Finance and Growth for Microenterprises

Using a survey dataset of Chinese rural households, we find that access to external finance is positively associated with the decision to become entrepreneur, the initial investment for microenterprises and the use of external finance. Also, we find that the use of informal finance is positively associated with sales growth of microenterprises with employees, but not of self-employed. We do not find any significant relationship between the use of formal finance and firm growth. Our findings underline the importance of finance for entrepreneurship and microenterprise growth, and the role of informal finance in the absence of efficient formal financial institutions.


INTRODUCTION
Firms face multiple growth constraints in developing countries, including the lack of access to markets, decrepit energy and transportation infrastructure and lack of security.A large and still growing literature, however, has documented the importance of financing constraints for small firms, especially in less developed countries (Ayyagari et al., 2008;Beck et al., 2005).
A recent controversy has focused on the relative importance of formal versus informal financial providers for alleviating firms' financing constraints in developing countries.Allen et al. (2005) posit that alternative financing channels, such as internal financing and trade credit as well as informal coalitions of firms, investors, and local governments are more important than formal bank credit in fostering the growth of private Chinese enterprises, while Ayyagari et al. (2010) show that it is bank rather than informal financing that fosters firm growth in China.Even less is known, however, about micro-or household-based enterprises, 1This paper gauges the importance of informal and formal financing sources for microenterprises in rural China.Using a unique household survey conducted in 2009, we relate the decision to start a microenterprise and its success to different gauges of access to and use of informal and formal financing.In doing so, this paper complements recent studies that focus on small-and medium-sized enterprises (SMEs), mostly in urban areas of China.especially in rural China.
Access to finance, whether from informal or formal sources, may alleviate credit constraints for initial investment of microenterprises and subsequent expansion.There are important differences, however, between informal and formal finance, related to screening and enforcement mechanisms (Jain, 1999).Banks rely on hard information such as financial reporting, often require collateral which is scarce in rural areas, and rely on government and courts for enforcement, while informal lenders rely on soft information, which is often proprietary, and on personal networks, and enforce loan contracts through reputation, social sanction, and coercion (Besley and Levenson, 1996).
Given the lack of formal financial statements, microenterprises face severe information asymmetries in developing countries, which hinder the credit availability from formal financial institutions, such as banks and credit cooperatives.Informal finance may therefore be essential for microenterprises in China under its special economic and legal institutions (Allen et al., 2005) and especially in rural areas that have experienced a withdrawal of formal financial institutions since the mid-1990s.On the other hand, formal finance might be more effective in alleviating financing constraints, as informal finance carries high interest rates, is of limited scale, and has pro-cyclical trends. 2he evidence on the effect of informal versus formal finance on firm growth is mixed in the literature.On the one hand, Allen et al. (2005) show that alternative financing channels and governance mechanisms support a high growth of the private sector in China.Degryse et al. (2013) show that co-funding of informal and formal finance is an optimal financing choice for firm growth in China as the informal finance has an information advantage while the formal finance has a cost advantage.On the other hand, Ayyagari et al. (2010) use a survey of It is therefore apriori not clear whether micro-and household-based enterprises can benefit more from informal or formal financing sources in rural China, which leaves the question for empirical research.
private firms in China, and find that bank finance is associated with faster growth while informal finance is not. 3e use data from a survey undertaken in 2009 of almost 2,000 households across nine counties in three provinces, 27 percent of which run microenterprises.This allows us to explore both the covariates of the decision to start a microenterprise and the co-variates of the use of informal and formal financial services and the role that the use of informal and formal financial service play in the growth process of microenterprises.Cheng and Degryse (2010) show that the development of banks has a positive impact on provincial economic growth in China, while non-bank financial institutions do not.Furthermore, the relative importance of informal versus formal finance is even less addressed in the literature.Tsai (2004) shows that the intended clients of microfinance continue to use informal finance in rural China, whose persistence may be due to the limited supply of formal credit and the institutional weaknesses of many microfinance programs.However, it's still unclear in the literature whether the widespread use of informal finance in rural China is a blessing or a curse for the microenterprises in the rural area.
Our results suggest that, controlling for characteristics of the family head and village characteristics, wealthier households with larger families and located closer to financial institutions are more likely to start microenterprises, and also have higher initial investment.
The same factors also predict the use of both informal and formal financial services.In addition, we find that higher use of informal finance is associated with higher sales growth for microenterprises with employees, while it has no effect for the self-employed.The use of formal finance, on the other hand, is not associated with higher sales growth, which is consistent with the inefficiency of bank loans in China (Bailey, Huang and Yang, 2011).
These findings confirm the importance of finance for entrepreneurship and firm growth, but also underline the important role that informal financial service providers play in the absence of efficient formal financial institutions.
Our paper contributes to the literature by using firm-and household-level survey data to explore the relationship between access to and use of financial services and the behavior of microenterprises.Beck et al. (2005) show that the small firms' growth rates are most constrained by financial, legal, and corruption obstacles, while financial and institutional development weakens these constraining effects, from which the small firms benefit the most.Beck et al. (Forthcoming) show that banking systems with more effective credit information sharing and higher branch penetration can help reduce the informality of enterprises.In addition, access to financial services can help new entrepreneurs survive beyond the first year, as evidence from Bosnia and Herzegovina shows (Demirgüç-Kunt et al., 2009), and can help enterprises innovate at a faster rate (Ayyagari et al., 2011).Furthermore, using subnational data for Italy, Guiso et al. (2004) show that local financial development enhances the probability that an individual starts a business, favors the entry of new firms, and promotes growth, and these effects are weaker for larger firms.We contribute to this literature by adding empirical evidence on the microenterprises in rural China.
