year dummies control for all time-province specic factors that aect the marriage market, including the changes in cultures and values across provinces and over time. Columns 1-3 in Table 9 show that having a cadre parent increases the probability of marrying a spouse with cadre backgrounds by 32 percentage points. It is not uncommon in China that cadre parents help potential children-in-law nd a job in governments, even before the marriage. Thus, the estimated coecients may partially reect the inuence of cadre parents that are extended to children-in-law. Columns 4-6 use only cadre father-in-law at the time of wedding as the dependent variable and re-run equation 6. In this case, the estimated ˆβ 6 better captures the expansion of political networks through marriage: a cadre parent increases the probability of marrying a spouse with a cadre father by 26 percentage points. 23 Interestingly, better education seems to help in marrying a cadre spouse, but less so in marrying a spouse with a cadre father. [Table 9 about here] 5.4 Business Opportunities Related to Government Size Besides broader networks, another key determinant of success among entrepreneurs from a cadre family is the connection to powerful ocials. Since those with a cadre parent are connected to the governments by birth, this subsection focuses on the power of the local governments in the economy, which is certainly the source of ocials' power. Particularly, we argue that for those from a cadre family, a larger government creates a less costly and more protable business environment. A big government is typically related to inecient and capricious bureaucracy, which raises the entry cost for start-up rms (Djankov et. al., 2002). Wang, Fan, and Li (2012) reports the evaluation of 4,586 Chinese entrepreneurs on the eciency of local governments. The correlation cocient between the evaluation and local government size is For those with government connections, dealing with bureaucracy would be easier, which reduces the cost of starting a rm. 23 Some papers estimate the eects of parents-in-law on married children's welfare and labor market outcomes. See, for example, Rosenzweig and Wolpin (1994), Behrman and Rosenzweig (2006), and Wang (2013). 18

20 More importantly, big governments with high scal spending could generate more rentseeking opportunities and corruption (Goel and Nelson 1998). One typical way of grabbing rents in many countries is that government ocials run business by themselves or by their relatives (Murphy, Shleifer, and Vishny, 1991). In China, there are at least three types of well-known business opportunities that are directly related to the size of local governments. First, at the expense of scal spending, government ocials spend a lot on eating and drinking, entertainment, and tourism and oversea travel. Various estimates show that these costs are approximately 5% of GDP (Zhu, 2010). In addition, a large part of entertainment and travel cost of Chinese rms are also spent on government ocials, such as eating, drinking, gifts, Karaoke, and sports club membership etc. (Cai, Fang, and Xu 2011). For those with government connections, it would be protable to start a business related to these activities and attract government clients. Second, scal spending expands with scal revenue for local governments, of which a large share comes from land sale. For example, leasehold sales revenues ranged from 2.6% to 5% of local GDP in three big cities in 2004 and 2005, and these sales are a major venue for corruption (Cai, Henderson, and Zhang 2009). Among convicted corrupted ocials, the cases related to land sale and real estate development increased from 6% to 23% between 2000 and 2009 (Gong and Wu, 2012), while the average provincial scal spending over GDP increased from 14% to 22%. Anecodotal evidence abounds that family members of government ocials who control land allocation become very wealthy in the business of real estate. Third, the expansion of certain government sectors directly create more business opportunities. Two salient examples are propaganda system and judical system, both are a part of Chinese governments. The huge propaganda system controls nearly all main TV channels and newspapers across China. As a result, many people from cadre families become advertising agencies. For example, the daughter and the son-in-law of Suixin Mu, the convicted former mayor of Shenyang, almost monopolized the business of billboard ads in the city. 24 Due to the lack of independence of judical system, it is only natural that cadre families start law rms and prot from their inuence on court decisions. Not surprisingly, according to Wang, Fan, and Li (2012), the provincial government size is negatively related to the quality of legal environment of running a business. All these bussiness practices discussed above are so prevalent and well-known that they were actually banned among the close relatives of top government ocials. In 2001, for the spouse and the children of top ocials, the Disciplinary Committee of the CPC banned those business activities that are under the ocial's authority and could conict with public 24 The story is from the People's Daily (the most inuential ocial newspaper from the Communist Party of China): accessed on September 23,

