When Shadow Banking Can Be Productive: Financing Small and Medium Enterprises in China

Abstract Small and medium enterprises (SMEs) represent the backbone of China’s economy, yet they face challenges in accessing bank credit. SMEs thus rely on a wide range of alternative sources, including informal finance, online peer-to-peer (P2P) platforms, registered non-banking financial institutions (NBFIs), and underground financiers. This article distinguishes among different types of ‘shadow banking’ to clarify popular misconceptions about the nature of risks associated with informal financial intermediation in China. Given their familiarity with local business conditions and needs, regulated and well-managed NBFIs could provide an enduring foundation for commercialised financial intermediation serving SMEs.

There is a mismatch between China's real economy and the financial system. The country's real economy is largely comprised of farmers, small and medium-sized businesses, and yet the financial sector is dominated by big banks that prefer to deal with big companies.
-Justin Yifu Lin, Peking University, 28 August 2014 The Asian financial crisis provides ample evidence of the risks associated with the absence of a balanced financial sector with multiple channels of financial intermediation. NBFIs play an important role in a balanced and diversified financial sector that is relatively robust and stable.
-The World Bank, 2002 In recent years, alarmist headlines about the rise of shadow banking in China have obscured fundamental structural realities about the relationship between the country's economic growth and financial development. Small and medium enterprises (SMEs) account for over 96 per cent of registered enterprises in China (National Bureau of Statistics [NBS], 2014), and 98 per cent of SMEs are controlled by private entities. 1 SMEs also employ nearly 65 per cent of the workforce and generate 60 per cent of China's GDP (Asian Development Bank [ADB], 2014). Yet state-controlled enterprises receive the majority of loans extended by state-owned commercial banks (Lardy, 2014, p. 161), and account for over 60 per cent of publicly listed businesses on the Shanghai Stock Exchange (China Securities Regulatory Commission [CSRC], 2014). Given such financing constraints, it is not surprising that SMEs rely on a remarkable variety of non-banking financing mechanisms. These include traditional forms of informal finance, online peer-to-peer (P2P) platforms, microfinance institutions, and legally registered non-banking financial private entrepreneurs, the Party made the strategic decision to incorporate and legitimise commercial profit (Dickson, 2003).
Official legitimation of red capitalists removed a major political barrier to larger-scale private enterprises. Concurrently, the state sector restructuring policy of 'grasping the large, letting go of the small' (zhuada fangxiao) in the late 1990s led to de facto privatisation of smaller state-owned enterprises (SOEs) through the establishment of limited liability companies (with fewer than 50 shareholders), employee shareholding cooperatives, and management buy-outs (Garnaut, Song, Tenev, & Yao, 2005). By 2003, over 85 per cent of small and medium-sized industrial SOEs had undergone restructuring (Zeng, 2013, p. 2). That same year, 'Temporary Regulations on the Classification of Large, Medium, and Small Enterprises' differentiated among seven types of industries; and categorised the size of enterprises according to revenues, total assets, and number of employees, rather than gross output. 4 The present 'Standards of Classifying SMEs', issued in 2011, differentiates among 15 sectors and introduced the category of 'microenterprises'. 5 Online Appendix Section A shows how business size (large, medium, small, micro) is defined by sales revenue and number of employees in different industries.
Estimating the percentage of SMEs among all enterprises and the percentage of SMEs that are privately owned, respectively, requires triangulating among different categories of data from the National Bureau of Statistics (NBS). The NBS's Third National Economic Census (2014) reports that as of year-end 2013, there were 8.2 million 'corporate enterprises' (qiye faren), of which 7.85 million or 96.5 per cent were micro and small corporate enterprises. 6 Estimates that include medium enterprises have found that SMEs account for at least 98 per cent of registered enterprises in China (Shi, 2012). Other sources echo this estimate that 98 per cent of businesses are SMEs, though some include over 51 million individual entrepreneurs (getihu) in the numerator (State Administration of Industry and Commerce [SAIC], 2015), leading to statements that over 99 per cent of businesses in China are SMEs (Lin, Qin, & Chi, 2014). Technically, getihu are not considered 'enterprises', so for the present purposes, we will use the 98 per cent figure for characterising the share of enterprises that are SMEs.
Although SMEs may be owned by state or private entities, official statistics on SMEs do not differentiate them by ownership type. Hence, to estimate the number of SMEs that are private, we start from the number of corporate enterprises that clearly report whether the majority of their holdings are controlled by private or state entities (excluding Hong Kong, Macau and Taiwan-holding enterprises and foreign funded enterprises) to reach a total of 7.5 million enterprises (http://data.stats. gov.cn). In other words, we count only the 7.5 million out of the total of 8.2 million corporate enterprises where the majority or dominant owners are either state or private. Starting from the base (denominator) of 7.5 million enterprises with clear majority or dominant holdings, and assuming that 98 per cent of all enterprises are SMEs, two estimates of the share of private SMEs can be derived depending on whether one assumes that the remaining 2 per cent of large enterprises are privately controlled or state controlled. Assuming that all large enterprises are privately controlled (2% share of large enterprises x 7.5 million enterprises = .15 million large enterprises), the number of private SMEs comes to 6.9 million, representing 94.1 per cent of all SMEs. If we assume that all large enterprises are state controlled, the number of private SMEs is 7 million, accounting for 96.1 per cent of all SMEs. (Table 1 summarises these calculations.) This article thus adopts the working assumption that approximately 95 per cent of China's SMEs are privately owned or controlled.
Private SMEs not only account for the overwhelming majority of registered enterprises, but have contributed significantly to the market-oriented core of China's reform era economy. Even during the Hu Jintao administration that advanced the official slogan of 'promoting the public sector, diminishing the private sector' (guojin mintui), SME growth persisted. Large SOEs benefited from the policy environment of the mid-2000s. Yet the development of SMEs in China more than kept up with the state sector. The following graphs show that despite occasional dips in growth rates, the general arc of private enterprise growth, private sector urban employment, and SME share of national industrial output has been robust (see Figures 1-3).
