WPS7454 Policy Research Working Paper 7454 Dual Credit Markets and Household Access to Finance Evidence from a Representative Chinese Household Survey Robert Cull Li Gan Nan Gao Lixin Colin Xu Development Research Group Finance and Private Sector Development Team October 2015 Policy Research Working Paper 7454 Abstract Using a new and representative data set of Chinese house- interest rates on loans similar to urban residents. Younger hold finance, this paper documents household access to residents pay higher rates, while households on firmer eco- and costs of finance, along with their correlates. As in most nomic footing face lower rates. Taking financial classes and developing countries, informal finance is a crucial element college education is associated with higher interest rates for of household finance, and wealth tends to be associated with urban residents, suggesting perhaps that financial knowl- better access to formal and informal finance. Better financial edge coincides with greater demand for credit in areas with knowledge shifts loan portfolios toward formal sources rela- more economic opportunity. Overall, the findings suggest tive to informal ones. Connections to the Communist Party that Chinese residents face dual credit markets, with the are associated with significantly better access to finance in poor, young, those with poor financial knowledge, and those rural areas but not in urban areas. A larger social network is with larger family sizes relying much more on informal positively associated with access to informal finance. Con- finance, while others are better able to access formal finance. trolling for household characteristics, rural residents pay This paper is a product of the Finance and Private Sector Development Team, Development Research Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at lxu1@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Dual Credit Markets and Household Access to Finance: Evidence from a Representative Chinese Household Survey1 Robert Cull World Bank Li Gan Texas A&M University and SWUFE Nan Gao SWUFE Lixin Colin Xu World Bank JEL codes: G00, O17, O53, P34. Key words: access to finance, costs of finance, household finance, political connections, financial knowledge, dual credit markets.                                                                 *  We have benefited from comments of Daniel Berkowitz, Jing Cai, Bert Hoffman, Xiaoning Long, Claudia Ruiz,  Thierry Tressel, and other participants of the Chinese Economist Society meeting at University of Michigan,  and a seminar at Xiamen University. The views are the authors and do not necessarily reflect those of the  World Bank, its executive directors, or the countries they represent.    Corresponding author: L. Colin Xu, World Bank, MC 3‐307, 1818 H Street, N.W. Washington, DC 20433;  lxu1@worldbank.org.  I. Introduction Household finance has always been an important lever to influence key policy outcomes. China, for instance, is hoping to increase household consumption to reduce its reliance on exports for growth, and household finance is clearly instrumental for purchase decisions for housing, durables, education, and medical care. Another key policy objective emphasized by the Chinese government in the past two years has been the facilitation of creation of new businesses, especially small and medium-sized ones. Again, only with good access to finance can households start businesses. Yet, partly due to the lack of detailed data on household usage of financial services, we have little in-depth knowledge about household financial arrangements in developing countries such as China. In this paper, we rely on a comprehensive new national data set to provide a more complete description of Chinese household finance. In particular, we address what determines a household’s access to finance. Why do certain segments of the society such as rural residents have significantly less access to finance? How do households use formal and informal finance differently? Do those factors that facilitate access to finance also facilitate lower costs of finance? What are the roles of political and private networks in facilitating access to finance? Does financial knowledge affect access to finance? The data we use, the Chinese Household Finance Survey (CHFS), is designed to be representative of China through a multi-stage stratified random sampling process. It covers 29 provinces, 262 counties, 28,100 households and 98,000 individuals, and represents the best data source available for studying the above questions. Our study contributes to an emerging literature on the relative importance of informal versus formal finance in explaining China’s recent growth. Some studies ascribe more importance to the role and efficiency of informal sources of finance in explaining the growth of Chinese firms (Allen et al., 2005) than others (Ayyagari et al., 2010; Cull and Xu, 2005; Cull 2    et al., 2009), though in general results indicate that many firms rely on both sources. Our analysis is most similar to others for China that examine how political connections and business and personal characteristics affect the sources of financing used by owners of small private firms (Tsai, 2002; Zhang, 2008). Zhang (2008), for instance, uses survey evidence to examine reliance on formal versus informal financing in Chengdu, one of China’s most important inland cities. He finds that proxies for reputation and relationships play crucial roles in explaining firms’ usage of formal finance, though those factors are typically associated with usage of informal finance. Because our survey is more detailed, we are better able to test whether proxies for social networks and other household characteristics related to reputation and relationships affect usage of both formal and informal finance. Another difference between that study and ours is that Zhang (2008) focuses on small private firms whereas the household is our unit of observation, though we admit that household and small business finance are often deeply intertwined. However, our key advantage is that we rely on nationally representative data for China enabling us to compare financial usage patterns across regions and to examine usage by different types of households within the same types of environments (e.g., rural versus urban). Our investigation yields a number of insights about how household finance works in China, many of them similar to those in other developing countries, but others that are distinctly Chinese. As in other contexts, informal finance proves to be of critical importance for our understanding of household finance, especially in rural areas. For example, we find that a larger social network (as measured by the number of siblings of the household head and his/her spouse) is positively associated with access to informal finance. Informal social networks can facilitate trade where legal enforcement of property rights is weak or uncertain (Grief, 1997; McMillan, 1997). For example, social networks based on family ties have been shown to support extension of credit in Thailand (Townsend, 1995) and Vietnam (McMillan and Woodruff, 1999). Given the nascent state of formal credit markets in many of the locales that we study, informal credit 3    proves to be relatively cheap, and with longer maturities. However, a small share of informal loans carry very high interest rates. Those loans highlight the limits of informal finance between parties that do not know each other well.2 A branch of the literature on firm creation and growth has emphasized the importance of government connections, particularly in obtaining access to finance. In developing economies, politically-connected firms have better access to loans (Li et al., 2008, Claessens et al., 2008, Fan et al., 2008) and to equity markets (Boubakri et al., 2012), and are more likely to be bailed out when facing financial stress (Faccio et al., 2006). In the Chinese context, government intervention has long biased credit allocation toward state-owned enterprises.3 Among private Chinese firms, political connections are also associated with better access to credit (Choi and Zhou, 2001; Cull et al., 2015) and to equity markets (Francis et al., 2009), and a growing literature indicates that government connections play a key role in explaining firm investment behavior (Chow and Fung 1998, Héricourt and Poncet 2009, Poncet, Steingress, and Vandenbussche 2010, Guariglia, Liu, and Song 2011, Cull et al., 2015) and profitability (Choi and Zhou, 2001). Again, however, our focus is on household usage of financial services. Our analysis confirms that connections to the Communist Party, the dominant party in China, are associated with better access to finance in rural areas (though, interestingly, not in urban areas). Our findings also contribute to a growing literature on how financial and economic knowledge affect usage of financial services by individuals and households, which in turn can affect their economic outcomes. For example, studies have shown that financial knowledge is linked to higher savings levels, better retirement planning, and better investment decisions (e.g.,                                                                 2   In contrast, McMillan and Woodruff (1999) find that business networks comprised of unrelated members  who knew each other well were effective in promoting trade credit in Vietnam, even more so than family‐ based social networks.    3   Brandt and Li 2003; Huang, 2003; Bai et al. 2006a; Li et al. 2008; Cull, Xu and Zhu 2009; Gordon and Li 2003,  2011.  4    in terms of diversification).4 Moreover, recent meta-analyses of studies of financial education generally show positive impacts of such training on financial outcomes (Fernandes et al., 2014; Miller et al., 2015). Although much is still to be learned about how the content, timing, and delivery of that training affects financial outcomes, a number of the most recent studies that employ rigorous identification methods such as randomized control trials (RCTs) also find positive impacts. Financial training is also likely to be potentially relevant in many environments, since cross-country evidence shows that a large share of the population struggles to understand basic financial concepts, even in developed countries (Lusardi and Mitchell, 2012). These factors also appear to be important in the Chinese context. Li, Rozelle, and Zhang (2004) find that a minimum level of educational competence is necessary for understanding the terms of micro-loans in China (durations, interest rates, and repayment conditions) and for reading the passbooks linked to those micro-credit accounts. And recent RCT evidence shows that financial education for rural farmers in China increased their uptake of insurance (Cai et al., 2015). We find that better financial knowledge is associated with loan portfolios that are tilted toward formal sources at the expense of informal loans. However, taking financial classes and college education is associated with higher interest rates on bank loans for urban residents, suggesting perhaps that financial knowledge coincides with greater demand for credit in areas with more economic opportunity. Household characteristics also play an important role in describing variation in usage of financial services. First and foremost, we find that household wealth (or its proxies) tends to be affiliated with better access to finance. While this is not surprising, it raises questions about how wealth can be pledged as collateral or whether wealth is simply a summary statistic for                                                                 4   See Miller et al. (2015) for an overview.  