While our paper's inference is based on household survey data, our paper is also related to a series of recent randomized control trials that explore the effect of interventions alleviating micro-entrepreneurs' financing constraints. 44 See Karlan and Morduch (2010) for a literature survey on the randomized controlled experiments on microenterprises.
De Mel et al. (2008) use shocks to capital stock through randomized grants in Sri Lanka, and find that the average real return to capital is much higher than the market interest rate, while a similar exercise in Philippines that expanded credit to micro-entrepreneurs did not show any positive effect on borrowers' business (Karlan and Zinman, 2011).More recent evidence has shown differential effects of credit on individuals and households with different characteristics, linked with different uses of credit.Specifically, Banerjee et al. (2013) and Crepon et al. (2011) use randomized controlled trials in India and Morocco, respectively, and show that only some clients use microcredit loans to start up enterprises.New clients that did not start businesses consumed more non-durable goods, with existing businesses reducing consumption and increasing savings.
Our paper relies on household survey data and thus faces the constraints of not being able to infer causality from cross-sectional data.On the other hand, we capture a larger array of financial service providers across a broader geographic area than randomized controlled trials can capture.While randomized controlled trials can often provide a cleaner identification on the effect of finance, they are also restricted in external validity as the experiments are often conducted in a specific location with limited participant groups.Being aware of the caveats with respect to survey data, we try to provide some first evidence on the role of finance plays in the rural microenterprises in China.As we do not have a clear instrument for the access to or use of finance, we describe our findings as correlations and associations rather than implying causality.Given the limited availability of data for rural China, however, we think that our findings still provide useful insights.
The paper proceeds as follows.Section 2 provides an introduction to the Chinese rural financial market.Section 3 introduces data and summary statistics, and develops the hypotheses and methodology.Section 4 presents the results, and section 5 concludes.

RURAL FINANCIAL MARKETS IN CHINA
Although China has had an annual GDP growth rate of around ten percent during the past three decades, rural areas have lagged behind urban areas, and the gap has been widening over the past decade.During the 1980s and 1990s, the rural areas were the engine of China's high economic growth, especially the town and village enterprises which were collectively owned by community governments (Jin and Qian, 1998).However, rural enterprises have witnessed a sharp decline since the mid-1990s due to a wave of bankruptcies and privatization.In contrast, the privatization of state-owned firms in the urban areas since the late 1990s has shifted the urban economy to a higher growth track.Non-farming activities in the rural areas are often undertaken by private microenterprises, and a large labor force has migrated to the urban areas for jobs since the mid-1990s.The outflow of labor force, most often young and talented, contributed to a further contraction of the rural economy in China.
Generally speaking, there are three types of financial institutions serving the rural credit market: the Agriculture Bank of China (ABC) which is a commercial bank, Rural Credit Cooperatives (RCCs), and the Agriculture Development Bank of China (ADBC) which is a policy bank.However, financial institutions, including the state-owned ABC and ADBC, have witnessed a contraction of branches and business in the rural area since the mid-1990s, which resulted in RCCs as the only provider of formal finance in most rural areas. 55 See Xie (2003) for an overview of the reform of RCCs in China.
RCCs, however, are often plagued by an unclear ownership structure and poor risk management, which often results in substantial non-performing loans6 and even bankruptcies.
Furthermore, RCCs usually focus on agricultural production loans, while loans to microenterprises only constitute a small proportion in its portfolio. 7hina still had a rural population of 700 million at the end of 2011, most of them far away from financial inclusion.An essential reason is the lack of collateral in the rural area.
All the rural land is owned by the collectives (i.e.village government), while farmers do not have full property rights on their houses.Generally speaking, houses can only be traded among farmers within the same village, which prevents their use as collaterals for bank loans.
On the other hand, financial institutions usually do not accept movable property, such as tractors, as collateral in the rural area due to the cumbersome registration and enforcement problems.Consequently, financial institutions mainly serve as depositary tool, which leads to a substantial surplus between the deposit and loans outstanding for RCCs.An outflow of credit from the rural area to the urban area widens the urban-rural gap.
In a nutshell, microenterprises face substantial obstacles in the access to formal finance in rural China.
Due to a lack of formal finance, farmers resort to informal sources, i.e. family members, friends, informal financial institutions, suppliers, and other sources, etc. Informal financial institutions (e.g.underground banks, loan sharks, moneylenders, and ROSCAs) have been prohibited since 1949 due to their usury characteristics.Although some informal financial institutions (e.g.underground banks, loan sharks, and moneylenders) have revived after the country's transition to the market economy after 1978, they are mainly in the urban areas which channel funds to the SMEs.In contrast, ROSCAs have not revived in the rural area after 1978 due to a crack-down of kinship and social network during the communist period between 1949 and 1976, especially the Cultural Revolution towards the end of Mao's reign, which often makes family members and friends the only informal source of credit in the rural area.Potential informal lenders are therefore restricted to a small group of people who often live within the same village.

DATA, HYPOTHESES AND METHODOLOGY
This section describes the survey data, discusses the different hypotheses and econometric methodology we will use to test them, and presents descriptive statistics.