When shadow banking can be productive 2007 In addition to these indicators, private firms in general (that is, including large ones, not just SMEs) have outperformed the state sector in return on assets (ROA). Figure 4 shows that the private sector has consistently maintained a higher ROA than state-owned industrial enterprises, and the gap in ROA has widened markedly since the global financial crisis. Although both private and stateowned industrial enterprises experienced declines in ROA between 2007 and 2009, the financial performance of private businesses recovered more quickly and continued on an upward trajectory. By contrast, even with access to subsidised credit, SOEs face continuing challenges.

Limits on bank lending to private SMEs
Despite their on-going contributions to China's economic development, SMEsespecially small and micro enterprisesface significant barriers in accessing credit from state-owned commercial banks. Based on data from the People's Bank of China (PBOC), Table 2 presents the distribution of total loans extended by formal financial institutions to large, medium, and small and micro enterprises from 2009 to 2014; and Table 3 summarises the overall percentage of loans extended to SMEs during the same period. (The PBOC did not report lending to firms by size until 2009.) Notes: a Out of the 7.5 million enterprises that clearly report whether the majority of their holdings are state or privately held, 7.06 million are privately controlled; 220,000 enterprises are state controlled, and 210,000 enterprises are collectively controlled.
As seen in the third column of Table 2, the percentage of bank loans extended to SMEs increased from 54.9 per cent in 2009 to 65.9 per cent in 2014, which suggests considerable progress in increasing SME access to bank finance. However, access to bank loans is inversely proportional to firm size. Table 3 shows that as of 2014, nearly 54 per cent of SME lending was concentrated among medium enterprises, which account for less than 5 per cent of total SMEs. By contrast, in aggregate, the 96.5 per cent of small and microenterprises received only 46 per cent of formal sector loans extended to SMEs.
The PBOC data is reflected in various surveys of enterprises. A national survey conducted by the All-China Federation of Industry and Commerce in 2009 found that 90 per cent of medium enterprises received bank loans, while only 30 per cent of small enterprises ('above scale', annual sales revenues over 5 million yuan), 10 per cent of small enterprises ('below scale', annual sales revenues less than 5 million yuan) and 5 per cent of micro enterprises were covered by bank loans (All-China Federation of Industry and Commerce [ACFIC], 2010). A 2011 NBS survey of 38,000 small and micro enterprises similarly concluded that only 15.5 per cent had access to credit from the regular banking system (Tencent Finance, 2011).  When shadow banking can be productive 2009 Access to working capital loans is even more restricted: as of 2009 only 4.85 per cent of short-term loans went to SMEs (Lardy, 2014;PBOC Online Database). Indeed, one of the paradoxes of China's reform-era growth is its maintenance of a bank-dominated financial system that privileges lending to SOEs. There are five main reasons why the most productive, market-oriented sector of China's economy is marginalised by the state banking system. 7 The first is rooted in political concerns about supporting the state sector and maintaining social stability. State banks have been pressured by local governments to provide soft loans to SOEs as a means to avoid mass unemployment. As a result, state banks accumulated high levels of nonperforming loans (NPLs) by the 1990s (Lardy, 1998). Although China's 'Big 4' commercial banks were subsequently recapitalised prior to their IPOs in the 2000s, NPLs have grown and re-emerged as a potential concern since the stimulus-induced bank lending of the late 2000s (Weinland, 2015;Zhang, Tian, & Li, 2012).  The second reason that SMEs face challenges in accessing bank credit reflects the state's developmental priorities. State banks extend loans to enterprises in sectors identified for preferential treatment as a matter of domestic industrial policy (for example, automobiles, pharmaceuticals, computer components). Official media occasionally exhorts banks to increase their lending to SMEs, but such statements have not been backed up by policy directives to ensure their implementation (Garaci, 2012).
Third and relatedly, financial repressionmeaning governmental suppression of interest rates below market levelshas served industrial policy at the expense of savers. Under financial repression, household savings have been transferred through the banking system to subsidise credit to SOEs, capital-intensive industry, and real estate developers (Lardy, 1998;Pettis, 2013). Interest rate ceilings also inhibit lending to SMEs because banks are not able to price loans to reflect risk.
The fourth reason for the private sector's exclusion from official credit relates to the limited organisational and technical capacity of state banks to serve the non-state sector. Credit officers have not been trained to evaluate loan applications according to commercial standards of creditworthiness, and until the late 1990s, they were not held accountable for high levels of SOE NPLs in their portfolios. The anti-corruption campaign launched by China's present leadership has reinforced the reluctance of state banks to lend to SMEs, which carry a higher perceived risk of payment delays, if  not defaults (field interviews by author in Beijing, Deqing, Guangzhou, Hangzhou, Ningbo, Shanghai, andWenzhou, 2013-2016). Fifth and finally, bank lending to private SMEs during the initial decades of reform was also limited due to the ideological and political sensitivity of using party-state resources to support capitalist ventures. Loan officers were hesitant to work with private entrepreneurs due to uncertainty about the potential consequences. If a SOE faces payment difficulties or defaults on a loan, the problem is contained within the public sector. However, as many bank managers have (repeatedly) explained to the author, they are much more likely to be reprimanded by superiors for losses incurred through loans to private individuals and SMEs (Field interviews by author, 1996-2016). In short, state banks remain institutionally biased against private sector clients.
Given these structural constraints on private sector borrowing from state banks, China's private SMEs have turned to non-banking sources of credit. In surveys of private businesses conducted during the mid-1990s and mid-2000s, over two-thirds of the respondents indicated that they had used some form of informal finance (Tsai, 2002). More recent research indicates that reliance on nonbanking financing mechanisms has not abated. A World Bank survey of 2700 private firms in 2011-2013 found that only 25 per cent had bank credit and 90 per cent drew on internal financing (World Bank, 2012). Within that period, a 2012 survey of SMEs in 15 provinces conducted by China's Central University of Finance and Economics found that 57.5 per cent had participated in informal credit markets (Li & Hu, 2013). The bi-annual national surveys of private enterprises administered by the All-China Federation of Industry and Commerce (AFIC, various years since 1992) consistently find that 'accessing bank credit' is among the top self-reported challenges facing the private sector. As such, China's SMEsespecially smaller onescontinue to rely on non-banking financial intermediaries.