5    other factors that indicate creditworthiness to potential lenders. Controlling for household characteristics, rural and urban residents pay surprisingly similar rates on loans. Less surprisingly, younger residents on average pay higher rates, while households on stronger economic footing face lower interest rates. Our data also enable us to examine how one key feature of a household’s locale, the density of bank branches in close proximity, is associated with usage of financial services. In line with studies of bank expansion on the uptake of financial services in developing countries (Burgess and Pande, 2005; Bruhn and Love, 2014), we find that greater branch density is positively associated with usage of loans, but only in the urban areas that we study. In short, this paper makes contributions to several strands of literature. First, while more is known about household/consumer finance in developed countries, we provide evidence on household financial usage from a key emerging economy. In particular, we provide detailed information on access to formal and informal finance, along with associated financing costs, and we document rural-urban differences in access to and costs of financing. Second, the paper offers insights about how access to political and social networks affects access to and costs of household finance (as opposed to firm finance), and what factors determine the relative attractiveness of formal and informal finance. Third, this paper provides fresh evidence of how financial knowledge influences access to and costs of financing. Finally, this paper offers a comprehensive picture of household finance in China, bringing to light details not seen in earlier portraits due to less representative data. Overall, this paper suggests that Chinese residents face dual credit markets, with the poor, young, politically underprivileged, those with poor financial knowledge, and those with larger family sizes relying much more on informal finance, while those that fare better on those dimensions have greater access to formal finance. 6    II. Data, Variables, and Conjectures We rely on the CHFS of 2013, which was collected by the Chinese Household Finance Research Center of Southwestern University of Finance and Economics. The data set is representative of Chinese households due to its stratified random sampling design in three stages: first, random sampling of counties; then communities within a county; finally random sampling from within a community.5 When choosing sample counties, all were first ranked according to GDP and classified into groups. Within each GDP group, a random sample of counties was then selected. In the end, a total of 262 counties, roughly 10% of the total number in China, were selected to be part of the sample. Within each county, four communities were chosen. And within a community, 25-50 households were chosen for urban areas, and 20 households for rural areas. The final CHFS sample consists of 28,100 households and 98,000 individuals in 1,029 communities, from 262 counties, spanning 29 provinces. The CHFS provides detailed information on households’ demographic characteristics, assets and liabilities, insurance and social protection, and income and expenditures. More importantly for our research purposes, the CHFS contains detailed information on household liabilities, such as debts incurred for agricultural production, for non-agricultural production of businesses owned by family members, for purchases of cars and housing, for schooling tuition, and many other purposes. Furthermore, the survey covers not only borrowing from banks but also debt from informal channels such as relatives, friends, and informal financial organizations. Moreover, we have information on whether the maturities of specific debts exceed one year, which allows us to classify some loans as long-term.6                                                                 5   A community is either a village in rural areas or a neighborhood committee (Ju wei hui in Chinese) in urban  areas.  6 For some loans, there are direct questions about maturity. For others, we know the timing of the debt, and  whether it has an outstanding balance. If the loan has an outstanding balance and was incurred at least a year  earlier, we classify it as long‐term. 7    We focus on 12 key indicators of financial usage, covering both access and costs. See Table 1 for a list of our variables and their definitions. First, Loan is a dummy variable indicating that the household has borrowed either from banks or from informal sources (which include loans from relatives and friends and other informal lenders). To capture the sources of loans, we rely on two variables, Bank Loan (an indicator that a household currently has at least one loan from a bank), and Informal Loan (indicating a current loan from informal sources). Second, since long-term finance is often particularly useful for households when they need to invest for the future or when dealing with risks (World Bank, 2015), we use three corresponding indicators of access to long-term finance: Long-Term Loan, Long-Term Bank Loan, and Long-Term Informal Loan. We consider a loan long term if its maturity exceeds one year. In large part, we define the long-term cutoff at one year because there are too few instances of households with loans carrying maturities much longer than that to permit meaningful analysis. We also have six variables capturing the costs of loans: for each type of loan, we also have the associated interest rate, which is used to capture the cost of various types of finance. We have several groups of variables that characterize a household’s basic characteristics, wealth, education, social network, political access, and financial and economic literacy. For basic household characteristics, we include the age of the household head, whether he/she works, the number of children (age 16 or younger), and the number of adults of working age (between 17 and 65). We also have several proxies of household wealth: whether the household has farmable land (Farmable Land),7 whether the housing unit in which the household is dwelling was inherited from parents or purchased at a subsidized price from the current or former employer                                                                 7   This variable is available for both urban and rural households. Due to the blurred nature of the urban/rural  classification, sometimes urban residents also have farmable land.  8    (Inherited Housing),8 and whether the household head is a manager in his/her place of work (Manager). The education level of the household is proxied by whether the household head graduated from high school (High School), or from a college (College). Since wealthy households have better collateral, longer credit histories that can be documented, and established relationships with both formal financial institutions and informal lenders, we expect them to have better access to finance in general. This implies not just better access to the overall level of finance, but also on better terms. To describe a household’s social network, we rely on the total number of siblings that the household head and his/her spouse have (“Number of Siblings”).9 We do this because it is natural to first borrow informally from siblings, who likely possess knowledge about one’s creditworthiness. Altruistic instincts and social norms might also induce them to lend at relatively low rates. Parents can also sometimes act as enforcement agents when default or other conflicts arise out of the credit relationship. Within-family lending can also act as an informal insurance mechanism (Townsend, 1995; Gan et al., 2012), especially where formal insurance markets are not well developed and where financial constraints are severe (both features germane to China). For political access, we rely on an indicator of whether any of the household members are affiliated with the ruling Communist Party. While no research has been conducted on how access to political power facilitates household finance in China, an emerging literature has shown that firms with strong political connections tend to have better access to bank finance, both within China (Bai et al., 2006a,b; Brandt and Li, 2003; Cull et al., 2015; Li et al., 2008 ) and in other developing countries (Shleifer and Vishny, 1994; Sapienza, 2004; Johnson and                                                                 8   Starting in the second half of the 1990s, the state sector in China’s urban areas began a series of housing  reforms in which employers sold (formerly free) apartments to employees at substantially subsidized prices  (Wang, 2011).  9   Yuan and Xu (2015) have also shown that the number of siblings is positively associated with usage of  informal credit for a smaller sample of rural Chinese households than the one we use in this paper.  9    Mitton, 2003; Dinc, 2005; Khwaja and Mian, 2005; Claessens, et al., 2008). A growing literature underscores the importance of financial and economic knowledge on financial access and financial decisions. 10 We rely on four proxies for financial/economic literacy. First, Risk for Return is a dummy variable indicating that the survey respondent prefers a lottery of higher expected return (but greater risk) to a risk-free return of lower value. Specifically, the respondent is presented with two options, one being 100% chance of getting 4,000 RMB, the other being a 50% chance of getting 10,000 RMB and 50% chance of getting nothing. Risk for Return is one if the respondent picked the second portfolio. Second, Fin Class is a dummy variable indicating that the respondent has attended a finance-related class. Third, Fin Literacy is a dummy variable indicating that the respondent displays financial literacy. During the interview, he/she was asked: “If you put 100 Yuan into a 5-year time deposit account, with an interest rate of 4%, 5 years later the total money you would receive is less than, more than or equal to 120 yuan?” We identify the respondent to have financial literacy if he/she answers correctly (greater than 120 Yuan). Finally, Econ Literacy is a dummy variable indicating basic economic literacy with regard to inflation. During the interview, the respondent was asked: “You have 100 Yuan, the bank interest rate is 5%, and the inflation rate is 3%. If you put this money into a bank, the amount of goods you can purchase one year later is: less than, more than, or equal to the products you could have purchased a year ago?” We identify the respondent to have economic knowledge if he/she answers correctly (“more than” since the interest rate exceeds the inflation rate). Overall, we expect households with better financial/economic literacy to have greater capacity to understand and use financial instruments such as borrowing to maximize household earning and welfare, and in responding to local economic opportunities. Thus, capable households are more likely to avoid exorbitantly high interest loans, while also borrowing more at reasonable rates when economic opportunities are                                                                 10   See Fernandes et al. (2014) and Miller et al. (2015) for overviews.  10    presented. Our final variable belongs to a category of its own: the local supply of finance, which we measure as the number of bank branches within a community. 11 This variable is directly obtained from the CHFS data. Based on the classifications in the CHFS, a village (i.e., a rural community) has 2,367 residents and 579 households on average, while a neighborhood committee (i.e., an urban community) has 8,147 residents living in 2,607 households. The ability to measure access to a financial facility at such a micro level is a clear advantage of this data set. Presumably, having greater access to financial facilities at the local level increases the supply of financial services (see e.g., Burgess and Pande, 2005; Bruhn and Love, 2014). We thus expect bank branch density to be associated with more borrowing and lower financing costs. III. Access to Finance, Household Characteristics and the Rural-Urban Differences Table 2 presents the summary statistics for the pooled, the urban, and the rural samples. The sample splits indicate that it is important to examine access to finance separately for rural and urban residents because they face vastly different levels of access. Relative to urban residents, those in rural areas are poorer, have significantly less usage of financial services, and likely face fewer opportunities to establish businesses, though the CHFS does not have questions that enable us to demonstrate this last point. Perhaps due to the disadvantages faced by rural residents, a key concern for policy makers in China is their access to and costs of finance. About 44% of our sample residents come from the rural areas. While the level of access to loans is reasonably high in China, it is achieved mainly through the informal channel. On average, 48.3% of households have some type of loan. However, only                                                                 11   Again, a community means either a village in rural areas or a neighborhood committee in urban areas.  