Access to the Internet and Access to Finance: Theory and Evidence

Abstract

This paper aims at investigating the relationship between the use of the Internet and access to external finance of small and micro businesses, both theoretically and empirically. We first develop a theoretical model to explore how access to the Internet affects the credit availability of firms. The model suggests that access to the Internet can effectively mitigate financing difficulty of firms by alleviating information asymmetry and reducing agency cost, and thus can promote the sustainable development of those firms. The model also shows that access to the Internet can improve social welfare based on aforementioned mechanism. Using China household finance data from China Household Finance Survey, we tested the impact of access to the Internet on access to finance of small and micro businesses. Our empirical results confirm the positive role played by access to the Internet in alleviating financing difficulty of those firms. Moreover, we also found evidence that access to the Internet can reduce borrowers’ dependence on physical branches of banks when making bank choice decision for loan applications. Our evidence also implies that access to the Internet is conducive to the sustainable development of small and micro businesses via mitigating their financing difficulty.

1. Introduction

What influence the Internet era is having or will have on personal or firm financing has been a hot topic throughout the world. In recent years, Internet-based technologies, such as mobile payments, social networks, search engines and cloud computation, are leading to a paradigm shift in financial sector. Internet finance has been rendering earth-shaking changes in our daily life and real economic activity, among which changes in individual or firm financing are far-reaching in terms of financial development. Even access to the Internet itself can make a big difference on personal or firms’ access to finance. Access to the Internet may alleviate information asymmetry between financial intermediaries and potential borrowers (another path for access to the Internet to play a part in access to finance is to disintermediate; a case in point for this is the online market for peer-to-peer lending), which in turn may be conducive to improving capital market efficiency, boosting social productivity, lending support to social sustainable development, and elevating social welfare. As evidenced by previous literature [1,2,3,4], financial sustainable development is important not only for economic growth itself but also for economic growth quality, and there exists a significant positive correlation between access to finance and poverty mitigation at the country level. In particular, Bruhn and Love [4] documented that access to finance has a sizeable effect on labor market activity and income levels, which is especially striking among low-income individuals and those located in areas with lower preexisting bank penetration, based on a quasi experiment from Mexico. Rusu and Roman [5] found that access to finance is one of the main macroeconomic determinants of entrepreneurship in 18 European Union countries. Allowing for the ever-growing Internet penetration rate, the rapid development of Internet finance in current era, and the importance of access to finance for real economic activity, therefore, the very basic but fundamental question we investigated in the study is whether access to the Internet can have an important effect on real economy, especially on individual or firms’ access to finance, in the context of emerging economies, such as China.

Small and micro businesses play a positive role in increasing employment, improving people’s living standard, promoting economic growth, etc. According to the data statistics from State Administration for Industry and Commerce of the People’s Republic of China (SAIC), in China, the number of small and micro businesses accounts for 76.57% of the total number of business enterprises as of 2013; meanwhile, if we incorporate individual-owned businesses into the category of small and micro businesses, then the ratio is about 94.15%. There are around 0.31 billion small and micro business employees as of 2016, increased by about 27.82 million from the previous year, according to SAIC. Although small and micro businesses are an important driver for economic development, the growth of small and micro businesses is often constrained by internal finance. In other words, the main development bottle neck of small and micro businesses lies in that they have less access to formal sources of external finance. In fact, financing difficulty of small and micro businesses is universal across the world for both developing and developed countries, which has been documented by substantial research [6,7,8,9]. On the other hand, rare literature directly explores the relationship between the use of the Internet and individual or firms’ access to external finance, although a strand of literature [10,11,12,13,14,15] has investigated the influence of online peer-to-peer lending as an appealing new channel of financing on people’s economic behavior. This study attempted to fill the void in the environment of an emerging economy, China. Put differently, we addressed the following research question: How does individual use of the Internet affect personal or small and micro businesses’ access to external finance or their sustainable growth, further social gross investment, and social welfare?

We studied the impact of economic mechanisms of individuals using the Internet on individual or small and micro businesses’ access to finance or sustainable growth, and also social welfare. Toward this end, we first build a simple theoretical model to analyze the economic mechanisms linking access to the Internet with access to finance. The building block of our model is the principal–agent model developed by Holmstrom and Tirole [16]. One implication from our model is that access to the Internet administers to moderate the agency problem between creditors and borrowers caused by the information asymmetry between the two kinds. Our model predicts that access to the Internet will have a prominent positive effect on access to finance. Since more valuable investment projects can be taken due to the positive effect, the use of the Internet will also shrink the gap between optimal social gross investment level and the actual one. Furthermore, access to the Internet will promote the sustainable growth of individual-owned or small and micro businesses on account of the alleviation of their financial constraints. Thus, a direct consequence of such effects will be the improvement of total social welfare.