Defining the universe of non-banking financial intermediation
Key conceptual issues need to be addressed before outlining the various expressions of non-banking financial intermediation in China. Above all, in China's regulatory context, 'informal' finance does not necessarily mean that it is illegal. Most forms of informal financeand what is now known as shadow bankingare 'semi-formal' in the sense that some financial institutions are not supervised by the Central Banking Regulatory Commission (CBRC), but are registered with local branches of other bureaucracies, such as the Industrial Commercial and Management Bureau, Civil Affairs Office, or local Finance Office. Table 4 delineates the registration and regulatory status of NBFIs used by SMEs.
The main difference between banks and licensed NBFIs is that the latter are not permitted to accept deposits. Meanwhile, some forms of informal finance are unregulated or loosely regulated; and others are indeed, illegal. The CBRC stipulates that only banking entities with financial licenses are permitted to mobilise savings deposits. (Online Appendix Section B summarises recent regulations governing various NBFIs.) However, these regulations are not well enforced outside of the formal financial sector. An innovative array of informal financing mechanisms has flourished, mainly in response to unmet demand for financial services, but also in response to arbitrage opportunities due to financial repression.
The diffusion of 'shadow banking' into popular discourse in the aftermath of the 2008 global financial crisis has complicated the preceding conceptual distinctions by casting a pejorative film over non-banking financial intermediation in general. 8 The irony is that shadow banking originally referred to opaque products issued by well-established formal banks. The term 'shadow banking' gained prominence in 2007 in the context of financial institutions in the United States that engaged in 'maturity transformation' by using short term funds (for example, deposits) to finance longer term assets (for example, real estate) (Kodres, 2013). Specifically, the Federal Reserve Bank of New York identifies securitised financial instrumentsincluding asset-backed commercial paper (ABCP), assetbacked securities (ABS), collateralised debt obligations (CDOs) and repurchase agreements (repos)as shadow banking (Pozar, Adrian, Ashcraft, & Boesky, 2010). When shadow banking can be productive 2013 The Financial Stability Board (FSB) (2013) later defined shadow banking more broadly as 'the system of credit intermediation that involves entities and activities fully or partially outside the regular banking system, or non-bank credit intermediation in short.' 9 Based on this expansive definition, the IMF (2014) observes that growth in shadow banking is outpacing that of the regular banking sector in emerging market economies, where shadow banking assets as a percentage of GDP expanded from 6 to 35 per cent during 2002-2012. Globally, the scale of shadow banking assets exceeds global GDP, and is regarded as a particular risk for the United States where shadow banking assets exceed the value of traditional bank assets. Credible estimates of the scale of shadow banking in China were 25 trillion yuan in 2012 and 32.7 trillion yuan in 2013 (Elliott, Kroeber, & Qiao, 2015;Shi, 2014;Wang & Li, 2013), representing 48 per cent and 57 per cent of China's GDP in those years, respectively. By contrast, in the United States the value of shadow banking exceeds GDP by over threefold. Figure 5 shows the global distribution of NBFI assets, with the United States, Euro area, and UK accounting for 79 per cent and China's shadow banking sector accounting for only 4 per cent of the total.  Notes: a Note that in English, '小额信贷公司' (xiaoe xindai gongsi) refers to microfinance, microcredit, and small loan companies. Some institutions that call themselves 'microfinance companies', however, provide larger loans to SMEs. Entities that refer to themselves as 'small loan companies (SLCs)', generally focus on SMEs rather than micro entrepreneurs. Depending on the nature of the license, some SLCs face caps on loan size, but others do not face such restrictions.
The FSB's encompassing definition of shadow banking may help regulators identify sectors that require prudential oversight. Given the term's negative connotation, however, lumping all forms of non-banking financial intermediation into 'shadow banking' offers limited analytic value for understanding the types of activities that carry systemic risk versus those that are established features of a healthy financial system serving diversified markets. Rather than mince the precise boundaries of formal versus informal and legal versus illegal finance, Table 5 presents the forms of non-banking financial intermediation used by SMEs in China according to their degree of institutionalisation. While recognising that informal norms and practices may be far more 'institutionalised' than those enshrined in formal institutions (Castells & Portes, 1989), in the context of financial intermediation, 'degree of institutionalisation' can be assessed by the extent to which they entail documented rules and procedures; the nature of their enforcement mechanisms; the stability of their physical or virtual operating space; and whether they are registered with a government entity. 10 Generally speaking, the more institutionalised a financing practice, the more likely it is to be subject to registration requirements and attract regulatory attention. As discussed below, however, the mere fact of registration does not mean that they are adequately regulated or well-managed internally.
Casual, interest-free, uncollateralised borrowing from family, friends, and business associates are ad hoc and lie at the un-institutionalised end of the spectrum. Legitimately registered NBFIs with modern corporate governance, accounting, information management, and credit-screening systems, lie at the other. 'Semi-institutionalised' practices include rotating credit and savings associations, which are unregistered but operate according to self-enforced written contracts binding group members (Geertz, 1962). Similarly, non-governmental investment alliances are documented by written agreements between private entrepreneurs who pool funds to make larger investments in member enterprises than would otherwise be feasible.
Various types of internet finance, such as crowdfunding sites (for example, DemoHour, DreamMore) and P2P lending platforms (for example, Ppdai.com, Renrendai.com, Dianrong.com) also fall into the semi-institutionalised category because they broker funds through parameters that are documented on-line, but are unregulated and not necessarily legitimately registered. Internet finance lies in uncharted regulatory waters. Because on-line platforms put private savings at risk, essentially substituting for bank deposits, the CBRC has been charged with developing a regulatory framework for internet finance. Over 1000 P2P sites have already ceased operating since 2011 (Weinland, 2014). Figure 6 shows the various forms of non-banking financial intermediaries according to their degree of institutionalisation and regulation.
'Degree of regulation' is defined by the extent to which: 1) there are official documents outlining the scope of permissible activity for a specific type of financial intermediation; and 2) the assignment of enforcing those rules to particular regulatory authorities (see the fourth column of Table 4). Hence, the most 'highly regulated' NBFIs are those regulated by ministerial-level financial supervisory entitiesalso called, 'one bank, three commissions' (yihang sanhui), meaning the PBOC plus the CBRC, CSRC, and the China Insurance Regulatory Commission (CIRC). Those reporting to provincial governments (that is, credit guarantee companies (CGCs), microfinance companies, and rural credit cooperatives (RCCs)) are considered to be more weakly regulated due to variation among local governments in managing NBFIs. The one exception to this regulatory classification is internet finance, which falls under the CBRC's regulatory scope, but in practice the regulatory framework is still evolving. Hence, we regard P2P lending and crowdfunding as weakly regulated.