21 interests. The banning order explicitly listed four types of business: real estate development, advertising agencies, law rms, and entertaining business such as Karaoke clubs and night clubs. 25 CPC. The banning order, however, only targetted very top ocials of governments or the We show several observations that are consistent with the discussions above. First, entrepereurs with cadre parents in provinces with big governments are more likely to enter the service sector that includes real estate development, restaurants and clubs, advertising agencies, and law rms etc. For non-entrepreneurs with cadre parents, however, there is no such a eect. Second, large governments do not promote political careers of people with cadre parents. For them, who already have access to governemnt resouces, running business, instead of pursuing a political career, seems a better way to exploit opportunities created by big governments. Among entrepreneurs, we test whether those with a cadre parent in provinces with big governments are more likely to enter the service sector. The specication is similar to the regression of business revenues in equation (4), and we replace the dependent variable with a dummy for the service sector. Columns 1 and 2 in Table 10 show that a cadre parent together with an one standard deviation of the government size above the mean (four percentage points) raises the probability of entering the service sector by seven percentage points. The eect of a cadre parent per se is close to zero, which also highlights the role of government size. This nding echoes the qualitative discussions above on business opportunities. As a comparison, columns 3 and 4 shows the probability of working in the service sector among non-entrepreneurs. In this case, the interaction between cadre parents and government size is insignicant, which suggests the interactive eects are particular for entrepreneurs. Among non-entrepreneurs, people with cadre parents are more likely to work in the service sector since this sector includes governments. So do college graduates, as we discussed in Section 4.1. Columns 5 and 6 replace the dependent variable with the dummy of working in governments. Although cadre parents do signicantly increase the choice of working in governments for their children, this eect does not change with government size. [Table 10 about here] 25 For more details of this banning order, see the ocial website of the Communist Party of China: accessed on September 23,

22 6 Conclusions In this paper, we provide new evidence that the political background of parents aects whether the children become entrepreneurs. We nd that in provinces with bigger government, those with cadre parents are more likely to become entrepreneurs. This nding suggests that the eect of parental characteristics emphasized in the existing literature is closely contingent on institutions. We also provide evidence on the mechanism behind the phenomenon: those with cadre parents have an advantage in becoming entrepreneurs in regions with bigger government, reected in their higher business revenue. There are at least two sources of this advantage: their cadre parents help them attract broad social networks, and the big local government creates more business opportunities for them. The allocation of entrepreneurship is a key element in understanding economic development and income distribution. With more data, future research may further investigate the welfare implications of the (mis)allocation of entrepreneurial talents. Are the rms owned by the entrepreneurs from cadre families less ecient? Do they contribute less to economic growth? To what extent can the dierent allocation of entrepreneurs across regions explain the divergence of regional development and income distribution? In terms of policies, our ndings suggest that restricting government size could mitigate the advantage of cadre families and encourage entrepreneurship among ordinary people. Future research may also investigate the eects of other policies, such as nancial liberalization or business education, on entrepreneurship. Finally, with more data, future research may generalize our main ndings to cross-country comparisons, particularly in developing countries. 21