Our empirical analysis employed detailed China household finance data from China Household Finance Survey (CHFS) released by Survey and Research Center for China Household Finance (SRCCHF), Southwestern University of Finance and Economics, China. CHFS is a national wide statistical investigation, which is performed every two years starting from 2011. In 2011, SRCCHF randomly sampled 8438 households distributed in 320 districts, 80 counties, and 25 provinces to visit, obtaining the first household micro finance survey data in China. Household finance information included in CHFS is mainly comprised of housing asset, financial asset, debt, payment practices, credit constraint, demographics, employment status, etc. Because SRCCHF has not yet publicly released the second round and the third round survey data, we relied on the 2011 CHFS data to explore the impact of the use of the Internet on the alleviation of the difficulty of access to external finance of households engaging in agriculture or industry and commerce. Even though we only have the 2011 CHFS data, the data still provide an ideal opportunity to investigate how access to the Internet can make a significant difference in households’ access to external finance. Consistent with our theoretical predictions, we found that whether the head of a household running individual-owned or small and micro businesses browses or searches information or news via the Internet has a significant positive effect on its loan availability. This implies that the use of the Internet has broadened the breadth of credit availability insofar as households run small and micro businesses. This effect is still highly significant after controlling householders’ risk preference and province heterogeneity. Such effects have important implications but have not been previously documented.

Since a variety of information can be reached via the Internet, which could have all kinds of uses, focusing only on the dependent variable, the use of the Internet, may include confounding factors that could potentially contaminate the association between individuals using the Internet and access to external finance. (For instance, Internet users may pay attention to political, military, or entertaining information, which could not help to enhance loan availability).

To more throughly disentangle the relationship between access to the Internet and access to external finance, therefore, we further examined the influence of the head of a household who concerns economic information provided by the Internet on loan availability. Our evidence suggests that households whose heads follow the economic information through the Internet are more likely to be funded. In other words, after purging much noise possibly included in the explanatory variable, individuals using the Internet, our empirical finding that the Internet use has enhanced individual-owned or small and micro businesses’ credit availability still holds and is even stronger, which at least has partially addressed the financing difficulty of households or those businesses run by them.

Our study is among the first to explore the relationship between access to the Internet and access to finance and its economic mechanisms and economic results both theoretically and empirically to understand their impact on economic agent behaviors, sustainable growth of individual-owned or small and micro enterprises, and social welfare. Our paper, therefore, contributes to the growing literature on Internet finance as well as the broader literature on crowdfunding. Recent investigations include [10,11,12,14,15,17,18,19,20,21,22,23,24,25,26,27]. Given that the explosive penetration of the Internet and Mobile Internet that has laid a sound foundation in China, which has been rapidly skyrocketing (China has greater Internet and mobile Internet development among emerging market economies with penetration ratios of 45% and 37.1%, respectively, in 2013, reported by Lei [19]), our study has important and timely implications not only for academics and practitioners, but for policy makers as well. Since the financing difficulty often faced by small and micro businesses comes from the information asymmetry between fund demanders and fund suppliers, which is essentially the classic agency problem emphasized by Jensen and Meckeling [28], our study also contributes to the large body of literature on information economics or agency theory (e.g., [16,28,29,30,31,32,33,34,35,36]). Furthermore, our work also enriches a long literature on sustainable growth of small business. For instance, Chittenden et al. [37] examined the effect of capital structure of small firms on growth and access to capital markets, documenting that the over-reliance on internal finance and the importance of collateral are likely to be major constraints on their sustainable growth, while Berger and Udell [38] thoroughly reviewed the relationship between access to external finance and sustainable growth of small firms. More recent literature on this line of research includes [9,39,40,41,42,43,44,45].

The remainder of the paper is organized as follows. Section 2 briefly discusses the research background and reviews the literature related to our study. Section 4 describes the data, variables and empirical methodology. Section 5 contains the empirical results of the study and discusses the robustness of the results to alternative specifications of controls and dependent variables. Section 6 concludes this study.

2. Research Background and Literature Review

2.1. Research Background

China is an ideal context to study such impact. As the largest emerging economy and the second largest economy in the world, China plays an extremely important role in the world economy. On the one hand, as aforementioned, there are substantial small and micro businesses that act an important part in economic development of China. At the same time, however, the financing difficulty of them is universal and increasingly urgent. Based on the data from the World Bank, on the other hand, Figure 1 shows that the Internet penetration rate in China has stably increased over the past decade. Clearly, both China’s and the world’s Internet penetration rates have increased over the years. Before 2008, the Internet penetration rate of China was lower than that of the world, but since 2008 it has exceeded the world’s average level. In fact, as of 2016, the Internet penetration rate of China is 53.20%, while the corresponding rate of the world is 45.91%. Moreover, when considering the Internet penetration rate distribution across countries, according to the World Bank, Figure 2 suggests that, as of 2016, China has relatively high Internet penetration rate among the developing countries, although it is still lower than those of most developed countries, such as U.S., Germany, etc. Therefore, in China, in terms of the time series dimension, individuals using the Internet has substantially expanded over the past decade, while, in the light of cross-section dimension, individuals using the Internet also has reached relatively high level. Now, simultaneously considering the financing difficulty of small and micro businesses and the fast growth of Internet using coverage, one may conjecture that the increase in individuals using the Internet may contribute to alleviating the external financing difficulty of small and micro businesses. This is the key issue investigated in this study.