Given that NBFIs are not as tightly regulated as state banks, one might question the empirical value of distinguishing among their degrees of comparatively weak regulation. Specifically, how can we tell if variation in the extent of NBFI regulation matters in practice? The use of the court system for resolving financial disputes is a potential indicator of regulatory effectiveness. Financial transactions involving more highly regulated NBFIs are more likely to be protected under law; hence, it is plausible that regulation would inspire a higher volume of litigation. In the case of China, however, it is apparent that the court system receives more cases involving less regulated informal financial intermediaries. PRC law distinguishes between debts to financial institutions, 'financial institution When shadow banking can be productive 2015 loans', and those to individual lenders, 'private loans'. Loans extended by NBFIs regulated by ministerial-level entities (trust companies, pawnshops, and financial leasing companies) are regarded as financial institution loans. NBFIs approved by local governments (CGCs, microfinance companies, and RCCs) and barely/un-regulated forms of financial intermediation (P2P, interpersonal loans) are treated as private loans. Litigation over private loans has escalated sharply in recent years: in 2011, China's courts received 594,000 private loan cases; by 2014, the number of private loan disputes reached 1.02 million, and there were already 526,000 cases by the first half of 2015.
Court cases involving private lending have become the second largest category of civil cases, following marriage and family disputes (Wang X, 2015). In response to this unexpected expansion in private financial disputes, in August 2015 the PRC Supreme Court issued 'Regulations on Applying Legal Provisions in Private Loan Cases' to clarify the parameters of legal protection for private debt transactions, including those brokered on P2P platforms. The notice indicated that if the private loan contract entails annual interest rates of up to 24 per cent, then lenders may request judicial assistance in requiring borrowers to pay the interest. If interest rates exceed 36 per cent, the upper limit for legal lending, then borrowers have the right to ask lenders to return the portion of interest paid exceeding 36 per cent. If a P2P operator implicitly or explicitly pledges to guarantee loans brokered on its platform, then it is legally obligated to honour that guarantee. When P2P platforms engage in fraudulent or illegal fundraising, however, then the dispute is treated as a criminal rather than a civil case. The combination of these judicial pronouncements and heightened use of the court system for resolving private financial disputes indicates that the regulatory status of various NBFIs is indeed relevant for understanding the nature of risk posed by shadow banking operators.
The above typology focuses on the types of non-banking financial intermediation used by SMEs in China rather than the entire universe of what is now associated with shadow banking. In 2014, the PRC State Council (2014) identified three categories of shadow banking from a regulatory perspective: (1) Unlicensed and unregulated financial operators (for example, underground banks, internet finance, P2P lending) (2) Entities operating without a finance license and inadequate supervision by credit agencies (for example, pawnshops, CGCs, microfinance companies) (3) Inadequately supervised activities of licensed financial institutions (for example, money market funds, wealth management products, asset securitisation).
The State Council's definition represents a reversal of the CBRC's statement, just two years earlier, that NBFIs (such as trust companies, finance companies, and off-balance sheet transactions of banks) did not constitute shadow banking (Gao & Wang, 2014). This reversal may be attributed to anxiety of regulators and industry observers about the unexpectedly rapid growth of two forms of shadow banking: local government financing vehicles (LGFVs) and wealth management products (WMPs). Neither is associated directly with the NBFIs serving SMEs, however.
LGFVs broker off balance sheet loans between state banks and local governments who offer revenues from land sales and real estate as collateral.
LGFVs have become the primary channel through which sub-national governments have financed public goods and large-scale infrastructure projects. Following an emergency national audit, the National Audit Office (2013) estimated that by mid-2013 local government debt had reached 17.9 trillion yuan ($2.9 trillion), an increase of 60 per cent since the end of 2010.
LGFVs pose systemic risk due to their reliance on new debt to finance longer-term investment (Nomura, 2013).
Meanwhile, in partnership with banks, trust companies have been offering WMPs that advertise returns two to four times higher than deposit rates in savings accounts (Zhang, 2014). Bank WMPs fall into two categories: 1) traditional bank business that is moved off balance sheet to evade reserve requirements and sector-specific lending quotas; and 2) speculative investment products that channel funds into projects that banks would not ordinarily support (primarily real estate). Both types entail maturity mismatch, but the second holds greater potential for triggering crisis when redemption is below par and either investors lose money or the banks are forced to absorb the losses. In 2014, near defaults of large WMPs issued by trust companies associated with China's largest state banks drew attention to the systemic risk posed by the rapid accumulation of assets held by trusts in recent years. 11 Unlike banks, which are required to set aside 20 per cent of deposits to protect savers against loan defaults, trusts are not subject to a reserve requirement.
The nature of risk posed by the proliferation of LGFVs and WMPs should be distinguished from that associated with NBFIs that provide financial services to SMEs. Rather than focusing on the current state of regulation, assessing the exposure of banks to different types of credit intermediation provides a better metric for analysing the risk posed to financial stability. A research report on shadow banking in China identifies three layers of shadow banking based on such criteria: the bank off-balance sheet financing layer, the credit enhancement layer, and the non-bank lending layer (Sheng, Edelmann, Sheng, & Hu, 2015).
In the top bank off-balance sheet financing layer, banks evade regulatory restrictions on capital requirements and loan-to-debt ratios by extending credit through off-balance sheet 'channelling.' The When shadow banking can be productive 2017 off balance-sheet supply of WMPs through trusts is the prime example of this top layer of shadow banking. The main source of risk stems from the mismatch between asset risk and investor appetite for risk; consumers may not understand the risk associated with securitised assets advertising returns substantially above the bank deposit rates.
In the second credit enhancement layer of shadow banking, non-banking financial companies extend loans to businesses unable to access bank credit, or enable lower-credit borrowers to access larger bank loans by providing guarantees. There is potential risk to banks because such institutions serve borrowers that unassisted, might not meet banks' credit standards. For example, when one business faces debt service challenges in a reciprocal loan guarantee network, all the other bank loans guaranteed by that business, either directly or indirectly, are also at risk.