We use a unique survey dataset for Chinese rural households (the "Rural Finance Survey"), conducted in 2009 by National School of Development at Peking University and funded by the Citi Foundation.The survey covered three provinces in China, Heilongjiang (Northeast), Hunan (Central), and Yunnan (Southwest). 88 These provinces are less developed than other provinces in the southeast part of China, and also ranks lower in terms of business environment.Accordingly to the indicators in "Doing Business in China 2008" by the World Bank, the northeast/central/southwest part of China ranks lower than the southeast part in the easiness of doing business, e.g. starting a business, registering property, getting credit, and enforcing contracts.Three counties were selected from each province, and three villages were selected from each county.This resulted in a final sample of 1,951 households across three provinces, nine counties and 81 villages.Provinces and counties were chosen based on representativeness, and villages and households are sampled randomly from the name lists.The sample distribution of households by region is shown in Panel A of Appendix 1.The questionnaire recorded detailed information on formal finance (loans from banks and rural credit cooperatives) and informal finance (loans from family, friends, and informal financial institutions, trade credit, and other sources), both at the startup year and the end of 2008 for microenterprises.9 [Table 1 here] In addition, demographic, income, and wealth profiles of households are also recorded in the survey questionnaire.All variables definitions are listed in Table 1.
Table 2 shows some summary statistics of the dataset.Among the 1,951 households surveyed 533 (27 percent) have a microenterprise.The sample distribution by industry is presented in Panel B of Appendix 1.Among the microenterprises, 70 percent have only the entrepreneur himself / herself, i.e. are self-employed.Among the 144 enterprises with employees, the mean (median) value of the number of employees is 8.37 (2), while the largest enterprise has 300 employees.The sample mean (median) value of the initial investment is 50,967 (10,000) RMB, which is about 7,473 (1,466) USD.The sample mean (median) value of the household income is 20,857 (11,000) RMB in 2007.Compared with the annual household income, initial investments are thus substantially large which indicates the need for external financing.Average growth of microenterprises between 2007 and 2008 was six percent, with a standard deviation of 37 percent, and ranging from -217 to 249 percent.
[Table 2 here] The survey provides us with several gauges of access to and use of informal and formal financial services.The use of informal and formal external finance is measured by dummy variables indicating whether a microenterprise has loans outstanding from informal financial institutions or family members and friends (informal finance), and banks or rural credit cooperatives (formal finance).At the startup year of microenterprises, 39 percent used informal finance, and 16 percent of microenterprises used formal finance.In contrast, at the end of 2008, 19 percent used informal finance, while 12 percent of microenterprises used formal finance.Appendix 2 shows the loan sources by region and industry both for the startup year and 2008 year end.
We also have several proxies for the access to financial services.First, we use the wealth of households, as measured by the value of durable goods 10 as a gauge of internal funds and proxy for possible collateral.Household wealth varies from zero to 684,182 RMB (about 100,000 USD), and the average value is 10,916 RMB (about 1,600 USD).11 10 In the survey questionnaire, the following items are included as durable goods: vehicle; tractor; motor or electro tricycle; other transportation tool; pump; other agriculture machine; motorcycle; electro bicycle; color TV; refrigerator; wash machine; computer; mobile phone; air conditioning; microwave oven; induction cooker; water heater; video recorder; audio system.Following the literature on depreciation rate of physical capital (0.106 by Bischoff and Kokkelenberg (1987); 0.126 by Epstein and Denny (1980); 0.100 by Ping et al. ( 2012)), we use an annual depreciation rate of ten percent calculated in a compound style from the initial purchase year to year 2008.We assume that all durable goods at the end of 2008 were purchased in the initial purchase year.As durable goods may be purchased after the initial purchase year and the purchase price is often increasing over the years, we are estimating a lower bound for the wealth of households at the end of 2008 in the above.In contrast, we can calculate an upper bound value of household wealth by treating all durable goods at the end of 2008 with the purchasing price at 2009 (current price) without any depreciation.Similarly, we can also estimate household wealth at the startup year of the microenterprises.
Second, we use the distance from the nearest financial institution to measure the access to finance.This might not only proxy for the geographic proximity to the formal financial institution, but also for proximity to the informal financial institutions.Informal and formal financial service providers cluster geographically in order to facilitate the banks' collection of soft information (Agarwal and Hauswald, 2010), and it can lead to a higher accessibility of bank loans for informal lenders which may be further intermediated to the final borrowers (Bose, 1998;Hoff and Stiglitz, 1997).As a result, it will lead to a positive association between the distance to informal and formal financial institutions.On average, households are 4.50 kilometers from the nearest formal financial institution.Third, we use several gauges of the access to funds from family members and friends, including the amount of gifts the household has received and given, the family size, and the number of family members and friends that are potential lenders.The amount of gifts received and given ranges from zero to 200,500 RMB with an average of 3,689 RMB.Table 2 shows that the family size, or rather, the number of family members, ranges from one to eleven with an average of 4.12.The number of family members and friends that are potential lenders ranges from zero to 200 with an average of 6.98.