11    16.2% have bank loans, while 40.9% have borrowed from informal sources.12 These figures are higher than those found for China in the World Bank’s Global Index for Financial Inclusion (Findex), which shows that 36.3% of Chinese respondents had borrowed any money in 2011, while 9.6% had borrowed from a financial institution. 13 The disparities could be due to differences in question wording, though it seems likely to us that the CFHS could simply be a more complete tool for summarizing Chinese household financial usage than the Global Findex.14 The same pattern is observed for access to long-term finance. Interestingly, roughly 92% of all loans are long-term, under the definition that we use (maturities greater than one year). But maturities differ across formal and informal sources: for bank loans, the share of long-term finance is 78%; for informal loans, 94%. Thus, informal loans are more common among Chinese households, especially for long-term finance. Rural residents borrow more frequently and rely relatively more on informal sources. While 41.7% of urban residents borrow, 56.7% of rural residents do. Moreover, the difference in the tendency to borrow stems mainly from informal borrowing: the share of residents borrowing from bank sources is roughly the same in rural and urban areas (about 16%), but the share borrowing from informal sources is higher by 20 percentage points in rural areas (i.e., 52% versus 32%). Our interpretation is that greater needs for finance in rural areas, perhaps due to lower wealth, higher risks associated with agricultural production and greater demand for productive inputs, along with worse access to formal financial services and medical insurance, results in stronger reliance on informal finance in rural areas.                                                                 12   Some households have access to both formal and informal loans, which explains why the share for loans is  not equal to the summation of the shares of households with bank and informal loans.  13  Figures taken from http://datatopics.worldbank.org/financialinclusion/country/china accessed 8/18/2015.    14   Of course, the Global Findex was designed to facilitate comparisons across countries, and thus the survey  instrument is shorter and less likely to fully capture the specific financial context of each country. The Findex  indicates that China is slightly less financially inclusive than countries with similar income levels. In other upper  middle income countries, 37.7% of respondents had borrowed over the past year, 10.4% from a financial  institution.  12    The shares of interest-free loans differ widely across types of loans (see Table 3). Overall, 67% of loans are interest-free, and more so in rural than in urban regions (74% vs. 60%). The prevalence of interest-free loans is mainly explained by a large percentage of interest-free informal loans: 86-88% of informal loans carry no interest for both rural and urban households. The share of long-term informal loans that are interest-free is even higher, 93 and 94 percent in rural and urban areas, respectively. Clearly, the prevalence of interest-free informal loans is another indication of the under-development of the formal banking sector and its inability to reach large segments of the Chinese population. On average, informal borrowers receive lower rates than those who borrow from banks, but that is largely due to the high incidence of interest-free informal loans. The average annual interest rate on bank loans is 5.2 percent, in contrast to only 0.4 percent for informal loans. If we focus on non-interest-free loans, the average rate on bank loans is 5.8 percent, while informal loans carry an average rate of 7.9 percent. There are some notable rural-urban differences in interest rates. On average, urban residents pay higher rates (1.4 vs. 0.9 percent). But rural residents pay higher rates for bank loans than urban residents (5.6 vs 5.0 percent), and for non-interest-free informal loans (8.6 vs 7.1 percent). Rural and urban residents also differ significantly on many key characteristics, which could also influence their sources of finance and credit terms. For example, family size tends to be slightly larger in rural areas, and social networks (in terms of the number of siblings) are therefore also larger than in urban areas. Obviously, rural households are more likely to own farm land. In contrast, urban residents are more likely to join the Communist Party (17.9% vs. 9.5%)15 and are better educated (12% have a college degree vs. 0.2% in rural areas). Not surprisingly, fewer bank branches are in close proximity in rural areas than in urban areas—the                                                                 15   The share of residents belonging to the Communist Party member is slightly higher than the national  average.  13    average number of branches per community is 1.59 for urban areas compared to only 0.32 in rural areas. Signals are mixed about which group has better financial/economic literacy: judging by Fin Class and Fin Literacy, urban residents are more likely to pursue financial education and have greater financial literacy. However, Econ Literacy and an understanding of expected value (as reflected in the Risk for Return variable) are similar for rural and urban residents. Reasons for borrowing CHFS data allow us to understand how various loans are used. For each respondent’s loans, the following potential uses are offered as responses, and multiple choices are allowed: agricultural production, own business, tuition for education, housing purchases, car purchases, financial product purchases, and other purposes. The last two options are available only for informal loans. Loans for “other purposes” would include medical expenses, marriages, funerals, and expenses associated with other important life events. We also know the specific amount of borrowing associated with each of these needs. The results are summarized in Table 5. The amount of and reasons for bank loans differ greatly between urban and rural regions. While roughly 16% of residents borrow from banks in both rural and urban regions (Table 2), the average amount for urban residents (115,488 RMB, roughly 18,000 U.S. dollars) is almost 9 times as large as that for rural regions (roughly 2,000 dollars). In urban areas, borrowing is most frequently for housing purchases: 12.4% of urban households borrow for this purpose, 1.9% of urban households borrow for business uses, 1.7% for car purchases, and 1.2% for tuition expenses (Table 5, Panel A). However, borrowing frequency patterns differ greatly from those for amounts borrowed. While housing purchases is by far the most frequently cited reason for borrowing, those loans represent only 47% of the total amount of borrowing from banks (Table 5, Panel A). In contrast, though only 1.9% of households take bank loans for business 14    uses, those loans account for 50.6% of total bank lending to urban households. Car purchases and tuition expenses represent small shares of total bank lending to urban households. In rural areas, however, the story is very different. 7.1% of rural households borrow for agricultural production, 6.2% for housing purchases, 3.1% for tuition expenses, 1.3% for car purchases, and only 1.1% for use in their own businesses. In terms of the shares of total bank lending to rural residents, housing purchases were by far the most important reason (58.5%), followed by agricultural production (18.6%), for use in own business (13.5%), car purchases (6.1%), and tuition expenses (3.3%). The amounts of and reasons for informal loans, in contrast, are much more similar across urban and the rural areas (Table 5, Panel B). The average amount of an informal loan in both areas is about 25,000 RMB (or 4,000 dollars). The key reasons for informal borrowing among urban households are for housing purchases (22.3%), other purposes (4.8%), tuition expenses (4.2%), business uses (4.2%), and car purchases (2.3%). Informal loans for housing purchases represent 66.8% of the volume of total informal borrowing by urban residents, business uses comprise 17%, other purposes 9.1%, while the remaining reasons are less important. For rural residents, the key reasons for taking on informal loans are for housing purchases (33.1% of informal loans), agricultural production (14.9%), tuition expenses (10.2%), and other purposes (9.2%). Again, the distribution of loan volumes associated with those reasons is more skewed: housing purchases account for 66.3% of all informal rural borrowing; other purposes, 10.8%; agricultural production, 10.3%; and business uses, 7.1%. Thus, housing purchases are by far the main reason for obtaining both formal and informal loans, and informal loans are also important for emergency uses (encompassed under “other purposes”) in both rural and urban areas. For rural residents, tuition expenses are also an important reason for informal borrowing, and the associated amounts tend to be quite small (about 150 dollars per rural borrower). 15    The emphasis on credit to support housing purchases is not surprising. While housing finance arrangements for high-income groups in developing countries are similar to those in developed countries, informal institutions for the provision and financing of housing are most common for lower-income groups, especially in cities (Pal and Van Vliet, 2012). While the state officially held urban land in China, a market has been emerging, in which leases to those lands can be bought and sold. One review of the Chinese housing situation notes, “Ill-defined property rights, incomplete real estate markets and competing forms of private-public supply, together with a lack of long-term finance have been the defining obstacles for home ownership, especially for the lower stratum of the middle class.” (Bardhan and Edelstein, 2007). Reliance on informal mechanisms to fund housing purchases is, therefore, pervasive. The lower half of Table 5 deals with long-term bank loans (Panel C) and long-term informal loans (Panel D). The urban (rural) share of long-term bank loans in total bank loans is about half (three-quarters).16 In contrast, long-term loans represent close to 100 percent of total informal loans in both urban and the rural areas. Thus informal loans tend to be long-term under our definition. In addition, the overwhelming share of long-term bank loans is for housing purchases: 94% for urban residents, and 78% for rural residents. Similarly, housing purchases are also a key reason for obtaining informal long-term loans: around 70 % of informal long-term loans are obtained for that reason in both urban and rural areas. A reasonably large share of the long-term loans taken by urban residents is to cover tuition expenses (4.1%). To summarize, the level of access to loans in general is reasonably high in China, though this is achieved mainly through the informal channel, especially in rural areas. Most loans, especially informal loans, have maturities longer than a year. Informal loans appear to be a                                                                 16   For urban residents, total loans in Table 5 are RMB 115,488 of which RMB 57,179 are long‐term. For rural  residents, total loans are RMB 13,317 while long‐term loans are RMB 9,811.    