The third non-banking lending layer is the most distant from the banking system itself. Pawnshops, leasing companies, P2P platforms, and microfinance companies serve private businesses in various sectors. These types of NBFIs face risk themselves, but pose limited systemic risk to the banking system. Credit failures are generally contained to the immediate operators and clients of such intermediaries. Moreover, non-deposit taking NBFIs do not create the same implicit obligation on the government to cover investment losses.
The triple-tiered framework for assessing the degree of systemic risk associated with shadow banking is helpful for differentiating the original type of shadow banking (the top bank off-balance sheet layer that involves maturity transformation through asset pooling and securitisation), from the more traditional forms of informal finance that have emerged in response to SME demand for credit (the credit enhancement and non-banking lending layers). Within the second and third layers of shadow banking, additional distinctions can be made in terms of risk assessment. The State Council's regulatory-based definition of shadow banking provides a starting point, along with the intermediary's degree of institutionalisation. Returning to Figure 6, the entities towards the upper right quadrant of the graph have a more secure registration status, regulatory standing, and level of institutionalisation. Those include a variety of NBFIs, such as microfinance, credit guarantee, and leasing companies. Further differentiation among registered, regulated, and well-institutionalised NBFIs requires firm-level analysis of corporate governance, management, internal systems, and so on.

Examples of Chinese NBFIs
For illustrative purposes, we present three examples of NBFIs that are registered, hold operating licenses, and benefit from local governmental supportbut vary in their management, scope of business, and performance. The first company, Zouli Kechuang Micro-Finance Company ('ZKMFC'), went public in 2014 after operating for only three years. The second, Credit Orienwise, developed an impressive market share in China's credit guarantee industry and attracted global venture capital, but was subsequently derailed by accusations of financial malfeasance. The third example, Gangyu Guaranty, is embedded within a broader network of NBFIs that provides different types of financial services for SMEs.

Zouli Kechuang Micro-Finance Company
Based in Deqing County in the southern coastal province of Zhejiang, Zouli Kechuang Micro-Finance Company ('ZKMFC') was the first microfinance company in China to list on the Hong Kong Stock Exchange (06866:HK). Relative to other provinces in China, Zhejiang is known for having a particularly innovative range of informal financial intermediaries that serve the province's thriving, but credit-constrained private sector. 12 By 2016, there were 336 microfinance companies in Zhejiang, and ZKMFC was the largest one with registered capital of 880 million yuan (US$141 million) (interviews by author, Deqing, June 2016).
Despite ZKMFC's rapid expansion and close relationship with county, municipal, and provincial government officials, its founding circumstances were complicated in China's emergent regulatory context for microfinance (interviews by author, Deqing, June 2016). After the central government granted legal status to microfinance companies (MFCs) in 2008, Zhejiang province encouraged counties to establish pilot MFCs; in turn, Deqing County decided that its Finance Office would only grant MFC licenses to entrepreneurs who operate a publicly-listed company. 13 Hence, in 2013 the key shareholders of Zuoli Yaoye, a pharmaceutical company on the Shenzhen Stock Exchange (SSE)'s Growth Enterprises Market (chuangye ban), applied for and received a MFC license in Deqing. However, SSE authorities informed Zuoli Yaoye that it would not receive further funding from the SSE if it entered the microcredit business. SSE's stated rationale was that it did not want its high growth-start-up companies to be exposed to the financial risk associated with microfinance. To circumvent this unexpected obstacle, the shareholders of Zuoli Yaoye established Zuoli Holdings as an umbrella entity to purchase another business, Puhua Energy. Since Puhua was part of an enterprise group that included a listed company (Zuoli Yaoye), it was able to contribute 200 million yuan, along with seven corporate shareholders and 15 individual shareholders to establish ZKMFC. In accordance with MFC registration requirements, loans cannot be extended to its shareholders.
Once established, ZKMFC expanded rapidly, outperforming the other four MFCs in the county. ZKMFC provides short-term loans (2-12 months) to local SMEs, microenterprises, and individual entrepreneurs in agricultural and rural industrial activities. Their portfolio includes both enterprise and individual loans ranging from 10,000 yuan (US$1,600) to 25 million yuan (US$4 million); as of 2015, 87.2 per cent of their loans were less than 1 million yuan. By 2015, ZKMFC had served over 1545 clients with short-term financing, and accounted for 65 per cent of the total loan balance of MFCs in Deqing (Zuoli Kechuang Micro-Finance Company [ZKMFC], 2016). ZKMFC manages credit risk by maintaining strict separation between the loan application and approval processes. ZKMFC's 'client relationship managers' conduct due diligence to verify information supplied in the loan application, including ascertaining the creditworthiness of the applicants and their guarantors, as well as the value of pledged collateral. Besides examining bank accounts, electricity bills, and on-site verification of production, the relationship managers also inquire into the applicant's reputation in the local community (shehui xinyu). As senior management explained, 'We have extensive networks in Deqing, so we can readily find out if a local entrepreneur has a history of being unreliable or unsuccessful at business. That is more reliable than only looking at balance sheets' (interview by author, Deqing, June 2016). Following these processes, the credit investigators submit a report to ZKMFC's credit approval department. The latter maintains three tiers of loan assessment and approval depending on loan size.
Meanwhile, the emergence of internet finance has broadened both their customer bases and sources of credit data. In April 2015, the Zhejiang Province Finance Office approved ZKMFC to extend loans of up to 500,000 yuan to online customers. Although ZKMFC has developed software for monitoring the condition of virtual borrowers (for example, through online sales records), it has been much more stringent in approving 'internet loans' because prospective borrowers apply from all over the country. As such, as of June 2016, clients based in Deqing still accounted for 90 per cent of ZKMFC's loan portfolio (versus 10% online borrowers), which gives them greater confidence in their ability to assess and manage credit risk. Indeed, a critical part of ZKMFC's system of risk control entails postlending reviews. The latter includes monitoring interest repayment patterns, tracking the value of collateral, and conducting on-site visits to identify any changes in business volume (ZKMFC 2014, p. 96).