As households' ability to access informal and formal sources of finance might be correlated with other family head characteristics, the relationship between access to finance and the decision to establish an enterprise and its growth might be spurious when ignoring these characteristics.Liu (2005) shows that the Hukou registration (registered as urban / rural citizens) is related with the career choice, i.e. rural citizens are less likely to work in the state sector jobs, and more likely to be self-employed.Li et al. (2008) show that the communist party membership can help private entrepreneurs obtain bank loans and further enhance firm performance in China.Age, education and gender are often related with human capital (Becker, 1993), which may also affect the behavior of microenterprises in rural China.We therefore control for an array of family head characteristics, which include Hukou registration, age, high school education or above, communist party membership, and gender of family head.Specifically, six percent of family heads have an urban Hukou registration; the average age is 49.50 years; twelve percent have an education of high school or above; ten percent have communist party membership; and seven percent of household heads are females.
Village characteristics might also be important for the decision to become an entrepreneur.Fan et al. (2004) show that increased government investment in infrastructure may reduce rural poverty by creating improved employment opportunities in the nonfarm sector in China.As a result, we also control for an array of village characteristics, including access to running water, arable land per capita, and access to bus service in the village.Table 2 shows that 69 percent of villages have access to running water service; the average arable land per capita in the village is 0.92 Mu (about 0.15 Acre); and 39 percent of villages have access to bus service.
Using this household survey, this paper tests several hypotheses related to microenterprises' access to different financing sources.First, we gauge the extent to which the decision to establish a microenterprise (i.e.entrepreneurship) is associated with access to informal and formal financing sources.As initial investments for the microenterprises often surpass self-owned funds, households need to borrow additional funds from informal or formal sources.Access to finance may enhance the decision to become entrepreneurs in rural areas by providing supplementary funds for initial investments for microenterprises.
Firstly, we use the following probit model to test the hypothesis on the entrepreneurship and the access to finance: (1) where i stands for a household, j for village, and k for county and equals 1 if a household has a microenterprise, 0 otherwise.On the right hand side, we have three is the heteroskedasticity robust error term.Given the binary nature of the dependent variable, Microenterprise, the regressions are run with the probit model.
Secondly, we relate the size of the initial investment to access to informal and formal financing sources.While initial investments mainly depend on the industry of microenterprises, access to finance may also affect their size.We use the following OLS regression model: (2) where l denotes industries.Initial Investment is the logarithm of 1 plus the amount of initial investment in RMB at the startup year of a microenterprise.By using logs, we can interpret the coefficient estimates as elasticities.In addition to all family head characteristics, family characteristics, and village characteristics in equation ( 1), we add a village-level variable to the initial investment equation ( 2), Competitors at Startup Year.We also include county and industry specific fixed effects, thus controlling for different investment levels across different counties and, critically, across different industries.
Thirdly, we relate the actual use of informal and formal external finance our proxies of access to finance.Specifically, we regress dummy variables indicating whether a microenterprise used informal or formal external finance in its startup year and in 2008 on our proxies for access to finance, controlling for characteristics of the family head and the family.Specifically, we run the following probit regression: (3) The dependent variable is the use of informal or formal finance at start year or in 2008.
We include the same variables as in the previous two models, including county and industry fixed effects, thus controlling for demand differences across counties and industries.
Finally, access to finance may also be related to the growth of microenterprises after the startup year.Microenterprises with employees may have more investment if they have access to finance (both informal and formal financing sources) and consequently a higher growth rate.However, those without employees may be run for subsistence purpose, and never grow even with sufficient access to finance.Following Ayyagari et al. (2010) and Beck et al. (2005), we gauge whether firms' sales growth is associated with the access to informal and formal financing sources.To test this hypothesis we utilize the following OLS regression: (4) The growth of microenterprise is measured by annual sales growth rate during 2007-2008; Finance includes a set of variables indicating the use of informal or formal finance; Self-Employed is equals one if there are no employees except for the entrepreneur himself / herself.We include the same set of control variables as in previous models, plus several firm characteristics, including Firm Size -measured as the logarithm of 1 plus the sales in 2007and Firm Age -measured as the logarithm of the number of years since start-up.As before, county and industry dummies are included in equation ( 4) to control for growth differences across counties and industries, orthogonal to the decisions of the individual microenterprises.
Table 3 shows the correlation coefficients between the variables.Panel A shows the correlation coefficients for all households, while Panel B shows the subsample results for the households with microenterprises.Panel A shows that households with more wealth in durable goods, higher amounts of gifts received and given, larger family size, and those with a head having urban Hukou registration and high school or above education, are more likely to start a microenterprise.In contrast, households located far away from formal financial institutions, residing in villages without access to running water and bus service, lower arable land per capita, and those with an older or female head are less likely to start a microenterprise.Panel B shows that all finance variables are positively correlated with sales growth except the use of informal finance in the startup year, although all correlation coefficients are statistically insignificant.In addition, all finance variables at startup year are statistically positively, while self-employed is negatively associated with the size of initial investment.Besides, firm size and age are negatively associated with the sales growth rate, which is consistent with stylized facts (Clementi and Hopenhayn, 2006).

RESULTS
This section presents our regressions results.We first gauge the relationship between access to informal and formal financial sources and the decision to become entrepreneur, initial investment and actual use of informal and formal external finance.We then relate the use of informal and formal external finance to firm growth.It is important to stress that we document correlations rather than implying causality with our regressions.