16    cheaper source of financing, as 92% of them are interest-free, and their overall average interest rate is 0.4%. However, those interest rates do not reflect the costs of arranging informal loans and the future social costs and obligations that those loans often entail. Moreover, interest- bearing informal loans are more expensive than bank loans. Housing purchases are by far the most prevalent purpose for both formal and informal loans, and informal loans also appear to be important for emergency uses (subsumed within the category “other purposes”) in both rural and urban areas. For rural residents, tuition payments are also an important motivation for informal borrowing, though the amounts borrowed tend to be quite small. Empirical specification We rely on multivariate regressions to investigate key determinants of access to and costs of credit, hoping to distinguish the influence of potentially correlated factors. We focus on the following aspects: the incidence of, amounts of, and interest rates for loans, bank loans, informal loans, long-term loans, long-term bank loans, and long-term informal loans. Such a multi-faceted investigation should further our understanding of both the quantity and cost of access to finance. We estimate the following equations on various aspects of access to finance: (1) Here, the subscripts i and c refer to household i and community c. F is financial access, which could be any of the indicators of quantity or price of access to finance. CHAR represents household characteristics including the age of the household head, whether he/she is employed, the number of children (i.e., residents 16 or younger), and the number of adults of working age (i.e., between the age of 17 and 65). Proxies for WEALTH include whether the household has 17    farmable land (“Farmable Land”),17 whether the household’s dwelling was inherited from parents or purchased from the (current or former) employer at a subsidized price (“Inherited Housing”), 18 and whether the household head is a manager in his/her place of work (“Manager”). The education level of the household is proxied by whether the household head graduated from high school (“High School”), or from a college (“College”). PARTY indicates whether any household member is affiliated with the ruling party, the Communist Party. SIBLING is the number of siblings that the household head and, if applicable, his/her spouse have, which is a proxy for the size of the household’s informal social network. LITERACY includes the variables Risk for Return, Fin Class, Fin Literacy, and Econ Literacy as described above. Finally, BRANCH is the number of bank branches within a community. To avoid overstating the precision of our estimates, our standard errors are clustered at the community level to allow for within-community correlation of our error terms (Moulton, 1990). Incidence of Loans In Table 6 we report regression results for the correlates of the incidence of different types of loans. The outcomes include a dummy variable for having a loan (columns 1 to 3), a dummy for having a bank loan (columns 4 to 6), and a dummy for having an informal loan (columns 7 to 9). For each outcome, we report the results for the pooled, urban, and rural samples, respectively. We report linear-probability model results for ease in interpretation, but results are robust when using either probit or logit specifications. Since we are primarily concerned about estimating average effects in this paper, linear probability models are as consistent as probit or logit (Angrist, 2001).                                                                 17   This variable is available for both urban and rural households. Due to the blurred nature of the urban/rural  classification, sometimes urban residents also have farmable land.  18   Starting in the second half of the 1990s, the state sector in China’s urban areas began a series of housing  reforms in which employers sold (formerly free) apartments to employees at substantially subsidized prices  (Wang, 2011).  18    As expected, household characteristics explain significant variation in loan usage, but urban and rural residents differ in terms of which characteristics matter most. From the pooled regressions, it is clear that even after holding constant a rich array of household characteristics and the availability of local bank branches, a higher percentage of rural residents borrow, roughly 6 percentage points higher (relative to a mean loan usage rate of 48.3% for the full sample in Table 2). Moreover, there is no difference in access to bank finance between urban and rural residents, but usage of informal finance is 6.6 percentage points higher for rural residents (model 7). Households with more children are more likely to borrow, both in terms of bank loans and informal loans. This pattern holds in the pooled and the urban samples, but not in the rural sample.19 This is not surprising since many factors in urban areas increase the cost of raising children and therefore create greater needs for borrowing. For example, urban children are more expensive to raise due to higher costs of schooling, and it is probably less likely that urban grandparents act as free care providers. Households with more working-age adults are more likely to borrow, both formally from the bank and from informal sources, in both rural and urban regions. This could reflect a greater perceived ability to pay back in the future and thus greater lender willingness to extend credit. In urban areas, a working household head increases the likelihood of borrowing from banks, but not from informal sources, though this has no association with borrowing in rural areas. Finally, households with younger heads are more likely to borrow from both formal and informal sources, in both urban and rural regions. This likely reflects both their lower levels of income and greater demand for credit to purchase durables and housing, and for human capital investment. In general, proxies for household wealth tend to be positively associated with more and/or                                                                 19   The one‐child policy of China is not strictly enforced, especially in rural regions. In recent years, members of  couples who themselves were only children are often allowed to have a second child. Twins are, of course,  automatic exemptions from the one‐child policy.    19    better access to finance. For example, Farmable Land is positively associated with access to finance, but only in urban areas, and only for informal loans. This also makes sense since land in urban areas is likely to have greater market value, but such land does not have firm legal protection as yet in Chinese law, so only informal lenders are willing to consider this as an indicator of repayment capacity. In addition, households with subsidized and/or inherited housing (i.e., those that have less need for a mortgage) are less likely to borrow, and this is true for both urban and rural residents. For urban residents, reductions in the probability of having bank loans or informal loans are 5.8 and 3.8 percentage points, respectively (models 5 and 8, Table 6). For rural residents, inherited housing reduces the probability of informal loans by 5.2 percentage points, though there is no effect on bank borrowing. Urban households whose head is a manager are significantly more likely to borrow from a bank (by 6.6 percentage points). No significant association is found for urban residents for informal financing, or for rural residents in general. This last non-result is not surprising since rural residents tend to work for themselves, and thus being a manager is both less likely and less meaningful. On average, households whose heads have high school education do not borrow more than those whose heads have less education. However, urban households whose heads have high school education are 4.3 percentage points more likely to borrow from banks (model 5), and 4.5 percentage points less likely to borrow from informal sources (model 8). Thus, these two effects largely wash each other out. Urban households whose heads have a college education (or more) are 3.8 percentage points more likely to have any type of loan; they are 13.9 percentage points more likely to have a bank loan, and 4.4 percentage points less likely to have informal loans. In light of their higher interest rates and longer maturities, it may come as a surprise that formal rather than informal loans are preferred by better-educated residents. Our conjecture is 20    that informal financing almost always carries associated informal costs (i.e., costs not reflected in interest rates), such as informal gifts and implied obligations to assist lenders in the future, making them relatively less attractive to borrowers with formal options. Another reason for their greater reliance on formal loans is that better-educated people tend to have fewer siblings and a higher opportunity cost of their time, which would raise their relative costs of using informal financing. They also have higher-paying jobs in general, and likely steadier incomes, making them better candidates for bank financing. The overall effect is therefore that better- educated residents rely more on formal financing. Contrasts in the configurations and effectiveness of various social networks on access to finance shed light on how finance works for Chinese households. Rural households with an affiliation to the Communist Party are 10.2 percentage points more likely to have a bank loan (model 6). For urban households, incidence of affiliation with the Communist Party is relatively high (18%, in contrast to 9.5% in rural areas),20 and party affiliation is not associated with a greater likelihood of having a bank loan. While party membership is associated with greater use of bank loans in rural areas, it is not associated with more usage of informal loans. These pieces of evidence suggest that affiliation with political power can affect access to formal credit in some rural Chinese contexts, perhaps because state-owned banks have greater influence over credit markets in rural areas. The number of siblings, a proxy for the extent of private social networks, also influences informal financing. Each additional sibling is associated with 4.7 and 3.9 percentage point gains in the likelihood of informal borrowing in urban and rural areas, respectively (models 8 and 9). Thus, private social networks also play an important role in facilitating informal credit markets.                                                                 20   Figures based on CHFS data.    21    Financial and economic literacy tend to shift household loan portfolios from informal to formal sources. Household borrowing from banks increases by 3.5 and 14.9 percentage points in urban and rural areas when survey respondents report having taken a financial class (models 4 and 5). Financial literacy is associated with a 3.9 percentage point gain in the likelihood of having a bank loan in rural areas. However, economic literacy appears to discourage informal borrowing (by 4.7 percentage points) in rural areas. For urban residents, understanding risk- return tradeoffs (as measured by our Risk for Return variable) is associated with a greater likelihood of bank borrowing and less likelihood of informal borrowing. Table 7 re-runs the regressions using the dummy variables for long-term loan, long-term bank loan, and long-term informal loan as dependent variables. Overall, qualitative results are remarkably similarly to what we find about loans in general in Table 6. Although a few coefficients differ in terms of statistical significance, they almost invariably have the same signs as in the prior table. To summarize, there is no significant difference in use of bank finance for urban and rural residents, but usage of informal finance is 6.6 percentage points higher for rural residents. In general, proxies for wealth (e.g., having farmable land) tend to be positively associated with more and/or better access to finance. Households with better education and stronger financial and economic literacy tend to rely relatively more on bank financing and less on informal financing, especially those with college educated heads. Rural households with connections to the Communist Party are more likely to obtain bank loans than other rural households; however, that connection does not appear to help in obtaining informal finance. In contrast, the extent of private social networks (as reflected in the number of siblings) is positively associated with access to informal finance for both urban and rural households. Finally, the determinants of access to long-term finance tend to be quite similar to those for access to credit in general. Amount of Loans 22    Table 8 replicates the models in Table 6, but the incidence of loans is replaced with the amount of loans (in logarithm, plus one) as the dependent variable. The aim is to examine whether the results for our key explanatory variables remain robust when access to finance is measured by the amount rather than the incidence of various types of loans. In fact, the results are remarkably consistent with those in