These risk management processes have been effective thus far. On the eve of its IPO in 2014, only 0.1 per cent of ZKMFC's loan portfolio was overdue and its six-month profit margin was 64.6 per cent. Despite the subsequent downturn in local economic conditions, as of year-end 2015, ZKMFC maintained an overdue ratio of 1.08 per cent and a profit margin of 68 per cent (ZKMFC, 2016, p. 12). General Manager Hu Haifeng explained that even though ZKMFC serves clients that are When shadow banking can be productive 2019 generally of 'lower quality' than those who are able to access credit from formal banks, they have managed risk by extending smaller and shorter term loans that pose less risk of delinquency and default.

Credit Orienwise Group
In 1999, a private entrepreneur who made his fortune in real estate, Zhang Kaiyong, founded the first CGC in Shenzhen. Initially, the CGCs owned by Credit Orienwise Group (COG) focused on guaranteeing loans from state banks extended to Shenzhen-based SMEs. Within a few years, COG had expanded to a dozen cities, becoming one of China's largest CGCs (Epstein, 2009a). COG was hailed by the Forum on Inclusive Finance as China's 'first comprehensive non-governmental corporation in the business of guarantees, investment and consulting' (Situ, 2011). By 2003, Zhang Kaiyong was on the Forbes list of the 400 wealthiest Chinese. Morgan Stanley backed a $100 million bond issuance on the Singapore Stock Exchange in 2005, and shortly thereafter, COG attracted over $100 million in private equity investment by the Carlyle Group, Citigroup Venture Capital International, General Electric Capital, and the Asian Development Bank (ADB). At its peak, COG guaranteed $2 billion worth of loans and engaged in $250 million of indirect lending (Epstein, 2009b). Domestically, COG received a steady stream of admiring political leaders, including party secretaries, governors, and Chinese People's Political Consultative Conference (CPPCC) delegates.
However, in August 2008, COG started to unravel from within. As reported in Caijing, Orienwise staff accused the company of fraudulent accounting, and claimed the general manager of a Fujian province subsidiary had exposed COG to 160 million yuan in failed loan guarantees (Caijing Staff, 2008). Domestic banks started to pull their loan guarantees to COG, and international investors learned about the internal crisis 'through an anonymous letter from a former Orienwise executive who accused the company of financial malfeasance' (Zhang, Zhang, Chen, & Fu, 2009). COG's investors threatened to pull out. But Zhang Kaiyong persuaded them to retain their investments and conduct an audit while holding his 54 per cent stake in COG as collateral. An investigation committee comprising of the four international investors, PriceWaterhouseCoopers, and two law firms proposed taking over the operations of COG in the interim. Zhang objected. Four international board members resigned and investors filed a lawsuit against COG in Hong Kong, though Carlyle and Citigroup subsequently backed out (Epstein, 2009a).
In speculative post-mortem accounts of COG's demise, observers note red flags such as COG's registration in the Cayman Islands. At the time, however, even ADB's external legal counsel concurred that working with an offshore entity would be less bureaucratic than a PRC-listed company (Sharman, 2012). Others reflect retrospectively, 'A diffuse holding structure that includes eight offshore and nine onshore entities makes it difficult to identify where assets may be hiding, or indeed, if the original structure still exists' (Robertson, 2014). Ultimately, COG's failure was probably a combination of non-transparent governance, coupled with internal mismanagement and under-regulation.
Although COG's case attracted international media coverage, it was not alone. During the same period, Huading Guarantee Company was similarly exposed for poor risk management and irregular practices (Dong & Shan, 2010). In 2012, the collapse of a prominent CGC in Beijing, Zhongdan Investment Credit Guarantee, revealed that it had been advertising high returns to its customers by using portions of guaranteed loans to invest in WMPs (Yang & Ma, 2012). Similarly questionable practices were linked with waves of CGC failures in Wenzhou and Zhengzhou that same year (Tsai, 2015).
Gangyu Guaranty and non-guaranty affiliates. The negative cases of CGCs discussed above receive more attention than NBFIs that have well-functioning risk management systems in place, externally audited financial statements, and transparent corporate governance structures. Gangyu Guaranty and its affiliated NBFCs provide a more comprehensive, business-oriented counterpoint to the CGCs that are limited to a single financial service. In 2007, Gangyu was established in partnership with a Hong Kong-based investment group, Aktis Hanxi, in the western municipality of Chongqing (Aktis Hanxi Group, 2011). 14 As the first Sino-foreign CGC in Chongqing, Gangyu prides itself on using tighter credit scoring standards, conducting more in-depth due diligence of clients, and closer monitoring of staff and workflow processes than other CGCs or even state banks. 15 With extensive experience in foreign commercial and investment banks, Gangyu's board members and senior leadership introduced a host of modern managerial practices, such as quarterly board meetings that include progress reports from various operations coordinators. In addition, unlike state banks and most CGCs, annual key performance indicators (KPIs) incentivise middle managers to maintain high quality portfolios, while generating increasing revenues. Quarterly board meetings are followed up with on-site monitoring by the board chairman.
When screening clients, Gangyu's marketing department verifies the legitimacy of operating licenses and ensures that the clients are not involved in any litigation. The finance department goes beyond standard review of cash flow statements by verifying tax records, and studying trends in electricity and gas bills to confirm production activity. They even examine owners' individual bank accounts. The credit department evaluates the value of the collateral and the ease of liquidating it. Each of these departments writes independent reports and submits them to a credit committee, which then makes a recommendation to the Board with a credit limit. The Board has the authority to veto a credit recommendation, but cannot autonomously accept one that has not been approved earlier on in the process.
Gangyu also has a designated post-lending monitoring department, which enables them to identify risk early on and take action to minimise losses. While banks assign SME risk according to five grades, Gangyu differentiates them by eight grades, which determines how frequently they visit clients. These practices earned Gangyu a national award for Best Risk Control in 2012 by the China Guarantee Association. As a further testament to its professional reputation, when banks became more cautious in extending credit to other CGCs due to declining economic growth in 2014, Gangyu was the only one of two CGCs in Chongqing that continued to receive increased credit lines from banks. As of year-end 2015, Gangyu's default rate since inception was only 0.79 per cent (interview by author, July 2016).