Finance, entrepreneurship and initial investment
Table 4 shows that our proxies for access to both informal and formal financial sources are significantly associated with the likelihood of running a microenterprise.Here we present Probit regressions for co-variates of entrepreneurship.Column (1) presents a regression with characteristics of the family head; column (2) adds family characteristics that proxy for access to financial services; column (3) adds village characteristics; column 4 restricts the model specification in column 2 to the sample that is used in column (3) to account for the potential sample selection problem.Given data constraints, the number of observations varies across the three columns, though our findings are robust to this variation. 12Table 4 here] Columns (2) -(4) show that the travel time to the nearest financial institution is negatively related with the likelihood of running a microenterprise.Specifically, a ten percent lower travel time to the nearest financial institution is associated with a six percent higher likelihood of running a microenterprise.Wealth, as measured by durable goods, is positively associated with the likelihood of running a microenterprise, with a ten percent increase in wealth associated with a one percent increase in the likelihood.Household wealth can be seen as both a proxy for internal financing capacity of microenterprises as well as for possible collateral to receive outside informal or formal funding.The size of the family also enters positively and significantly, suggesting that access to more family members for possible funding or other support (e.g.labor) increases the likelihood that households run their own microenterprises.On the other hand, the number of family members and friends as potential lenders is not significantly associated with the likelihood of running a microenterprise.
Similarly, gifts received and given enters only weakly significant (at most at the ten percent level) with a positive sign.Together, these results suggest that access to both formal (financial institutions) and informal sources (family members) as well as own funds (household wealth) are important for the decision to start a microenterprise in China.
Turning to the control variables, we note that -not surprisingly -households in urban areas (i.e.family heads with urban Hukou registration) are more likely to run a microenterprise.Among the other characteristics of the family, only age enters consistently across the regression with a significantly negative coefficient; families with older family heads are less likely to run a microenterprise.Families whose heads have education levels of high school or above are more likely to run microenterprises, although the coefficient is not significant in columns ( 2) -( 4) when we control for family and village characteristics.
Families whose heads are female are six to seven percent less likely to run microenterprises, although the coefficient enters at most at the ten percent significance level.Party membership and thus political connection cannot explain whether a household has a microenterprise or not.
Among the village variables, only the dummy variable indicating bus service enters positively and significantly.Neither access to running water nor the average arable land per capita is significantly correlated with the probability of running a microenterprise.
The regressions in Table 5 show that access to informal and formal financial services is also important for the size of the initial investment for the microenterprises.Column (1) presents a regression with characteristics of the family head; column (2) adds family characteristics that proxy for the access to financial services; column (3) adds village characteristics; column 4 restricts the model specification in column 2 to the sample that is used in column (3) to account for the potential sample selection problem.
[Table 5 here] Here, we use OLS regressions to relate family head, family and village characteristics to the size of the initial investment in an microenterprise.Starting with the regressions in this table, we now work with the smaller sample of micro-entrepreneurs.The results suggest that the size of the initial investment increases with the wealth of the household, as measured by durable goods, with the gifts that the family has received and given, and the size of the family.
The size of the initial investment decreases with the distance to the nearest financial institution, while the number of family members and friends as potential lenders does not enter significantly.Interestingly and unlike in Table 4, none of the characteristics of the family head enters significantly.Among the village characteristics, the competition at the start of the enterprise, as measured by the log one plus the number of families with similar microenterprises in the village enters significantly and negatively, suggesting that households are investing less if there are already many other similar enterprises around, which could be both demand-driven in the product market (e.g.caution in investment decision) and supplydriven in financial market (e.g.caution by lenders).None of the other village characteristics enters significantly.As for the decision to establish or not a microenterprise, the size of the initial investment thus depends critically on the access to own funds, informal and formal external funds.
Table 6 shows that access to external financial resources explains the use of informal or formal finance in the startup year of the enterprise, but not in 2008.Here we regress dummy variables for the use of external informal or formal finance on our set of explanatory variables, both in the startup year and in 2008.The distance to the nearest financial institution can explain the use of external finance only in the startup year of the microenterprise, but not in 2008.This suggests that close geographic proximity is important for startup microenterprises but not for more established ones.One potential explanation is that mature firms often have more internal finance and collaterals than new startups, which makes the access to external finance easier for mature firms.
Consistent with the previous findings, we find that wealthier households are more likely to have had informal or formal external finance in the startup year of the microenterprise and in 2008, although it is insignificant in column (3) of [Table 6 here] Most of the characteristics of the family head do not enter significantly in the regressions, at least not in a consistent manner.An urban Hukou registration is negatively associated with the use of informal finance at startup year and positively associated with the use of formal finance at startup year, while the age of family head is negatively associated with the use of informal finance in both startup year and in 2008.

Finance and growth for microenterprises
The results in Table 7 show that the use of informal but not of formal external finance is related with higher firm growth, except for self-employed entrepreneurs.Here we use OLS regressions to relate the sales growth between 2007 and 2008 with the dummy variables indicating the use of formal and/or formal finance plus their interaction with a dummy variable for self-employed micro-entrepreneurs, i.e. entrepreneurs who do not have any employee from outside their immediate family.The results in column (1) of Table 7 shows that the use of informal finance is associated with faster firm growth in the case of multipleemployee firms, while the results in column (2) show that the use of formal finance is not related to firm growth.Both the informal finance dummy and its interaction with self-employed dummy enter significantly, with coefficients that are of similar size but of opposite sign (column (1)).This suggests that the use of informal external resources help enterprises grow, but only if they have employees outside the immediate family.Neither the formal finance dummy nor its interaction with self-employed dummy enter significantly (column (2)), while the combination of formal / informal external finance dummy enters positively and significantly and its interaction with self-employed dummy enters negatively but only marginally significant (P-value is 0.124 in column (3)).We get similar results in columns ( 4) -( 6) when controlling for family head characteristics. 13 [Table 7 here] The P-value of the interaction for informal finance dummy and self-employed dummy is 0.148 in column (4), and the P-value of the interaction for finance dummy and self-employed dummy is 0.229 in column ( 6).