Table 6. The vast majority of significant results remain so, and when statistical significance differs the signs of the coefficients are almost always the same. There are, however, a few notable exceptions. First, affiliation with the Communist Party is significantly associated with larger loan amounts in rural areas (model 3); in contrast, this variable was insignificant when loan incidence was the dependent variable. Second, Econ Literacy is not significantly associated with total loan amounts in rural areas; it was negative and significant when loan incidence was the outcome of interest. Third, Risk for Return is not significantly associated with informal loan amounts in urban areas; it was negative and significant for incidence of informal loans. Table 9 replicates the results for various types of long-term loans. Again, the results are very similar, both to Table 7 and to Table 8. Thus, our conclusions on access to loans are not highly sensitive to whether it is measured by incidence or by amounts. Cost of Loans To round out our portrait of Chinese household finance, we examine how costs of financing are related to key household characteristics. One concern is that some groups might appear to have good access to finance but have to pay dearly for it. In short, we ask whether groups with better access to finance also enjoy better interest rates on their loans. Since we observe multiple loans for each loan type, we compute the average interest rate for each type of loan in our analysis (i.e., all loans, bank loans, informal loans, long-term loans, long-term bank loans, and long-term informal loans) weighting by the amount of credit for 23    each loan type.21 Results for various types of loans are shown in Table 10. Perhaps surprisingly, many characteristics that are associated with loan incidence and amounts are not significantly associated with interest rates. Holding basic household background, financial/economic literacy, and local banking facilities constant, rural residents do not appear to pay higher interest rates than urban residents. This is true for both bank loans and informal loans, and for long-term loans as well (See Appendix Table A.1). A key concern for policy makers in China is whether rural people face exorbitant interest rates, and thus these results suggest that there is not an inherent bias against rural borrowers, and that reducing those rates could be achieved through improvement on other factors that we control for in our models. Relatedly, the number of working household members does not appear to have strong effects on the rates at which a household borrows. Perhaps even more interestingly, affiliation with the Communist Party does not affect the interest rate that a household pays on its loans. This suggests that political benefits are mainly reflected in access to credit (as found earlier), not in its cost. Our proxy for the extent of private social networks, the number of siblings, is also not significantly associated with interest rates. However, households with more children tend to face higher bank loan rates, perhaps due to their greater needs to borrow for consumption and human capital investment that stretch their repayment capacities in the eyes of lenders. Recall that Tables 8 and 9 showed that the total amount of borrowing is increasing in the number of children, and the result is significant only for bank loans and in urban areas. Households with younger survey respondents (typically the household head) tend to borrow at higher interest rates in the pooled sample, and for informal finance in urban areas. This is consistent with our priors since young people likely face greater information asymmetry challenges in their dealings with lenders. They are more likely to lack an established network                                                                 21   Although the interest rate computed in this way is censored at zero, our key interest is the marginal effect  of a variable on interest rates, and thus ordinary least square estimation delivers unbiased and easier‐to‐ interpret estimates (Angrist, 2001).  24    of friends and relatives who have sufficient money to lend, and thus are perceived as more risky clients and must incur a risk premium. A stronger economic situation tends to have mixed effects on interest rates paid by households. The coefficient for Farmable Land is insignificant, while that for Inherited Housing is negative and significant for the pooled and the urban samples, though it is not significant for the rural subsample. Rural households whose head is a firm manager enjoy lower overall rates, perhaps indicating some privilege associated with better economic status. Somewhat surprisingly, households whose heads have high school education tend to pay higher interest rates than those that do not in both rural and urban areas. Similarly, urban households whose heads have college education tend to pay higher interest rates, though the opposite is true for rural households. The relatively higher rates paid by highly-educated households in urban areas should be interpreted in the context of the better investment opportunities that they face. Those opportunities mean highly-educated urban residents are likely to have greater demand for credit, which could contribute to higher rates. Financial/economic literacy tends to have little influence on bank interest rates, though Fin Class is associated with higher overall loan rates and overall rates for urban residents (models 1 and 2). Our interpretation again hinges on the assumption that urban residents have more economic opportunities, and that more financially literate households are keener to pursue those investment opportunities and thus have greater demand for loans. The number of bank branches in a community is also associated with higher interest rates in urban areas. While greater supply of finance should intuitively result in lower rates, our interpretation is that higher rates and more branch density are driven by the same underlying factor, ample growth opportunities. Thus high demand for credit pushes up interest rates in high-growth areas that also tend to have more dense networks of bank branches. We have also examined the determinants of interest rates for long-term finance (see Table A1 in the appendix). The 25    qualitative results are very similar. IV. Conclusions Using a representative data set summarizing household finance in China in 2012, we focus on how household structure, wealth, political and social networks, financial literacy levels, and the density of the local bank branch network are associated with access to, and the costs of, credit. We pay particular attention to usage of formal versus informal financial sources, and how access to sources of finance differs for rural and urban residents. We find that the access to loans is reasonably good in China, but this is achieved mainly through the informal channel, especially in rural areas. Surprisingly, many of these informal loans carry maturities longer than a year. However, interest-bearing informal loans are significantly more expensive than bank loans. The most important uses for formal and informal loans are for housing purchases, though informal loans are also important for emergency uses in both rural and urban areas. There is no difference in the rate of access to bank finance for urban and rural residents (both are low), but the usage of informal finance is 6.6 percentage points higher for rural residents. In general, proxies for wealth tend to be associated with access to more and/or better finance. Households with better education and stronger financial and economic literacy tend to rely more on bank financing relative to informal financing, especially those with college education. Rural households with connections to the Communist Party obtain more bank loans; such connections do not appear to help in obtaining informal finance, however. In contrast, the extent of private social networks (as reflected in the number of siblings) is positively associated with access to informal finance for both urban and rural households. Access to long-term finance tends to be determined by the same factors that drive access to finance in general. Moreover, our conclusions on access to finance are robust whether we measure it by the 26    incidence of access to loans or their amounts. Holding constant basic household characteristics, financial/economic literacy levels, and local bank branch presence, rural residents do not appear to pay higher interest rates than urban residents. Relatedly, household structure as reflected in the number of employed members is not significantly associated with the interest rates at which households borrow; nor is affiliation with the Communist Party. Households with younger survey respondents tend to borrow at higher interest rates. Rural households with members that are firm managers enjoy lower interest rates, signaling perhaps some privileges associated with higher economic status. Urban households whose heads have a college education tend to pay higher interest rates, while the opposite is true for rural households with college education. Taking financial classes is associated with higher bank rates for urban residents. Perhaps a surprising finding is that in urban communities with more dense networks of bank branches, households tend to pay higher rates. Clearly, however, the associations between financial education, branch density and higher interest rates should not be interpreted as causal. They do, however, suggest great demand for credit in high-growth urban areas that stretches the available supply of credit, even in those areas with a relatively high number of bank branches. The rate determination in urban areas thus differs from that in rural areas, hinting at the importance of greater economic opportunity. To summarize, our paper offers evidence that Chinese residents face dual credit markets, with the poor, young, those with poor financial knowledge, and those with larger family sizes relying much more on informal finance, while those that fare better on those dimensions are able to rely more on formal finance. Our research has several implications for research and policies. First, informal finance is overwhelmingly prevalent throughout China. Future research on household finance should pay 27    attention to the workings of informal finance, especially in developing countries, and how and when it gives way to formal lending that is better tailored to borrowers’ needs. Second, a key component of informal finance in the Chinese context is for housing, and more research could be devoted to the financing of housing, and reliance on informal sources, in other developing country contexts (though data availability is likely to be a binding constraint in the near term at least). Third, and as shown in other developing country contexts (see, e.g., Collins et al. 2009), Chinese household finance is comprised of an array of different types of arrangements. Some items involve small amounts but are high frequency (such as tuition); others are the reverse (such as funding for starting businesses). The challenge is for the supply side (and especially for formal providers) to meet the demands for affordable financial products with distinct, well-defined features tailored to households’ needs. Much more research could be done on the costs and benefits of informal finance, and what the government can do to reduce the costs of relying on informal finance for a large segment of the population. For instance, anecdotal evidence suggests that default is rampant among informal borrowers. What can be done to better safeguard the interests of informal lenders? From the borrower’s perspective, how much risk do SMEs face when they rely heavily on informal finance, especially in times of economic crisis? Finally, how can the government facilitate the transition from a dual credit market to a formal credit market that allows the less privileged to have good access to formal finance as well? 28    References Allen, Franklin, Jun Qian, and Meijun Qian, 2005. "Law, Finance, and Economic Growth in China." Journal of Financial Economics 77: 57–116. Angrist, Joshua. 2001. "Estimation of Limited Dependent Variable Models with Dummy Endogenous Regressors: Simple Strategies for Empirical Practice," Journal of Business and Economic Statistics 19(1): 2-16. Ayyagari, Demirguc-Kunt, Maksimovic. 