Meanwhile, despite softening in macroeconomic conditions, as of 2016 the profit rate of Gangyu's clients in Chongqing remained robust, albeit with typical sectoral variation. In the medical equipment industry, the average profitability of Gangyu borrowers ranged from 15 to 18 per cent; the average profitability among logistics clients was 12 per cent; among producers of automobile and motorcycle parts, the average profit rate was 8 per cent; and even within the increasingly beleaguered manufacturing sector, Gangyu's clients were still yielding 5 per cent profitability as of 2016. As one of the most stable and professional CGCs, Gangyu has built a vast network of working relationships with the local government, local banks, and SMEs; and maintains a database with the credit history for hundreds of local businesses. Through other Aktis-invested NBFCs, Gangyu is linked to a broader platform of credit guarantee, small loan, financial leasing, and fund management companies in Chongqing, Hubei, and Chengdu. Individually, none of them are as large in scale as ZKMFC or Credit Orienwise; taken together, they deploy 1.4 billion yuan of equity capital, enabling them to serve different financing needs of SMEs as they move through various cycles of development. For example, financial leasing is more practical for longer-term investments and expansion of production than a short-term loan from a small loan company. In other circumstances, going through a CGC to secure a medium-term loan from a bank may not be as convenient when a business faces urgent bridge financing. Given the fragmented nature of NBFI regulation and licensing in China, most NBFIs focus on developing clients for a single product, which facilitates economies of scale, but limits flexibility in adapting to client needs at varying stages of their business cycle (interviews with Gangyu managers, December 2014; March 2015; June 2016). If well managed, as seen in Gangyu, a diversified network of licensed NBFIs provides a more comprehensive package of products, while diversifying risk exposure.
When shadow banking can be productive 2021

20.68
Notes: 1) % of SME lending from CGCs came from the China MSME Finance Report 2014. 2) % of SME lending from micro and small loan companies the 39 per cent only includes small and micro enterprises (excluding getihu); medium enterprises were not included in the survey. 3) % of SME lending from pawn shops came from the MSME Finance Report (pp.18), 80 per cent of pawn shop loans went to SMEs by the end of 2013; it is unclear whether getihu were included. 4) % of SME lending from trust companies is not aumulative. Out of the 13 trillion in total funds, 3.13 trillion went to industrial and commercial enterprises by the end of 2014. In 2014, 990 trust products were issued with a total financing scale at 128.2 bn, of which 30.6 bn was SME financing, accounting for 38.9 per cent of total trust products. 5) The China MSME Finance Report only mentioned that the 'small and micro enterprises contract balance reached around 200 bn RMB by the end of 2013, out of the 2.1 trillion contract balance from all financial leasing companies. 2.1 trillion is the balance from all financial leasing companies including those managed by the CBRC (around 30 NBFIs) and those managed by MOCOM (around 1,000 non-NBFIs); and '10 per cent' is a rough estimation (MSME Report pp. 171). 6) The number of P2P lending platforms is the number of normally operating platforms instead of total platforms (from Wangdaizhijia http://www.wangdaizhijia.com/news/baogao/24321.html). Total loan balance (wangdai hangye daikuan yu'e) comes from the Wangdaizhijia monthly report (Oct. 2015). 7) The number of crowdfunding platforms is the number of "normally operating" platforms instead of total platforms (from Wangdaizhijia http://www.wangdaizhijia.com/news/baogao/24489.html). Wangdaizhijia started providing monthly reports on crowdfunding after June 2015. The reported total crowdfunding scale is the total financing scale (项目总融资额) of 6.86 bn RMB, with 70.8 per cent equity financing crowdfunding, and actual financing scale (实际已募集金额) of 3.7 bn RMB, with 29.9 per cent equity financing crowdfunding by the end of June 2015 (from PEDATA report, pp.33, http://www.pedata.cn/resource/freeReport/1440998436813408). 8) The total scale of RCC loans the comes from PBOC annual report; the column is named 'loans extended to non-financial corporations'. a Pawnshops report that 90 per cent of their clients are private enterprises and individual entrepreneurs.
Given that 95 per cent of private enterprises are SMEs and SMEs account for 10.2 per cent of all private businesses (7 mil enterprises /(7 mil enterprises + 61 mil getihu)), we estimate that 9.8 per cent of pawnshop clients are SMEs.

Contribution of NBFIs to SME finance in China
Given that NBFIs serve clients beyond SMEsand the fact that SMEs access financing from non-NBFIsthis section briefly reviews available aggregate data on the scale and client base of various NBFIs. As shown in Table 6, we estimate the scale of credit-related NBFI financing in 2014 at 20.68 trillion yuan. The fourth column of the table reports the proportion of SME clients for each type of NBFI. At the high end, for example, SMEs account for 75.8 per cent of the business of credit guarantee companies, while only 9.8 per cent of pawnshop customers are SMEs (because pawnshops predominantly serve individual entrepreneurs). Note that we were not able to derive confident estimates for the percentage of SME lending by Internet finance (P2P and crowdfunding) and RCCs because they include microenterprises and individual entrepreneurs in their loan portfolio reports. Interviews with P2P operators suggest that the vast majority of their clients are individual entrepreneurs rather than SMEs ( In brief, although the volume of bank lending to SMEs (29.33 trillion yuan in 2013) exceeds the total scale of NBFI financing activity (20.68 trillion yuan in 2014), it is apparent that NBFIs contribute to the financing of SMEs, particularly those smaller in scale.

Conclusion
As in other parts of the world, China's SMEs face a financing gap that is structural in nature. Even though SMEs account for the overwhelming number of businesses, generate employment, and contribute meaningfully to GDP, traditional commercial banks are nonetheless biased towards larger-scale borrowerseven in highly liberalised financial systems. The near-universal bias against SME financing has an economic logic: it is less efficient to process numerous smaller loans; and due to limited credit history, collateral, and/or scale, SMEs carry a higher risk profile. With the exceptions of a handful of OECD countries, 16 the SME funding gap has proven to be an enduring structural feature of both emerging and advanced market economies. This remains the case even in countries that have enacted a variety of policy measures to support SMEs and enhance financial inclusion more broadly.