23 References Banerjee, Abhijit V. and Andrew F. Newman (1993), Occupational Choice and the Process of Development, Journal of Political Economy, Vol. 101(2): Baumol, William J. (1990), Entrepreneurship: Productive, Unproductive, and Destructive, Journal of Political Economy, Vol. 98 (5): Behrman, Jere R. and Mark R. Rosenzweig (2006), Parental Wealth and Adult Children's Welfare in Marriage, Review of Economics and Statistics, Vol. 88: Bian, Yanjie, and Lulu Li (2012), The Chinese General Social Survey (2003-8), Chinese Sociological Review, Vol. 45: Blanchower, David G., and Andrew J. Oswald (1998), What Makes an Entrepreneur? Journal of Labor Economics, Vol. 16: Bowles, Samuel, and Herbert Gintis (2002), The Inheritance of Inequality, Journal of Economic Perspectives Vol. 16: Cagetti, Marco, and Mariacristina De Nardi (2006), Entrepreneurship, Frictions, and Wealth, Journal of Political Economy, Vol. 114: Cai, Hongbin, Hanming Fang, and Lixin Colin Xu (2011), Eat, Drink, Firms and Government: An Investigation of Corruption from Entertainment and Travel Costs of Chinese Firms, Journal of Law and Economics, Vol. 54: Cai, Hongbin, J. Vernon Henderson, and Qinghua Zhang (2009), China's land market auctions: evidence of corruption. NBER Working Papar Cameron, Colin A., Jonah B. Gelbach, and Douglas L. Miller (2008), Bootstrap-based Improvements for Inference with Clustered Errors, Review of Economics and Statistics, Vol. 90: Cameron, Stephen V., and James J. Heckman (2001), The Dynamics of Educational Attainment for Black, Hispanic, and White Males, Journal of Political Economy, Vol. 109: Cesarini D., C.T. Dawes, M. Johannesson, P. Lichtenstein and B. Wallace (2009a), Genetic Variation in Preferences for Giving and Risk-Taking, Quarterly Journal of Economics, Vol. 124: Cesarini D., M. Johannesson, P. Lichtenstein and B. Wallace (2009b), Heritability of Overcondence, Journal of the European Economic Association, Vol. 7: Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer (2002), The Regulation of Entry, Quarterly Journal of Economics, Vol. 117: Djankov, Simeon, Yingyi Qian, Gerard Roland and Ekaterina Zhuravskaya (2006), Who Are China's Entrepreneurs? American Economic Review, Vol. 96, Doepke, Matthias, and Fabrizio Zilibotti (2013), Culture, Entrepreneurship, and Growth, 22

24 mimeo. Durevall, Dick, and Magnus Henrekson (2011), The Futile Quest for a Grand Explanation of Long-Run Government Expenditure, Journal of Public Economics, Vol 95: Faccio, Mara (2006), "Politically Connected Firms", American Economic Review 96: Fan, Joseph PH, Tak Jun Wong, and Tianyu Zhang (2007), Politically Connected CEOs, Corporate Governance, and Post-IPO Performance of China's Newly Partially Privatized Firms, Journal of Financial Economics, Vol. 84: Fatas, Antonio, and Ilian Mihov (2003), The Case for Restricting Fiscal Policy Discretion, Quarterly Journal of Economics, Vol. 118: Ferguson, Thomas and Voth, Hans-Joachim (2008), "Betting on Hitler: the value of political connections in Nazi Germany", Quarterly Journal of Economics 123, pp Ferreira, Fernando, and Joseph Gyourko (2009), Do Political Parties Matter? Evidence from U.S. Cities, Quarterly Journal of Economics, Vol. 124: Fisman, Raymond (2001), Estimating the Value of Political Connections, American Economic Review, Vol. 91: Goel, Rajeev K., and Michael A. Nelson (1998), Corruption and Government Size: a Disaggregated Analysis, Public Choice, Vol. 97: Goldfajn, Ilan, and Rodrigo O. Valdes (1999), The Aftermath of Appreciation, Quarterly Journal of Economics, Vol. 114: Gong, Ting and Wu, Muluan (2012), A Report on China's Corruption from 2000 to 2009: an Analysis of over 2000 Reported Cases, Sociology Research, Vol. 4: Gordon, Roger H. and Wei Li (2011), Provincial and Local Governments in China: Fiscal Institutions and Government Behavior, NBER Working Paper Han, Jun, Runjuan Liu, and Junsen Zhang (2012), Globalization and Wage Inequality: Evidence from Urban China, Journal of International Economics, Vol. 87(2): Khwaja, Asim Ijaz and Atif Mian (2005), Do Lenders Favor Politically Connected Firms? Rent Provision in an Emerging Financial Market, Quarterly Journal of Economics, Vol. 120: Li, Hongbin, Lingsheng Meng, Xinzheng Shi and Binzhen Wu (2012), Does Having a Cadre Parent Pay? Evidence from the First Job Oers of Chinese College Graduates, Journal of Development Economics, Vol. 99: Li, Hongbin, Lingsheng Meng, Qian Wang and Li-An Zhou (2008), Political connections, nancing and rm performance: Evidence from Chinese private rms, Journal of Development Economics, Vol. 87: Li, Hongbin, Lingsheng Meng, and Junsen Zhang (2006), Why Do Entrepreneurs Enter 23