Besides, we also find similar results in columns ( 7) -( 9) when controlling for village characteristics, where both the coefficients for informal finance dummy and its interaction with self-employed dummy are statistically significant.
The results are not only statistically, but also economically significant.Using informal finance in the case of multiple-employee enterprises is associated with 17 percent higher growth, which compares to a mean growth rate of six percent and a standard deviation of 37 percent.In contrast with Ayyagari et al. (2010), we find that informal finance may enhance the growth of firms with employees while formal finance is irrelevant for the growth of all firms.It is important to note the difference in samples, however.Ayyagari et al. (2010) use a sample of relatively larger firms with a minimum of 20 (15) employees in the manufacturing 13 We also run the regression when controlling for family characteristics, i.e.Durable Goods, Travel Time to Financial Institution, Family and Friends as Potential Lenders, Gift, and Family Size.We find similar results with a slight difference in magnitude and significance.The same case applies to Table 8 when we use loan size as finance variables.The results are suppressed for conciseness but are available upon request.
(service) industry, of which 22% are state-owned and 40% are registered as corporations.In contrast, 70% of our sample firms are self-employed microenterprises (i.e.without employees), while only 12 firms have more than 20 employees (see footnote 1).In addition, all microenterprises are privately owned and registered as household-based enterprises (i.e. the so-called "individually-owned business").The behavior of microenterprises is essentially different with the SMEs and large firms, which may explain the conflicting results with Ayyagari et al. (2010).
The results in Table 8 confirm our previous findings that informal financing is associated with faster growth of enterprises with non-family employees, while formal finance is not.Rather than dummy variables for the use of informal and formal finance, here we use the logarithm of one plus total informal, total formal and total external financing, as well as their interaction with a self-employed dummy.While the amount of total informal financing enters positively and significantly, its interaction with the self-employed dummy enters negatively and significantly (column (1)).We find the same results for the log of one plus total external financing and its interaction with the self-employed dummy (column (3)), while neither the amount of formal financing nor its interaction with the self-employed dummy enter significantly in our regression (column (2)).We find similar results in columns ( 4) -( 6) when controlling for family head characteristics, although the sign of informal finance in column (4) turns insignificant (P-value is 0.105).Furthermore, we find similar results when controlling for village characteristics in columns ( 7) -( 9) with a slight difference in magnitude and significance.
[Table 8 here] Together, the results in Tables 7 and 8 suggest that growth of microenterprises in rural China can be mainly explained by the use of informal external finance, but not by the use of formal external finance.These results stand in contrast to our findings from the previous section that access to both informal and formal finance can explain whether households start their own enterprise.They suggest an important advantage for informal financial service providers in screening and monitoring creditworthy micro-entrepreneurs, while they shed doubt on the ability of formal financial institutions in rural China, especially Rural Credit Cooperatives, to contribute to firm growth.They underline the important role of informal finance in the presence of regulatory restrictions on formal financial institutions.
While we find that the access to formal finance (i.e.distance to the nearest financial institution) is positively associated with the decision to become an entrepreneur and the size of the initial investment, we do not find any significant effect of the use of formal finance (i.e. loans outstanding from banks and credit cooperatives) on the growth of microenterprises.
One potential reason is that households may be severely constrained in obtaining finance at the start-up year for the microenterprises, which makes the access to finance (i.e. both informal and formal finance) essential for the decision to become an entrepreneur.Banks may turn out be to inefficient lenders with the expansion of microenterprises (e.g.due to information asymmetry and connected borrowers), while informal financial institutions fill in the gap using their proprietary information.Another reason could be that banks lend to informal financial institutions, which further intermediate loans to borrowers (Hoff and Stiglitz, 1997).As a result, the access to formal finance is positively associated with the access to informal finance (i.e.see panel A of Table 3 for the negative correlation between the travelling time to the nearest financial institution and the number of family members and friends as potential lenders) and the use of informal finance (i.e.see column (1) of Table 6).
Thus, the access to formal finance may enhance the start-up decision of microenterprises and the size of initial investment.Finally, the access to formal finance may proxy the access to markets in general, i.e. the distance to the nearest financial institution is often correlated with the distance to the major markets, which may enhance the decision to become entrepreneurs and the size of initial investment.In sum, the access to formal finance is different with the use of formal finance, which may lead to diverging results among the start-up decision, the initial investment, and the subsequent growth of microenterprises.

CONCLUSION
The effect of access to finance on entrepreneurship and microenterprise growth is of vital importance for poverty reduction and economic growth in developing countries.Using a survey dataset on Chinese rural households, we find that access to finance is positively associated with the decision to start microenterprises and with the size of initial investment.