In China, shadow banking represents a market response to a combination of policy restrictions and related political priorities. At the most basic level, financial repression allows SOEs to receive subsidised credit, while inhibiting the ability of banks to price loans for higher risk SMEs. As such, various types of informal financial intermediaries and NBFIs have emerged to fill the SME funding gap. Some lend directly to private businesses, while others guarantee loans from commercial banks. Meanwhile, suppression of deposit rates has driven savers to seek higher returns from other investment opportunities. Banks thus turned to off-balance sheet products to generate earnings from alternative sectors. The recent rise of online P2P lending and crowdfunding platforms bypasses the banking system altogether by brokering between SMEs and private lenders/investors. All of the above practices are regarded as shadow banking under the Financial Stability Board's expansive definition. But they present different types of risk, and in turn, invite different regulatory responses. The maturity transformation associated with off-balance sheet channelling poses the greatest systemic risk to the banking system. But even within this top bank off-balance sheet layer of shadow banking, China has not developed the volume or complexity of securitised products that triggered the global financial crisis. To avert such snowballing, since 2009 the CBRC has issued various notices requiring banks to disclose risks associated with WMPs, and ensure that WMPs are not linked with deposits (Hsu, 2014).
In the second, credit-enhancement layer of shadow banking, loans backed by CGCs expose banks to more traditional credit risk. Unlike complex derivatives, which may be repackaged by multiple NBFIs, the risk from guaranteed bank loans stems from specific borrowers and should be known to the guarantor, if not the lender as well. However, as seen in the CGC case studies, considerable variation exists in the corporate governance, management, and sophistication of credit scoring systems employed by CGCs. This reflects in part the decentralised regulation of CGCs by local government entities.
In the third non-bank lending layer, risk is concentrated with private lenders and borrowers, so client defaults in NBFIs such as microfinance/small loan companies pose less risk to social stability or the financial system. On the other hand, P2P platforms, and rotating savings associations mobilise individual savings without regulation or government safeguards, posing more of a risk to social stability than the banking system as a whole.
As in the second credit-enhancement layer, the non-banking layer of NBFIs encompasses a diverse range of entities. Some are unregistered, unlicensed, and limited to discrete local networks of lenders and borrowers. Such forms of informal finance avoid regulatory scrutiny as long as they are functional. Many are not, however, which is reflected in the heightened case load involving private lending disputes in China's courts. At the other extreme are licensed NBFIs with transparent governance, and strong information management and credit systems. Some of these NBFIs have attracted foreign equity investment or listed publicly. A handful of CGCs, microfinance/small loan companies, and pawnshops fall into this subset of NBFIs. P2P and crowd funding have been operating in a regulatory void even while mobilising millions of users. Given that internet financing platforms essentially mobilise deposits, the CBRC has been working on appropriate regulations.
China's regulators are actively cognisant of the relationship between financial repression and the modalities of risk posed by different forms of shadow banking. Incremental liberalisation of interest rates on loans has occurred since the late 1990s, and the upper limit on lending rates was removed in 2013. The ceiling on deposit rates was ultimately lifted in September 2015, following creeping increases over the previous year. (See Online Appendix Section C for a chronology of interest rate reform measures.) The implications of recent liberalisation are mixed. On the one hand, removing deposit rate ceilings should dampen demand for WMPs, which pose the greatest systemic risk. On the other hand, offering higher deposit rates would narrow the already shrinking profit margins of banks and SOEs, and could lead to greater interest rate volatility.
Interest rate liberalisation is unlikely to narrow the SME financing gap significantly, however. Besides the sheer volume of SMEs in China, the experience of other countries suggests that interest rate liberalisation does not translate into greater willingness of commercial banks to extend credit to SMEs. Traditional constraints on limited economies of scale, credit evaluation, loan monitoring, and collection are operational, rather than purely financial issues. Furthermore, SME promotion policies, such government-backed loan guarantee schemes, have not proven to be effective for supporting the most promising segment of SMEs.
In short, China's SME financing gap will persist. Despite official efforts at financial inclusion and interest rate liberalisation, NBFIs will remain an important source of financing for China's SMEs. Even in advanced industrialised countries, NBFIs continue to serve critical segments of the economy due to the incentives of banks and securities markets to privilege larger-scale businesses. China's state-owned commercial banks and stock markets face similar incentives, but these are compounded by the political prioritisation of state firms in strategic sectors. Formal financial institutions in China are biased against SME clients for both economic and political reasons. By contrast, NBFIs are unencumbered by these dual constraints. Motivated by the glaring SME financing gap, a remarkable variety of NBFIs have flourished and floundered in China in the past three decades, often in collaboration with local governmental authorities who recognise their contribution to the local economy. Sensationalised accounts of shadow banking obscure the fact that an elite subset of credit guarantee, small loan, and financial leasing companies has developed locally grounded expertise in serving the particular needs of SMEs. Such NBFIs employ rigorous methods of financial accounting, credit scoring, and loan monitoring because they seek risk-adjusted returns on their investment. They are motivated by profit rather than politics.
Going forward, the policy challenge thus lies in establishing a regulatory framework for NBFIs that promotes best practices in a sector that has inspired both financial innovation as well as malfeasance. Fortunately, a modest core of well-managed, legally registered NBFIs provides a foundation for guiding future regulation. Within this core, foreign-invested and listed NBFIs are poised to provide a positive demonstration effect for corporate governance, credit evaluation techniques, and risk management. This is not to say that foreign-invested NBFIs are immune from scandal (for example, Credit Orienwise). But such cases should be distinguished from legitimate NBFIs that are closely supervised by private stakeholders, both domestic and foreign, who are invested in building commercially successful NBFIs that serve SMEs.
Regulators have an opportunity to move such NBFIs out of the shadows. This sector is likely to grow and innovate rapidly and should be managed as a key source of SME funding. Reducing the opportunities for inter-bureaucratic and cross-jurisdictional regulatory arbitrage could be a guiding principle for a process that is bound to be complex. Multiple bureaucracies govern the spectrum of quasi-regulated NBFIs. The non-banking sector has this in common with other sectors in China's fragmented authoritarian political economy. However, social stability remains a leading concern for the regime, and financial stability is viewed as an essential component of that overarching goal. Given that China's present leadership has demonstrated willingness and capacity to centralise authority in priority areas, efforts to consolidate NBFI regulation are likely forthcoming as a component of China's on-going economic reforms and development.