2010. “Formal versus Informal Finance: Evidence from China.” Review of Financial Studies 23(8): 3050-3097. Bai, Chong-En, Jiangyong Lu, and Zhigang Tao. 2006a. “Property Rights Protection and Access to Bank Loans,” Economics of Transition 14(4): 611-628. Bai Chong-En, Jiangyong Lu, and Zhigang Tao. 2006b. “The Multitask Theory of State Enterprise Reform: Empirical Evidence from China. American Economic Review, 96(2): 353-357. Bardhan, Ashok and Robert H. Edelstein. 2008. “Housing Finance in Emerging Economies: Applying a Benchmark from Developed Countries.” In Mortgage Markets Worldwide Danny Ben-Shahr, Charles Ka Yui Leung, Seow Eng Ong, eds. Oxford: Blackwell Publishing Ltd. Boubakri, N., O. Guedhami, D. Mishra, W. Saffar. 2012. “Political Connections and the Cost of Equity Capital,” Journal of Corporate Finance 18(3), 541-549. Brandt, Loren and Hongbin Li, 2003. “Bank Discrimination in Transition Economies: Ideology, Information, or Incentives?” Journal of Comparative Economics 31:387–413. Bruhn, Miriam and Inessa Love, 2014. “The Real Impact of Improved Access to Finance: Evidence from Mexico.” Journal of Finance 69(3):1347-1376. Burgess, Robin, and Rohini Pande. 2005. “Do Rural Banks Matter? Evidence from the Indian Social Banking Experiment.” American Economic Review, 95(3): 780-795. Cai, Jing, Alain De Janvry, and Elisabeth Sadoulet. 2015. “Social Networks and the Decision to Insure.” American Economic Journal: Applied Economics 7(2): 81-108. Choi, Eun Kyong and Kate Xiao Zhou. 2001. “Entrepreneurs and Politics in the Chinese Transitional Economy: Political Connections and Rent-Seeking.” The China Review 1(1):111- 135. Chow, C. K. and Fung, M. K. Y. 1998. “Ownership Structure, Lending Bias, and Liquidity Constraints: Evidence from Shanghai’s Manufacturing Sector,” Journal of Comparative Economics, 26: 301-316. Claessens, Stijn, Erik Feijen, and Luc Laeven, 2008. “Political Connections and Preferential Access to Finance: The Role of Campaign Contributions.” Journal of Financial Economics 88(3): 554-580. Collins, Daryl, Jonathan Morduch, Stuart Rutherford and Orlanda Ruthven. 2009. Portfolios of the Poor: How the World’s Poor Live on $2 a Day. Princeton, NJ: Princeton University Press. Cull, Robert and Lixin Colin Xu, 2005. “Institutions, Ownership, and Finance: The Determinants of Investment among Chinese Firms.” Journal of Financial Economics 77(1): 117-146. Cull, Robert, Lixin Colin Xu, Tian Zhu. 2009. "Formal Finance and Trade Credit During China's Transition," Journal of Financial Intermediation 18(2): 173-192. Cull, Robert, Wei Li, Bo Sun, Lixin Colin Xu. 2015. “Government Connections and Financial Constraints: Evidence from a Large Sample of Chinese Firms,” Journal of Corporate Finance. 32: 271-294. 29    Dinc, Serdar I., 2005. “Politicians and Banks: Political Influences on Government-owned Banks in Emerging Countries.” Journal of Financial Economics 77(2): 453-479. Faccio, Mara. 2006. “Politically Connected Firms,” American Economic Review 96(1): 369-386. Faccio, Mara, John J. McConnell, and Ronald W. Masulis, 2006. “Political Connections and Corporate Bailouts,” Journal of Finance 61(6): 2597-2635. Fan, Joseph P.H., O.M. Rui, M.X. Zhao. 2008. “Public Governance and Corporate Governance: Evidence from Corruption Cases,” Journal of Comparative Economics 36(3): 343-364. Fernandes, Daniel, John G. Lynch, Jr., and Richard G. Netemeyer. 2014. “Financial Literacy, Financial Education, and Downstream Financial Behaviors.” Management Science 60(8): 1861-1883. Francis, B.B., I. Hasan, X. Sun. 2009. “Political Connections and the Process of Going Public: Evidence from China.” Journal of International Money and Finance 28(4): 696-719. Gan, Li, Lixin Colin Xu, and Yang Yao. 2012. "Risk Sharing and Local Elections: Evidence from Chinese Villages," Economics of Transition 20(3): 521-547. Gordon, R. H. and W. Li. 2003. “Government as a Discriminating Monopolist in the Financial Market: The Case of China,” Journal of Public Economics, 87:283-312 Gordon, R. H., and W. Li. 2011. “Provincial and Local Governments in China: Fiscal Institutions and Government Behavior,” in Capitalizing China, National Bureau of Economic Research and University of Chicago Press. Grief, Avner. 1997. “Contracting, Enforcement, and Efficiency: Economics Beyond the Law,” In Annual World Bank Conference on Development Economics, 1997, Michael Bruno and Boris Pleskovic, eds. Washington DC: World Bank. Guariglia, Alessandra, Xiaoxuan Liu, and Lina Song. 2011. “Internal Finance and Growth: Microeconometric Evidence on Chinese Firms.” Journal of Development Economics 96(1): 79-94. Héricourt, J. and S. Poncet. 2009. “FDI and Credit Constraints: Firm Level Evidence in China”. Economic Systems 33: 1-21. Huang, Yasheng. 2003. Selling China: Foreign Direct Investment during the Reform Era. New York: Cambridge University Press. Johnson, Simon and Todd Mitton, 2003. “Cronyism and Capital Controls: Evidence from Malaysia.” Journal of Financial Economics 67(2): 351-382. Li, H.B., L.S. Meng, Q. Wang, and L.A. Zhou. 2008. “Political Connections, Financing and Firm Performance: Evidence from Chinese Private Firms.” Journal of Development Economics 87(2): 283-299. Li, Hongbin, Scott Rozelle, and Linxia Zhang, 2004. “Micro-credit Programs and Off-farm Migration in China. Pacific Economic Review 9(3): 209-223. Lusardi, Annamaria and Olivia S. Mitchell. 2011. “Financial Literacy around the World: An Overview.” Journal of Pension Economics and Finance 10(4): 497-508. McMillan, John. 1997. “Markets in transition.” In Advances in Economics and Econometrics: Theory and Applications, David M. Kreps and Kenneth Wallis, eds. Cambridge: Cambridge University Press. McMillan, John and Christopher Woodruff. 1999. “Interfirm Relationships and Informal Credit in Vietnam.” Quarterly Journal of Economics 114(4): 1285-1320. Miller, Margaret, Julia Reichelstein, Christian Salas, and Bilal Zia. 2015. “Can You Help Someone 30    Become Financial Capable? A Meta-Analysis of the Literature.” World Bank Research Observer 30(2): 220-246. Moulton, Brent. 1990. “An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Units.” Review of Economics and Statistics 72 (2): 334-38. Pal, A. and W. Van Vliet. 2012. “Housing Institutions in Developing Countries.” In International Encyclopedia of Housing and Home, Susan J. Smith, ed. Amsterdam: Elsevier. pp. 509-517. Poncet, Sandra, Walter Steingress, and Hylke Vandenbussche, 2010. “Financial Constraints in China: Firm-level Evidence.” China Economic Review 21(3): 411-422. Sapienza, Paola, 2004. “The Effects of Government Ownership on Bank Lending.” Journal of Financial Economics 72(2): 357-384. Shleifer, Andrei, and Robert W. Vishny, 1994. “Politicians and Firms.” Quarterly Journal of Economics 109(4): 995-1025. Tsai, K. S. (2002). Back-Alley Banking: Private Entrepreneurs in China. Ithaca: Cornell University Press. Townsend, Robert. 1995. “Financial Systems in Northern Thai Villages,” Quarterly Journal of Economics 110(4): 1011-1046. Wang, Shing-Yi. 2011. “State Misallocation and Housing Prices: Theory and Evidence from China.” American Economic Review 101: 2081-2107. World Bank. 2015. Global Financial Development Report. Yuan, Yan and Lihe Xu. 2015. “Are Poor Able to Access the Informal Credit Market?” Evidence from Rural Households in China.” China Economic Review 33(C): 232-246. Zhang, Guibin. 2008. “The Choice of Formal or Informal Finance: Evidence from Chengdu, China.” China Economic Review 19(4): 659-678. 31    Table1: Definition of key variables variables Definition loan loan dummy(=1 if have loan) bank loan bank loan dummy(=1if have bank loan) Informal loan informal loan dummy(=1 if have nonbank loan) long term loan long term loan dummy(=1 if have long term loan) LT bank loan bank long term loan dummy(=1 if have bank long term loan) LT Informal loan informal long term loan dummy(=1 if have non-bank loan) log(1+children) log(1+# of member <=16) log(1+labor) log(1+# of member>16 and <=65) land dummy =1 if household have farmable land inherited housing =1 if housing is inherited from parents or working unit log(1+sibling) log(1+ # of siblings),restricted to respondent and spouse. log(1+bank branch) log(1+number of bank branch within the community of the respondent). head work dummy =1 if house head have work work leader =1 if house head is a leader in work unit log(age) log(age of house head) high school =1 if have high school(or similar) degree college =1 if have college(or above) degree party =1 if any of house member is communist party member Fin class =1 if respondent attended financial related class before =1 if respondent have financial literacy (during the interview, the respondent is asked to answer a question about interest rate: if you put 100 Yuan into a 5-year time deposit Fin literacy account, the interest rate is 4%, 5 years later, total money you will get is: less, more than or equal to 120 yuan? We identify the respondent have financial literacy if he/she answers correctly.) =1 if respondent have economy knowledge. (During the interview, the respondent is asked to answer a question about inflation: you have 100 Yuan, the bank interest rate is 5%, inflation rate is 3%, if you put these money into a bank, one year later, the amount Econ literacy of goods you can purchase is: less, more than or equal to the products you can purchase a year ago? We identify the respondent have economy knowledge if he/she answers correctly.) =1 if respondent prefer higher risk and higher return. (During the interview, the respondent is provided with a lottery, one is 100% chance of getting 4,000 yuan, another Risk for return one the 50% chance of getting 10,000 yuan, 50% chance of get nothing. We identify the respondent as prefer risk if he/she picked the second one.) 32    Table 2: Summary statistics for key variables Pool sample Urban area Rural area variables Obs Mean Std.Dev Min Max Obs Mean Std.Dev Min Max Obs Mean Std.Dev Min Max loan 28132 0.483 0.500 0.000 1.000 18705 0.417 0.493 0.000 1.000 9427 0.567 0.495 0.000 1.000 bank loan 28132 0.162 0.369 0.000 1.000 18705 0.164 0.371 0.000 1.000 9427 0.160 0.366 0.000 1.000 informal loan 28132 0.409 0.492 0.000 1.000 18705 0.323 0.468 0.000 1.000 9427 0.519 0.500 0.000 1.000 LT loan 28132 0.445 0.497 0.000 1.000 18705 0.388 0.487 0.000 1.000 9427 0.517 0.500 0.000 1.000 LT bank loan 28132 0.126 0.332 0.000 1.000 18705 0.143 0.350 0.000 1.000 9427 0.104 0.306 0.000 1.000 LT informal loan 28132 0.384 0.486 0.000 1.000 18705 0.304 0.460 0.000 1.000 9427 0.485 0.500 0.000 1.000 log(1+loan amount) 28132 3.928 5.414 0.000 14.193 18705 3.601 5.483 0.000 14.193 9427 4.342 5.295 0.000 14.193 log(1+bank loan amount) 28132 1.792 4.199 0.000 14.298 18705 1.942 4.504 0.000 14.298 9427 1.601 3.767 0.000 14.298 log(1+informal loan amount) 28132 2.689 4.713 0.000 13.304 18705 2.085 4.386 0.000 13.304 9427 3.456 4.995 0.000 13.304 log(1+LT loan amount) 28132 3.560 5.283 0.000 14.152 18705 3.336 5.365 0.000 14.152 9427 3.845 5.163 0.000 14.152 log(1+LT bank loan amount) 28132 1.413 3.839 0.000 14.286 18705 1.695 4.279 0.000 14.286 9427 1.056 3.160 0.000 14.286 log(1+LT informal loan amount) 28132 2.554 4.632 0.000 13.305 18705 1.987 4.307 0.000 13.305 9427 3.274 4.921 0.000 13.305 Interest rate(>=0) 9885 0.011 0.029 0.000 0.240 5623 0.014 0.029 0.000 0.240 4262 0.009 0.029 0.000 0.240 - bank loan(>=0) 2001 0.052 0.035 0.000 0.156 1439 0.050 0.029 0.000 0.156 562 0.056 0.043 0.000 0.156 - informal loan(>=0) 8792 0.004 0.025 0.000 0.360 4712 0.004 0.025 0.000 0.360 4080 0.004 0.026 0.000 0.360 Interest rate(>0) 2199 0.052 0.042 10-7 0.240 1561 0.050 0.036 10-5 0.240 638 0.056 0.050 10-7 0.240 - bank loan(>0) 1782 0.058 0.032 5*10-5 0.156 1328 0.053 0.029 10-4 0.156 454 0.066 0.039 5*10-5 0.156 - informal loan(>0) 474 0.079 0.079 2*10-4 0.360 273 0.071 0.077 2*10-4 0.360 201 0.086 0.080 7.5*10-4 0.360 LT loan interest rate 10134 0.010 0.028 0.000 0.240 5746 0.013 0.028 0.000 0.240 4388 0.008 0.028 0.000 0.240 LT bank loan interest rate 1801 0.053 0.034 0.000 0.156 1344 0.051 0.028 0.000 0.156 457 0.058 0.042 0.000 0.156 LT informal loan interest rate 9177 0.004 0.025 0.000 0.360 4916 0.004 0.024 0.000 0.360 4261 0.004 0.025 0.000 0.360 LT loan interest rate 2066 0.053 0.043 4*10-7 0.240 1498 0.050 0.035 6*10-5 0.240 568 0.057 0.052 