In addition, we find that the use of informal finance is associated with higher growth of rural microenterprises with employees, while there is no effect for those without (i.e.selfemployed).We do not find any significant relationship between the use of formal finance and firm growth.Increasing the competition in the credit market, for example, may be a solution (Chong et al., 2013;Park et al., 2003).On the one hand, a fiercer credit market competition can drive banks and credit cooperatives to expand the borrower set through more investment in relationship lending (in a similar style as informal financial institutions), which may help pick those "good" borrowers in the rural area (i.e.those households running microenterprises with external employees).One the other hand, it may also push these financial institutions to provide better service for existing borrowers, i.e. simpler loan applications procedures, faster loan approval process, and less corruption of loan officers.While it is infeasible to eliminate the informal finance, the formal financial institutions may utilize its cost advantage (i.e.lower interest rate than informal finance) to provide complementary funding for those rural households in shortage of fund.For column (1), the dependent variable is Informal Start Year which is a dummy variable which equals 1 if a microenterprise has informal finance at the start year, 0 otherwise; for column (2), the dependent variable is Formal Start Year which is a dummy variable which equals 1 if a microenterprise has formal finance at the start year, 0 otherwise; for column (3), the dependent variable is Informal which is a dummy variable which equals 1 if a microenterprise has informal finance at the end of 2008, 0 otherwise; for column (4), the variable is Formal which is a dummy variable which equals 1 if a microenterprise has formal finance at the end of 2008, 0 otherwise.Hukou Urban equals 1 if a family head has a urban Hukou registration, 0 otherwise; Head Age is the logarithm of the family head's age; Senior Education equals 1 if a family head has an education of high school or above, 0 otherwise; Political Connection equals 1 if a family head is party member, 0 otherwise; Female equals 1 if a family head is female, 0 otherwise; Durable Goods is the logarithm of 1 plus the value of durable goods at the start year for column (1)-( 2), and the logarithm of 1 plus the value of durable goods at end of 2008 for column (3)-(4); Travel Time to Financial Institution is the logarithm of 1 plus travel time in minutes from a family to the nearest financial institution; Family Members and Friends as Potential Lenders is the logarithm of 1 plus the number of family members and friends who are potential lenders; Gift is the logarithm of 1 plus gift received and given in 2008; Family Size is the number of family members; Running Water equals if a village has running water service; Land Per Capita in the Village is the average arable land per capita in a village; Bus equals 1 if a village has bus service, 0 otherwise.Marginal effects and Pseudo R-squared are reported in the table.All regressions include county and industry dummies.Heteroskedasticity robust standard errors are in brackets, significance * at 10%, ** at 5%, *** at 1%. (1) ( Our findings are complementary to the literature on the relationship between informal and formal finance and firm growth in China.Unlike previous papers we focus on rural China and micro-or household-based entrepreneurs.Our findings provide important insights into the relative importance of different financial service providers in rural China.They confirm previous evidence on the dearth and inefficiency of formal financial institution in rural China and point to the important role of informal financial service providers in filling the gap left by banks and credit cooperatives.China's government has been actively pushing for the development of rural areas and non-export domestic sectors.Our findings point to the important role that access to financial services should have on this agenda.They point to the urgency of strengthening the contractual and regulatory frameworks for formal financial institutions in rural areas.The formal financial institutions have not played a proper supporting role in rural China since the retreat of branches in the mid-1990s.There are a number of things that the financial institutions and government should do in order to promote the development of microenterprises and poverty reduction in rural China. categories of variables: family head characteristics (Hukou Urban, Head Age, Political Connection, Senior Education, and Female), family characteristics (Durable Goods, Travel Time to Financial Institution, Family Members and Friends as Potential Lenders, Gift, and Family Size), and village characteristics (Running Water, Land Per Capita in the Village, and Bus).The county dummies control for other important cultural and socio-economic differences across counties and provinces that are not captured by the village-level variables.
Table 6 for informal finance in 2008.Family size can explain whether firms use informal finance in the startup year and in 2008, while gifts can, somewhat surprisingly, explain the use of formal finance both in the startup year and in 2008, but not the use of informal finance either in startup year or in 2008.As before, family members and friends as potential lenders cannot explain the use of informal or formal external finance neither in the startup year nor in 2008.

Table 2 .
Summary StatisticsMicroenterprise equals 1 if a family has a microenterprise, 0 otherwise.Hukou Urban equals 1 if a family head has an urban Hukou registration, 0 otherwise; Head Age is the logarithm of a family head's age; Senior Education equals 1 if a family head has an education of high school or above, 0 otherwise; Political Connection equals 1 if a family head is party member, 0 otherwise; Female equals 1 if a family head is female, 0 otherwise; Durable Goods is the logarithm of 1 plus the value of durable goods; Travel Time to Financial Institution is the logarithm of 1 plus the travel time in minutes from a family to the nearest financial institution; Family Members and Friends as Potential Lenders is the logarithm of 1 plus the number of family members and friends who are potential lenders; Gift is the logarithm of 1 plus the gift received and given in 2008; Family Size is the number of family members; Running Water equals if a village has running water service; Land Per Capita in the Village is the average arable land per capita in a village; Bus equals 1 if a village has bus service, 0 otherwise.Sales Growth 2008 is the annual sales growth rate calculated as log difference between sales in 2008 and sales in 2007; Initial Investment is the logarithm of 1 plus the initial investment in the startup year of a microenterprise in RMB; Informal equals 1 if the microenterprise has informal finance at the end of 2008, 0 otherwise; Formal equals 1 if a microenterprise has formal finance at the end of 2008, 0 otherwise; Finance equals if a microenterprise has either informal or formal finance at the end of 2008, 0 otherwise; Informal at Start Year equals 1 if a microenterprise has informal finance at start year, 0 otherwise; Formal at Start Year equals 1 if a microenterprise has formal finance at start year, 0 otherwise; Finance at Start Year equals 1 if a microenterprise has either informal or formal finance at start year, 0 otherwise; Informal Finance Size equals the logarithm of 1 plus informal finance at the end of 2008 are positive; Formal Finance Size equals the logarithm of 1 plus formal finance at the end of 2008 are positive; Finance Size equals the logarithm of 1 plus the informal and formal finance; Self-Employed equals 1 if a microenterprise does not have other employees excluding the entrepreneur himself / herself, 0 otherwise; Firm Size is the logarithm of 1 plus the sales in 2007; Firm Age is the logarithm of the number of years since firm startup.