25 Politics? Evidence from China, Economic Inquiry, Vol. 44.: Lieberthal, Kenneth (1995). Governing China: from revolution through reform (2nd Edition). New York: WW Norton. Lin, Justin Yifu, and Zhiqiang Liu (2000), Fiscal Decentralization and Economic Growth in China, Economic Development and Culture Change, Vol. 49(1): Lou, Jiwei (2008), The Reform of Intergovernmental Fiscal Relations in China: Lessons Learned, in Public Finance in China: Reform and Growth for a Harmonious Society, edited by Jiwei Lou and Shuilin Wang, The World Bank, Washington D.C. Lindquist, Matthew, Joeri Sol, and Mirjam Van Praag (2012), Why Do Entrepreneurial Parents Have Entrepreneurial Children? mimeo. Murphy, Kevin M., Andrei Shleifer, and Robert W. Vishny (1991), The Allocation of Talent: Implications for Growth, Quarterly Journal of Economics, Vol. 106: National Population and Family Planning Commission of China (2011), Report on China's Migrant Population Development (in Chinese), China Population Publishing House, Beijing. Peng, Yusheng (2004), Kinship Networks and Entrepreneurs in China's Transitional Economy, American Journal of Sociology, Vol. 109: Persson, Torsten, and Guido Tabellini (1999), The Size and Scope of Government:Comparative Politics with Rational Politicians, European Economic Review, Vol. 43: Quadrini, Vincenzo (2000), Entrepreneurship, Saving, and Social Mobility, Review of Economic Dynamics, Vol. 3: Ram, Rati (2009), Openness, Country Size, and Government Size: Additional Evidence from a Large Cross-country Panel, Journal of Public Economics, Vol. 93: Rodrik, Dani (1996), Why Do More Open Economies Have Bigger Governments? NBER Working Papar Reynolds, Paul, David J. Storey, and Paul Westhead (1994), Cross-national Comparisons of the Variation in New Firm Formation Rates, Regional Studies, Vol. 28: Rosenzweig, Mark R., and Kenneth I. Wolpin (1994), Parental and Public Transfers to Young Women and Their Children, American Economic Review : Shelton, Cameron A. (2007), The Size and Composition of Government Expenditure, Journal of Public Economics, Vol. 91: Shih, Victor, Christoper Adolph and Mingxing Liu (2012), Getting Ahead in the Communist Party: Explaining the Advancement of Central Committee Members in China, American Political Science Review, Vol. 106: Tiebout, Charles M. (1956) A Pure Theory of Local Expenditures, Journal of Political Economy 64(5):

26 Wang, Xiaolu, Fan Gang, and Li Feiyue (2012), Business Environment Index for China's Provinces: 2011 Report, China CITIC Press, Beijing. Wang, Shing-Yi (2013), Marriage networks, nepotism, and labor market outcomes in china, American Economic Journal: Applied Economics 5(3), Xu, Chenggang (2011), The Fundamental Institutions of China's Reforms and Development, Journal of Economic Literature 49 (4): Zhang, Tao, and Heng-fu Zou (1998),Fiscal Decentralization, Public Spending, and Economic Growth in China, Journal of Public Economics, Vol. 67, Zhu, Miaokuan (2010), Deter High Consumption with Government Money, National Development and Reform Commission, 25

27 Figure 1: The Spatial Distribution of Government Size in 2005 Notes: This map shows the distribution of government size in 2005 across 28 provinces in the sample. As the east provinces are more developed than the west, it suggests a negative correlation between GDP per capita and government size (with a correlational coefficient of -0.46). However, government size cannot be perfectly explained by GDP per capita. 26

28 Figure 2: The Difference in the Probability of Entrepreneurship (Cadre Parent minus Non-cadre Parents) in Each Province and Government Size, 2005 Notes: This figure shows that the effect of cadre parents on the probability of entrepreneurial children increases with government size 27

29 Figure 3: Probability of Entrepreneurship and Government Size, Notes: This figure shows the national pattern of entrepreneurship and government size from 1980 to 2005, divided into three decades and by the type of parents. 28