4*10-7 0.240 LT bank loan interest rate 1646 0.058 0.031 4*10-4 0.156 1261 0.054 0.026 0.001 0.156 385 0.065 0.039 4*10-4 0.156 LT informal loan interest rate 468 0.079 0.079 2*10-4 0.360 271 0.071 0.077 2*10-4 0.360 197 0.087 0.081 0.001 0.360 rural 28132 0.441 0.496 0.000 1.000 - - - - - - - - - - 33    Table 3. The share of interest‐free loans for various types of loans      Loans  Bank loans  Informal loans  Pooled sample  67.1%  4.9%  86.8%  Urban    59.7%  3.7%  85.5%  rural  74.0%  6.6%  87.8%    Long‐term Loans  Long‐term Bank loans  Long‐term Informal loans  Pooled sample  64.3% 3.2% 91.1% Urban    81.9% 6.6% 94.1% rural  73.3% 4.4% 92.8%     34    Table 4: Summary statistics for key explanatory variables Pool sample Urban area Rural area variables Obs Mean Std.Dev Min Max Obs Mean Std.Dev Min Max Obs Mean Std.Dev Min Max log(1+children) 28132 0.336 0.435 0.000 2.303 18705 0.298 0.397 0.000 2.079 9427 0.384 0.474 0.000 2.303 log(1+labor) 28132 1.171 0.469 0.000 2.890 18705 1.118 0.452 0.000 2.303 9427 1.238 0.482 0.000 2.890 log(age) 28085 3.894 0.296 2.944 4.727 18668 3.857 0.322 2.944 4.727 9423 3.941 0.253 2.944 4.727 land dummy 28132 0.543 0.498 0.000 1.000 18705 0.282 0.450 0.000 1.000 9427 0.874 0.332 0.000 1.000 inherited housing 28132 0.134 0.341 0.000 1.000 18705 0.161 0.368 0.000 1.000 9427 0.100 0.300 0.000 1.000 head work 28132 0.840 0.366 0.000 1.000 18705 0.781 0.413 0.000 1.000 9427 0.915 0.279 0.000 1.000 work leader 28132 0.050 0.218 0.000 1.000 18705 0.078 0.268 0.000 1.000 9417 0.014 0.119 0.000 1.000 party 28132 0.142 0.349 0.000 1.000 18705 0.179 0.384 0.000 1.000 9427 0.095 0.294 0.000 1.000 high school 28118 0.249 0.432 0.000 1.000 18695 0.345 0.475 0.000 1.000 9423 0.126 0.332 0.000 1.000 college 28118 0.068 0.252 0.000 1.000 18695 0.120 0.325 0.000 1.000 9427 0.002 0.050 0.000 1.000 log(1+sibling) 28132 1.751 0.623 0.000 3.091 18705 1.638 0.663 0.000 3.091 9427 1.896 0.535 0.000 3.045 log(1+bank branch) 28132 0.657 0.712 0.000 3.611 18705 0.951 0.737 0.000 3.611 9427 0.283 0.461 0.000 2.996 Fin class 28132 0.072 0.258 0.000 1.000 18705 0.109 0.311 0.000 1.000 9427 0.025 0.155 0.000 1.000 Fin Literacy 28132 0.226 0.418 0.000 1.000 18705 0.258 0.437 0.000 1.000 9427 0.185 0.389 0.000 1.000 Econ Literacy 28132 0.160 0.366 0.000 1.000 18705 0.154 0.361 0.000 1.000 9427 0.168 0.374 0.000 1.000 Risk for Return 28132 0.268 0.443 0.000 1.000 18705 0.273 0.446 0.000 1.000 9427 0.262 0.440 0.000 1.000 35    Table 5: Percentages of households that have loans and long term loans (from banks & informal sources) Urban Rural Pooled Share with Amount in % of total Share with Amount in % of total Share with Amount in % of total PANEL A. Total Loans access RMB loans access RMB loans access RMB loans Loans 41.70% 141,063.3 56.70% 38,055.4 48.30% 95,670.7 Bank Loans 16.40% 115,488.0 81.90% 16.00% 13,316.7 35.00% 16.20% 70,464.0 73.70% Amount in % of bank Amount in % of bank Amount in % of bank Purpose of Bank Loans Share for: RMB loans for: Share for: RMB loans for: Share for: RMB loans for: --of which: For agriculture use 0.70% 875.6 0.80% 7.10% 2,476.4 18.60% 3.50% 1,581.0 2.20% For business use 1.90% 58,400.7 50.60% 1.10% 1,798.5 13.50% 1.50% 33,457.8 47.50% For educational tuitions 1.20% 240.1 0.20% 3.10% 444.3 3.30% 2.10% 330.1 0.50% For housing purchases 12.40% 54,138.5 46.90% 6.20% 7,787.7 58.50% 9.70% 33,713.1 47.80% For car purchases 1.70% 1,832.8 1.60% 1.30% 809.6 6.10% 1.50% 1,381.9 2.00% total 100% 100% 100% Share with Amount in % of total Share with Amount in % of total Share with Amount in % of total PANEL B. access RMB loans access RMB loans access RMB loans Informal loan 32.30% 25,575.3 18.10% 51.90% 24,738.7 65.00% 40.90% 25,206.6 26.30% Amount in % of inf Amount in % of inf Amount in % of inf Purpose of Informal Loans Share for: RMB loans for: Share for: RMB loans for: Share for: RMB loans for: --of which: For agri. use 1.80% 608.1 2.40% 14.90% 2,544.9 10.30% 7.60% 1,461.6 5.80% For business use 4.20% 4,356.3 17.00% 2.30% 1,750.5 7.10% 3.40% 3,208.0 12.70% For financial product purchases 0.20% 260.7 1.00% 0.00% 1.6 0.00% 0.10% 146.5 0.60% For educational tuitions 4.20% 441.6 1.70% 10.20% 933.6 3.80% 6.80% 658.4 2.60% For housing purchases 22.30% 17,075.7 66.80% 33.10% 16,389.7 66.30% 27.00% 16,773.4 66.50% For car purchases 2.30% 517.5 2.00% 2.40% 452.5 1.80% 2.40% 488.8 1.90% For other purposes 4.80% 2,315.0 9.10% 9.20% 2,665.9 10.80% 6.70% 2,469.6 9.80% total 100% 100% 100% PANEL C. Urban Rural Pooled Share with Amount in % of total Share with Amount in % of total Share with Amount in % of total Long-term loans: access RMB loans access RMB loans access RMB loans LT loan 38.80% 81,382.6 51.70% 32,909.0 44.50% 60,021.7 LT bank loan 14.30% 57,178.5 71.50% 10.50% 9,811.4 31.20% 12.60% 36,305.2 61.90% % of LT % of LT % of LT Amount in Amount in Amount in Purpose of LT Bank Loans Share for: RMB bank Share for: RMB bank Share for: RMB bank loans for: loans for: loans for: --of which: For agriculture use 0.20% 267.0 0.50% 1.60% 647.3 6.60% 0.80% 434.6 1.20% For business use 0.70% 2,349.0 4.10% 0.30% 821.7 8.40% 0.50% 1,676.0 4.60% For educational tuitions 1.20% 236.8 0.40% 3.00% 437.4 4.50% 2.00% 325.2 0.90% For housing purchases 12.30% 53,745.1 94.00% 6.00% 7,661.5 78.10% 9.50% 33,437.4 92.10% For car purchases 0.30% 580.3 1.00% 0.30% 243.3 2.50% 0.30% 431.8 1.20% Share with Amount in % of total Share with Amount in % of total Share with Amount in % of total PANEL D. access RMB loans access RMB loans access RMB loans LT informal loan 30.40% 24,204.1 28.50% 48.50% 23,097.5 68.80% 38.40% 23,716.5 38.10% % of LT % of LT % of LT Amount in Amount in Amount in Purpose of LT informal Loans Share for: RMB inf loans Share for: RMB inf loans Share for: RMB inf loans for: for: for: --of which: For agriculture 1.70% 600.3 2.50% 14.50% 2,407.5 10.40% 7.40% 1,396.7 5.90% use For business use 4.00% 3,972.2 16.40% 2.20% 1,557.7 6.70% 3.20% 2,908.2 12.30% For financial product purchases 0.10% 246.0 1.00% 0.00% 1.6 0.00% 0.10% 138.3 0.60% For educational tuitions 41.40% 441.6 1.80% 10.00% 930.0 4.00% 6.70% 656.8 2.80% For housing purchases 22.00% 17,075.7 70.50% 32.80% 16,314.5 70.60% 26.8% 16,740.3 70.60% For car purchases 2.30% 487.5 2.00% 2.30% 426.7 1.80% 2.30% 460.7 1.90% For other purposes 2.40% 1380.7 5.70% 4.00% 1459.6 6.3% 3.10% 1415.4 6.0% 36    Table 6: Determinants of access to loans y=loan dummy y=bank loan dummy y=informal loan dummy Explanatory variables urban rural pool sample urban area rural area pool sample pool sample urban area rural area area area rural 0.058*** 0.010 0.066*** (0.013) (0.011) (0.012) log(1+children) 0.041*** 0.063*** 0.018 0.028*** 0.052*** 0.012 0.027*** 0.033** 0.014 (0.010) (0.015) (0.015) (0.009) (0.011) (0.015) (0.010) (0.014) (0.015) log(1+labor) 0.180*** 0.153*** 0.205*** 0.070*** 0.043*** 0.106*** 0.159*** 0.138*** 0.171*** (0.010) (0.012) (0.016) (0.008) (0.009) (0.013) (0.010) (0.012) (0.016) land dummy 0.027** 0.030** 0.002 -0.005 -0.016 -0.008 0.044*** 0.057*** 0.011 (0.011) (0.013) (0.023) (0.009) (0.010) (0.015) (0.011) (0.013) (0.023) inherited housing -0.070*** -0.071*** -0.049** -0.048*** -0.058*** -0.009 -0.046*** -0.038** -0.052** (0.013) (0.016) (0.022) (0.008) (0.009) (0.014) (0.013) (0.015) (0.025) log(1+sibling) 0.037*** 0.037*** 0.036*** 0.007 0.006 0.015 0.047*** 0.047*** 0.039*** (0.008) (0.010) (0.012) (0.005) (0.006) (0.010) (0.008) (0.009) (0.014) log(1+bank branch) -0.012 0.001 -0.042* 0.009 0.016** -0.009 -0.019** -0.010 -0.048** (0.009) (0.010) (0.023) (0.008) (0.008) (0.019) (0.008) (0.009) (0.023) head work dummy 0.013 0.026* -0.000 0.039*** 0.048*** 0.006 -0.020 -0.012 -0.008 (0.013) (0.015) (0.025) (0.008) (0.011) (0.012) (0.013) (0.015) (0.026) work leader 0.067*** 0.068*** 0.076* 0.053*** 0.066*** -0.025 0.018 0.014 0.062 (0.019) (0.021) (0.046) (0.017) (0.018) (0.038) (0.017) (0.018) (0.049) log(age) -0.147*** -0.129*** -0.158*** -0.080*** -0.097*** -0.016 -0.108*** -0.077*** -0.158*** (0.018) (0.022) (0.030) (0.014) (0.017) (0.022) (0.017) (0.019) (0.031) high school -0.017 -0.009 -0.036 0.033*** 0.043*** -0.011 -0.044*** -0.045*** -0.034 (0.010) (0.011) (0.022) (0.009) (0.009) (0.019) (0.011) (0.011) (0.023) college 0.026 0.038** -0.125 0.133*** 0.139*** -0.089 -0.056*** -0.044** -0.108 (0.018) (0.019) (0.110) (0.016) (0.015) (0.076) (0.017) (0.018) (0.114) party -0.003 -0.022 0.34 0.034*** 0.005 0.102*** -0.006 -0.022 0.014 (0.013) (0.016) (0.021) (0.011) (0.010) (0.025) (0.013) (0.015) (0.022) Fin class 0.010 0.003 0.061* 0.052*** 0.035** 0.149*** -0.029* -0.025 -0.025 (0.017) (0.019) (0.035) (0.014) (0.013) (0.041) (0.017) (0.018) (0.042) Fin literacy 0.005 0.002 0.013 0.016** 0.002 0.039*** -0.004 -0.006 0.002 (0.010) (0.011) (0.021) (0.008) (0.008) (0.014) (0.009) (0.011) (0.018) Econ literacy -0.021 -0.007 -0.045* -0.003 0.008 -0.024 -0.026* -0.013 -0.047* (0.013) (0.013) (0.024) (0.010) (0.009) (0.018) (0.013) (0.012) (0.025) Risk for return -0.012 -0.005 -0.022 0.009 0.016** -0.000 -0.022** -0.018* -0.027 (0.010) (0.010) (0.018) (0.007) (0.008) (0.013) (0.010) (0.010) (0.018) constant 0.681*** 0.581*** 0.985*** 0.252*** 0.340*** -0.056 0.526*** 0.378*** 1.004*** (0.080) (0.096) (0.136) (0.060) (0.072) (0.103) (0.081) (0.089) (0.137) Observation 28084 18668 9416 28084 18668 9416 28084 18668 9416 R square 0.135 0.119 0.140 0.115 0.124 0.154 0.140 0.118 0.115 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 37    Table 7: Determinants of access to long term loans y=long term loan dummy y= long term bank loan dummy Y=long term informal loan dummy Explanatory variables pool urban pool sample urban area rural area rural area pool sample urban area rural area sample area rural 0.046*** -0.007 0.058*** (0.013) (0.009) (0.012) log(1+children) 0.032*** 0.059*** 0.004 0.008 0.046*** -0.019* 0.023** 0.029** 0.009 (0.011) (0.015) (0.017) (0.007) (0.010) (0.012) (0.010) (0.013) (0.015) log(1+labor) 0.173*** 0.153*** 0.187*** 0.056*** 0.042*** 0.081*** 0.156*** 0.138*** 0.162*** (0.011) (0.012) (0.017) (0.007) (0.008) (0.011) (0.010) (0.011) (0.016) land dummy 0.025** 0.023* 0.015 -0.015* -0.027** -0.020 0.045*** 0.054*** 0.022 (0.012) (0.012) (0.027) (0.008) (0.010) (0.012) (0.012) (0.012) (0.027) inherited housing -0.070*** -0.070*** -0.053** -0.050*** -0.057*** -0.017 -0.047*** -0.038*** -0.057** (0.013) (0.016) (0.022) (0.007) (0.008) (0.012) (0.013) (0.015) (0.024) log(1+sibling) 0.037*** 0.034*** 0.039*** 0.003 0.004 0.009 0.046*** 0.043*** 0.042*** (0.008) (0.010) (0.012) (0.005) (0.006) (0.008) (0.008) (0.009) (0.014) log(1+bank branch) -0.013 -0.001 -0.048** 0.011 0.014* 0.004 -0.021*** -0.010 -0.060*** (0.009) (0.010) (0.024) (0.007) (0.007) (0.013) (0.008) (0.009) (0.023) head work dummy 0.014 0.023 0.000 0.033*** 0.036*** -0.003 -0.016 -0.009 -0.003 (0.013) (0.015) (0.026) (0.007) (0.010) (0.011) (0.013) (0.015) (0.027) work leader 0.072*** 0.074*** 0.067 0.059*** 0.067*** -0.015 0.017 0.011 0.066 (0.019) (0.020) (0.047) (0.016) (0.017) (0.032) (0.017) (0.018) (0.048) log(age) -0.136*** -0.106*** -0.170*** -0.060*** -0.081*** -0.002 -0.100*** -0.061*** -0.162*** (0.019) (0.022) (0.034) (0.013) (0.017) (0.019) (0.017) (0.019) (0.031) high school -0.014 -0.004 -0.039* 0.038*** 0.043*** 0.008 -0.040*** -0.038*** -0.034 (0.011) (0.012) (0.022) (0.008) (0.008) (0.015) (0.011) (0.011) (0.023) college 0.026 0.042** -0.107 0.142*** 0.141*** -0.069 -0.053*** -0.039** -0.101 (0.019) (0.019) (0.107) (0.015) (0.015) (0.060) (0.017) (0.018) (0.114) party 0.002 -0.018 0.041* 0.022** 0.005 0.066*** -0.000 -0.018 0.023 (0.013) (0.016) (0.021) (0.010) (0.010) (0.023) (0.013) (0.015) (0.022) Fin class 0.014 0.002 0.088** 0.038*** 0.028** 0.098*** -0.022 -0.023 0.010 (0.017) (0.019) (0.035) (0.012) (0.012) (0.028) (0.016) (0.018) (0.040) Fin literacy 0.004 0.001 0.011 0.018** 0.010 0.029* -0.007 -0.010 0.004 (0.011) (0.012) (0.024) (0.008) (0.007) (0.016) (0.010) (0.011) (0.021) Econ literacy -0.023** -0.003 -0.053** -0.005 0.009 -0.026 -0.028** -0.010 -0.054** (0.012) (0.013) (0.021) (0.010) (0.009) (0.020) (0.012) (0.012) (0.022) Risk for return -0.016* -0.008 -0.026 0.012** 0.014* 0.010 -0.022** -0.018* -0.028* (0.009) (0.010) (0.017) (0.007) (0.007) (0.012) (0.009) (0.010) (0.017) constant 0.629*** 0.480*** 1.021*** 0.186*** 0.278*** -0.084 0.485*** 0.309*** 1.006*** (0.085) (0.097) (0.150) (0.056) (0.071) (0.093) (0.080) (0.088) (0.136) Observation 