Table 3
Panel A. Correlation Table for Microenterprise.Microenterprise equals 1 if a family has a microenterprise, 0 otherwise; Initial Investment is the logarithm of 1 plus the initial investment in the startup year of a microenterprise in RMB; Hukou Urban equals 1 if a family head has an urban Hukou registration, 0 otherwise; Head Age is the logarithm of a family head's age; Senior Education equals 1 if a family head has an education of high school or above, 0 otherwise; Political Connection equals 1 if a family head is party member, 0 otherwise; Female equals 1 if a family head is female, 0 otherwise; Durable Goods is the logarithm of 1 plus the value of durable goods; Travel Time to Financial Institution is the logarithm of 1 plus the travel time in minutes from a family to the nearest financial institution; Family Members and Friends as Potential Lenders is the logarithm of 1 plus the number of family members and friends who are potential lenders; Gift is the logarithm of 1 plus the gift received and given in 2008; Family Size is the number of family members; Running Water equals if a village has running water service; Land Per Capita in the Village is the average arable land per capita in a village; Bus equals 1 if a village has bus service, 0 otherwise.* Significance at 5% level.

Correlation Table for Initial Investment and Sales Growth.
Initial Investment is the logarithm of 1 plus the initial investment in the startup year of a microenterprise in RMB; Sales Growth 2008 is the annual sales growth rate calculated as log difference between sales in 2008 and sales in 2007; Informal equals 1 if a microenterprise has informal finance at the end of 2008, 0 otherwise; Formal equals 1 if a microenterprise has formal finance at the end of 2008, 0 otherwise; Finance equals if a microenterprise has either informal or formal finance at the end of 2008, 0 otherwise; Informal at Start Year equals 1 if a microenterprise has informal finance at start year, 0 otherwise; Formal at Start Year equals 1 if a microenterprise has formal finance at start year, 0 otherwise; Finance at Start Year equals 1 if a microenterprise has either informal or formal finance at start year, 0 otherwise; Self-Employed equals 1 if a microenterprise does not have other employees excluding the entrepreneur himself / herself, 0 otherwise; Firm Size is the logarithm of 1 plus the sales in 2007; Firm Age is the logarithm of the number of years since the startup of a microenterprise.* Significance at 5% level.

Table 4 .
Probit Regressions for Microenterprise.Dependent variable is Microenterprise, which equals 1 if a family has a microenterprise, 0 otherwise.Hukou Urban equals 1 if a family head has an urban Hukou registration, 0 otherwise; Head Age is the logarithm of a family head's age; Senior Education equals 1 if a family head has an education of high school or above, 0 otherwise; Political Connection equals 1 if a family head is party member, 0 otherwise; Female equals 1 if a family head is female, 0 otherwise; Durable Goods is the logarithm of 1 plus the value of durable goods; Travel Time to Financial Institution is the logarithm of 1 plus the travel time in minutes from a family to the nearest financial institution; Family Members and Friends as Potential Lenders is the logarithm of 1 plus the number of family members and friends who are potential lenders; Gift is the logarithm of 1 plus the gift received and given in 2008; Family Size is the number of family members; Running Water equals if a village has running water service; Land Per Capita in the Village is the average arable land per capita in a village; Bus equals 1 if a village has bus service, 0 otherwise.Marginal effects and Pseudo Rsquared are reported.Column (4) restricts the specification of column (2) to the sample of column (3).All regressions include county dummies.Heteroskedasticity robust standard errors are in brackets, significance * at 10%, ** at 5%, *** at 1%.

Table 5 . OLS Regressions for Initial Investment.
Dependent variable is the logarithm of 1 plus the initial investment for the microenterprise in RMB.Hukou Urban equals 1 if a family head has an urban Hukou registration, 0 otherwise; Head Age is the logarithm of a family head's age; Senior Education equals 1 if a family head has an education of high school or above, 0 otherwise; Political Connection equals 1 if a family head is party member, 0 otherwise; Female equals 1 if a family head is female, 0 otherwise; Durable Goods is the logarithm of 1 plus the value of durable goods; Travel Time to Financial Institution is the logarithm of 1 plus travel time in minutes from a family to the nearest financial institution; Family Members and Friends as Potential Lenders is the logarithm of 1 plus the number of family members and friends who are potential lenders; Gift is the logarithm of 1 plus gift received and given in 2008; Family Size is the number of family members; Competitors at Start Year is the logarithm of 1 plus the number of families with similar microenterprises within a village; Running Water equals if a village has running water service; Land Per Capita in the Village is the average arable land per capita in a village; Bus equals 1 if a village has bus service, 0 otherwise.Column (4) restricts the specification of column (2) to the sample of column (3).All regressions include county dummies.Heteroskedasticity robust standard errors are in brackets, significance * at 10%, ** at 5%, *** at 1%.