28084 18668 9416 28084 18668 9416 28084 18668 9416 R square 0.122 0.111 0.130 0.095 0.119 0.098 0.135 0.115 0.118 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 38  Table 8: Determinants of the amount of loans Explanatory y=log(1+loan amount) y=log(1+bank loan amount) y=log(1+informal loan amount) variables pool sample urban area rural area pool sample urban area rural area pool sample urban area rural area rural 0.276* -0.042 0.436*** (0.141) (0.117) (0.119) log(1+children) 0.409*** 0.643*** 0.151 0.351*** 0.645*** 0.166 0.204* 0.130 0.149 (0.119) (0.185) (0.156) (0.105) (0.132) (0.163) (0.105) (0.150) (0.150) log(1+labor) 1.665*** 1.339*** 1.947*** 0.743*** 0.450*** 1.155*** 1.234*** 1.034*** 1.287*** (0.125) (0.134) (0.204) (0.092) (0.108) (0.130) (0.108) (0.124) (0.187) land dummy 0.037 0.050 -0.228 -0.154 -0.286** -0.170 0.225** 0.384*** -0.148 (0.117) (0.138) (0.220) (0.100) (0.126) (0.154) (0.101) (0.104) (0.226) inherited housing -0.886*** -1.048*** -0.359* -0.608*** -0.736*** -0.112 -0.468*** -0.518*** -0.319 (0.122) (0.144) (0.209) (0.091) (0.104) (0.152) (0.110) (0.121) (0.214) log(1+sibling) 0.411*** 0.385*** 0.457*** 0.064 0.064 0.161 0.479*** 0.447*** 0.455*** (0.081) (0.099) (0.147) (0.059) (0.071) (0.103) (0.067) (0.076) (0.132) log(1+bank branch) -0.047 0.017 -0.315 0.109 0.174* -0.043 -0.198** -0.154** -0.469** (0.100) (0.108) (0.251) (0.092) (0.098) (0.198) (0.082) (0.075) (0.211) head work dummy 0.179 0.463*** -0.359 0.471*** 0.597*** 0.001 -0.292** -0.080 -0.406 (0.127) (0.150) (0.271) (0.090) (0.124) (0.122) (0.115) (0.127) (0.275) work leader 0.732*** 0.754*** 0.750 0.643*** 0.761*** -0.201 0.109 0.042 0.784 (0.234) (0.253) (0.556) (0.205) (0.225) (0.406) (0.179) (0.181) (0.558) log(age) -1.629*** -1.324*** -1.757*** -1.036*** -1.264*** -0.242 -0.826*** -0.372* -1.547*** (0.206) (0.265) (0.321) (0.164) (0.211) (0.241) (0.176) (0.190) (0.334) high school 0.126 0.107 0.078 0.422*** 0.527*** -0.078 -0.289** -0.405*** 0.086 (0.114) (0.127) (0.233) (0.095) (0.109) (0.186) (0.113) (0.111) (0.264) college 0.968*** 1.063*** -0.056 1.689*** 1.733*** -0.951 -0.504*** -0.452*** 0.701 (0.210) (0.221) (1.226) (0.194) (0.197) (0.870) (0.148) (0.149) (1.200) party 0.150 -0.176 0.780*** 0.379*** 0.062 1.135*** -0.122 -0.209 -0.081 (0.144) (0.180) (0.227) (0.124) (0.126) (0.255) (0.116) (0.133) (0.225) Fin class 0.433** 0.255 1.654*** 0.613*** 0.447*** 1.532*** -0.051 -0.119 0.595 (0.195) (0.207) (0.405) (0.158) (0.157) (0.384) (0.143) (0.146) (0.397) Fin literacy 0.198* 0.131 0.309 0.204** 0.043 0.462*** 0.062 0.033 0.122 (0.109) (0.117) (0.226) (0.087) (0.095) (0.149) (0.095) (0.100) (0.197) Econ literacy -0.028 0.064 -0.193 -0.002 0.135 -0.231 0.014 0.004 0.000 (0.134) (0.150) (0.231) (0.106) (0.112) (0.188) (0.132) (0.120) (0.257) Risk for return 0.072 0.155 -0.038 0.125 0.197** 0.035 -0.081 -0.064 -0.119 (0.105) (0.115) (0.193) (0.077) (0.092) (0.132) (0.102) (0.092) (0.207) constant 6.430*** 5.228*** 8.684*** 3.431*** 4.532*** -0.313 3.540*** 1.715** 8.469*** (0.895) (1.126) (1.530) (0.714) (0.890) (1.142) (0.805) (0.870) (1.558) Observation 28084 18668 9416 28084 18668 9416 28084 18668 9416 R square 0.113 0.112 0.141 0.116 0.126 0.155 0.105 0.094 0.105 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 39  Table 9: Determinants of the amount of long term loans Explanatory y=log(1+long term loan amount) y=log(1+long term bank loan amount) y=log(1+long term informal loan amount) variables pool sample urban area rural area pool sample urban area rural area pool sample urban area rural area rural 0.122 -0.205** 0.372*** (0.136) (0.100) (0.118) log(1+children) 0.330*** 0.626*** 0.028 0.153* 0.593*** -0.150 0.222** 0.154 0.173 (0.119) (0.172) (0.169) (0.085) (0.119) (0.125) (0.107) (0.150) (0.156) log(1+labor) 1.471*** 1.276*** 1.625*** 0.570*** 0.422*** 0.869*** 1.183*** 1.010*** 1.196*** (0.125) (0.134) (0.207) (0.082) (0.101) (0.118) (0.107) (0.118) (0.185) land dummy 0.038 0.007 -0.102 -0.228** -0.366*** -0.236* 0.276*** 0.401*** -0.009 (0.122) (0.131) (0.258) (0.093) (0.120) (0.133) (0.106) (0.103) (0.248) inherited housing -0.825*** -0.996*** -0.308 -0.611*** -0.714*** -0.170 -0.445*** -0.496*** -0.325 (0.126) (0.146) (0.226) (0.081) (0.101) (0.128) (0.108) (0.118) (0.213) log(1+sibling) 0.403*** 0.371*** 0.444*** 0.034 0.046 0.111 0.460*** 0.416*** 0.444*** (0.077) (0.098) (0.132) (0.052) (0.069) (0.083) (0.067) (0.076) (0.131) log(1+bank branch) -0.057 -0.005 -0.271 0.129 0.168* 0.070 -0.180** -0.150** -0.452** (0.099) (0.105) (0.235) (0.082) (0.091) (0.145) (0.077) (0.073) (0.202) head work dummy 0.190 0.410*** -0.317 0.417*** 0.469*** -0.068 -0.248** -0.061 -0.313 (0.130) (0.145) (0.282) (0.084) (0.116) (0.106) (0.116) (0.123) (0.285) work leader 0.836*** 0.844*** 0.791 0.717*** 0.796*** -0.110 0.144 0.080 0.796 (0.229) (0.247) (0.567) (0.198) (0.221) (0.345) (0.177) (0.179) (0.551) log(age) -1.447*** -1.140*** -1.695*** -0.784*** -1.075*** -0.025 -0.754*** -0.286 -1.527*** (0.219) (0.273) (0.357) (0.154) (0.208) (0.215) (0.177) (0.186) (0.334) high school 0.171 0.130 0.177 0.473*** 0.521*** 0.117 -0.272** -0.378*** 0.072 (0.120) (0.129) (0.240) (0.088) (0.100) (0.154) (0.113) (0.110) (0.264) college 1.031*** 1.072*** 0.586 1.768*** 1.721*** -0.684 -0.461*** -0.408*** 0.932 (0.209) (0.219) (1.214) (0.188) (0.192) (0.703) (0.147) (0.145) (1.226) party 0.029 -0.180 0.443** 0.231** 0.073 0.690*** -0.111 -0.200 -0.061 (0.138) (0.175) (0.215) (0.110) (0.122) (0.220) (0.112) (0.130) (0.216) Fin class 0.378* 0.198 1.541*** 0.423*** 0.323** 0.986*** -0.022 -0.104 0.683* (0.195) (0.203) (0.427) (0.139) (0.153) (0.286) (0.139) (0.142) (0.391) Fin literacy 0.189 0.187 0.184 0.208** 0.121 0.328** 0.061 0.044 0.108 (0.117) (0.118) (0.254) (0.085) (0.090) (0.166) (0.098) (0.099) (0.208) Econ literacy -0.021 0.090 -0.184 -0.036 0.141 -0.272 -0.021 0.011 -0.087 (0.125) (0.146) (0.220) (0.109) (0.107) (0.207) (0.122) (0.119) (0.236) Risk for return 0.013 0.056 -0.044 0.143* 0.171* 0.105 -0.102 -0.098 -0.122 (0.107) (0.115) (0.201) (0.074) (0.090) (0.122) (0.096) (0.090) (0.195) constant 5.758*** 4.475*** 8.295*** 2.603*** 3.805*** -0.904 3.198*** 1.368 8.031*** (0.954) (1.164) (1.603) (0.677) (0.875) (1.034) (0.801) (0.852) (1.508) Observation 28084 18668 9416 28084 18668 9416 28084 18668 9416 R square 0.099 0.106 0.117 0.101 0.120 0.101 0.103 0.094 0.103 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 40    Table 10. Determinants of the interest rates on loans (1) (2) (3) (4) (5) (6) (7) (8) (9) Interest rate of loan(*100) Interest rate of bank loan(*100) Interest rate of informal loan(*100) Pool Urban Rural Pool Urban Rural Pool Urban Rural rural -0.118 0.381 -0.067 (0.119) (0.469) (0.132) log(1+children) 0.065 0.299** -0.075 0.878*** 0.814*** 1.354** -0.058 0.078 -0.133 (0.106) (0.152) (0.143) (0.273) (0.265) (0.601) (0.121) (0.217) (0.144) log(1+labor) 0.234* 0.157 0.358* -0.307 -0.431 -0.228 0.092 0.212 0.044 (0.121) (0.152) (0.197) (0.484) (0.462) (1.087) (0.119) (0.137) (0.206) land dummy 0.074 0.049 -0.105 0.193 0.431 -1.256 0.279* 0.371 0.071 (0.126) (0.169) (0.200) (0.300) (0.340) (0.825) (0.151) (0.232) (0.184) inherited -0.287*** -0.287** -0.234 -0.184 -0.297 -0.046 0.032 0.136 -0.055 housing (0.108) (0.136) (0.177) (0.331) (0.337) (0.743) (0.088) (0.124) (0.136) log(1+sibling) -0.039 -0.145 0.087 0.011 -0.333 0.413 0.056 -0.048 0.153 (0.088) (0.100) (0.138) (0.200) (0.208) (0.527) (0.092) (0.067) (0.157) log(1+bank 0.222*** 0.189** 0.102 0.070 0.154 1.002* 0.172 0.096 -0.043 branch) (0.079) (0.092) (0.103) (0.230) (0.193) (0.605) (0.121) (0.091) (0.098) head work 0.112 0.279** -0.275 -0.506 0.315 -3.474 0.012 -0.025 0.001 (0.121) (0.138) (0.259) (0.736) (0.804) (2.306) (0.119) (0.139) (0.249) work leader -0.081 0.029 -0.601** -0.051 -0.096 0.141 -0.080 0.003 -0.404 (0.148) (0.167) (0.295) (0.228) (0.226) (1.250) (0.113) (0.125) (0.298) log(age) -0.764*** -0.766*** -0.644** -0.299 -0.056 0.031 -0.395** -0.268** -0.515 (0.183) (0.207) (0.322) (0.483) (0.496) (1.263) (0.179) (0.128) (0.319) high school 0.525*** 0.423*** 0.533** -0.199 -0.397 0.151 0.239 -0.003 0.540 (0.144) (0.163) (0.269) (0.223) (0.250) (0.516) (0.216) (0.227) (0.419) college 1.211*** 1.043*** -0.845** -0.420 -0.537 0.064 -0.115 -0.028 (0.197) (0.209) (0.356) (0.392) (0.361) (0.210) (0.223) (0.323) party 0.060 0.032 0.159 0.029 -0.042 0.090 -0.164 -0.040 -0.276 (0.128) (0.129) (0.243) (0.257) (0.211) (0.770) (0.106) (0.107) (0.210) Fin class 0.373** 0.345** 0.549 0.217 0.260 0.044 -0.057 -0.005 0.075 (0.161) (0.151) (0.460) (0.233) (0.196) (1.146) (0.139) (0.139) (0.315) Fin literacy 0.199 0.196 0.194 0.003 0.078 0.120 0.221 0.248 0.229 (0.151) (0.178) (0.257) (0.242) (0.221) (0.579) (0.198) (0.272) (0.296) Econ literacy -0.022 0.110 -0.131 0.048 0.055 0.480 -0.158 -0.125 -0.160 (0.113) (0.157) (0.158) (0.234) (0.190) (0.659) (0.127) (0.164) (0.150) Risk for return 0.054 0.205 -0.085 -0.029 -0.129 0.406 0.000 0.297 -0.206 0.098 (0.148) (0.134) (0.211) (0.178) (0.532) (0.128) (0.236) (0.141) constant 3.092*** 3.310*** 2.352* 6.586*** 5.618*** 7.647 1.103 0.643 1.673 (0.766) (0.874) (1.297) (2.032) (2.082) (5.476) (0.724) (0.518) (1.280) Observation 9863 5609 4254 1998 1436 562 8771 4699 4072 R square 0.111 0.128 0.122 0.175 0.209 0.333 0.051 0.073 0.073 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 41    Table A1. Determinant of interest rates for long term loans (1) (2) (3) (4) (5) (6) (7) (8) (9) Interest rate of LT loan Interest rate of LT bank loan Interest rate of LT informal loan (*100) (*100) (*100) Pool Urban Rural Pool Urban Rural Pool Urban Rural rural -0.117 0.464 -0.084 (0.115) (0.478) (0.129) log(1+children) 0.035 0.234 -0.076 0.783*** 0.589** 1.480** -0.046 0.067 -0.107 (0.107) (0.152) (0.146) (0.277) (0.271) (0.621) (0.120) (0.206) (0.147) log(1+labor) 0.214* 0.001 0.425** -0.208 -0.636 0.319 0.135 0.135 0.143 (0.120) (0.129) (0.203) (0.494) (0.466) (1.072) (0.116) (0.102) (0.206) land dummy 0.020 0.060 -0.240 0.199 0.423 -1.066 0.263* 0.415* 0.005 (0.129) (0.180) (0.208) (0.322) (0.367) (0.844) (0.148) (0.245) (0.171) inherited -0.294*** -0.285** -0.246* -0.136 -0.510 0.570 -0.018 0.132 -0.154 housing (0.094) (0.131) (0.137) (0.306) (0.321) (0.751) (0.073) (0.123) (0.095) log(1+sibling) -0.054 -0.123 0.047 -0.090 -0.361 -0.019 0.032 -0.038 0.093 (0.083) (0.097) (0.130) (0.226) (0.229) (0.577) (0.086) (0.067) (0.145) log(1+bank 0.233*** 0.224** 0.165 0.169 0.240 1.719*** 0.177 0.116 -0.030 branch) (0.075) (0.093) (0.106) (0.229) (0.187) (0.629) (0.119) (0.088) (0.093) head work 0.151 0.319** -0.193 -0.286 0.550 -2.575 0.021 0.014 0.045 (0.119) (0.139) (0.250) (0.753) (0.845) (2.443) (0.118) (0.159) (0.232) work leader -0.097 0.043 -0.701*** 0.040 -0.029 -1.024 -0.141 -0.051 -0.450** (0.142) (0.163) (0.232) (0.214) (0.209) (1.337) (0.092) (0.116) (0.185) log(age) -0.695*** -0.778*** -0.476 -0.304 0.150 -0.605 -0.314* -0.194 -0.406 (0.177) (0.212) (0.304) (0.530) (0.518) (1.357) (0.168) (0.123) (0.296) high school 0.504*** 0.406** 0.525* -0.283 -0.487* 0.391 0.213 0.012 0.482 (0.146) (0.164) (0.282) (0.245) (0.266) (0.532) (0.212) (0.218) (0.402) college 1.228*** 1.065*** -0.634* -0.535 -0.620 0.090 -0.057 0.066 (0.196) (0.211) (0.351) (0.409) (0.379) (0.200) (0.205) (0.308) party 0.027 -0.055 0.178 -0.133 -0.182 -0.034 -0.178* -0.077 -0.270 (0.129) (0.127) (0.241) (0.264) (0.207) (0.749) (0.105) (0.098) (0.204) Fin class 0.369** 0.395*** 0.414 0.344 0.481** -0.570 -0.051 -0.011 0.130 (0.162) (0.148) (0.452) (0.228) (0.200) (1.261) (0.139) (0.137) (0.314) Fin literacy 0.209 0.220 0.178 -0.096 -0.002 0.172 0.251 0.276 0.254 (0.151) (0.177) (0.254) (0.247) (0.221) (0.679) (0.202) (0.278) (0.297) Econ literacy -0.050 0.121 -0.188 -0.075 -0.004 0.070 -0.202* -0.112 -0.258** (0.108) (0.150) (0.151) (0.243) (0.201) (0.707) (0.110) (0.156) (0.118) Risk for return 0.048 0.206 -0.102 -0.045 -0.117 0.527 -0.034 0.265 -0.245* (0.098) (0.152) (0.131) (0.212) (0.178) (0.548) (0.126) (0.244) (0.126) constant 2.839*** 3.407*** 1.657 6.508*** 5.028** 8.808 0.807 0.376 1.248 (0.744) (0.878) (1.245) (2.151) (2.150) (6.081) (0.687) (0.536) (1.192) Observation 9581 5467 4114 1797 1342 455 8585 4617 3968 R square 0.105 0.130 0.104 0.197 0.222 0.410 0.048 0.073 0.068 Standard errors clustered at the community level in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01; 42