2.2. Related Literature

There has been an extensive discussion on how factors, such as tax policy, government subsidy, intellectual property protection, and various legal processes, affect entrepreneurial activity or firm growth. However, the most frequently cited obstacle to the sustainable growth of small firms is the large amount of financing needed to support their healthy development. A vast amount of literature has documented that the lack of ability to access necessary external finance is one of the crucial unfavorable factors that negatively affect innovation and sustainable growth of small businesses [17,38,46,47]. For instance, Nowak et al. [17] found that, to mitigate financing difficulty, small business loan descriptions can be used as a tool to signal borrowers’ worthiness, which can predict the likelihood that the loan will be funded, and investors can make investment decisions based on proper and relevant signals given by the small business borrowers through the loan description. Cosh et al. [47] considered what affects rejection rates for financing from different types of investors, including private individuals. Based the data from the United Kingdom, they showed that it is small businesses that are most likely to obtain financing from private individuals.

With the advent of Internet finance era, individual borrowers or small firms may access external funding via online financing markets. The ability of online markets to efficiently bring together buyers and sellers has transformed the paradigms of businesses, and refined the roles of traditional intermediaries. Crowdfunding and peer-to-peer lending are in the ascendant. A variety of literature has made efforts to delve into the forms of crowdfunding, such as an equity purchase, loan, or pre-order on a product [48,49,50,51,52], or the economic mechanisms of peer-to-peer lending [15,53,54,55]. For example, Wei and Lin [15] both theoretically and empirically investigated how the supply and demand of funds in online peer-to-peer lending markets are matched, and the prices at which transactions will occur. They concluded that, under platform-mandated posted prices, loans are funded with high probability, but at the same time with relatively high preset interest rates. Einav et al. [55] showed that the choice between auctions and posted prices in online markets is essentially a trade-off between competitive discovery and convenience. Moreover, Gong [20] systematically analyzed the current situation, future prospect, and possible economic mechanisms of Internet finance, and the potential links between the use of the Internet and credit availability in China.

The vast majority of studies on the growth of small and micro firms and their access to external finance suggest that the bottle neck that impedes the healthy progress of small businesses is the lack of ability to obtain necessary external finance to satisfy their regular development and sustainable growth [7,8,9,38,39,40,41,42,43,44,45]. However, we may question this issue further: Why is this the case? The short answer may be summarized as information asymmetry. It is the information asymmetry or agency problem between borrowers and lenders that engenders the financing difficulty of most small businesses and further curbs their healthy growth. Information asymmetry in the process of financing constrains the development and growth of a large spectrum of firms as well as the development and growth of small firms. For small firms, nevertheless, this issue is more eye-catching and more prominent, which may partially come from their limited financing sources. Starting from Jensen and Meckeling [28], a growing body of literature [16,29,30,31,32,33,34,35,36] has proposed a variety of theoretical framework and applied various empirical models to understand financing difficulty of firms in depth and to estimate the important effect of information asymmetry on firm growth. Feltham and Hofmann [31] proposed a principal–agent model in a multi-agent/multi-task context; as shown by most principal–agent models, information asymmetry cannot be perfectly addressed in their model, but second-best solution can be reached through proper incentive mechanism designs as proved in their paper. Holmstrom and Tirole [16] developed an incentive model of financial intermediation in which firms as well as intermediaries are capital constrained. They show that all forms of capital tightening (capital tightening forms include a credit crunch, a collateral squeeze, or a savings squeeze) hit poorly capitalized firms the hardest.

Although extant literature has examined firm financing problem, Internet finance, or information asymmetry in the process of external financing met by firms from different dimensions, to our best knowledge, there is little literature to directly investigate whether or not access to the Internet itself helps to alleviate the information asymmetry between potential borrowers and lenders, and furthermore to mitigate financial constraints of small firms. Allowing for the explosive penetration of the Internet (as of 2016, the Internet penetration rate of China is 53.20%, while the corresponding rate of the world is 45.91%, as reported by World Bank)and flourishing development of Internet finance in China and the world, this study attempted to fill the gap. In this study, we proposed a simple principal–agent model built on Holmstrom and Tirole [16], developed testable implications, and empirically tested the model’s predictions.

3. The Model

Our model builds on the previous literature on information asymmetry between borrowers and lenders in capital markets. In particular, the paper most closely related to ours is by Holmstrom and Tirole [16], who employed the same basic moral hazard model as we do to analyze how the firm’s balance sheet strength determines its choice between direct and indirect financing. Our model has three types of agents: firms, intermediaries, and investors. There are two periods. In the first period, financial contracts are signed, and investment decisions made. In the second period, investment returns are realized, and claims are settled. All parties are assumed to be risk neutral and protected by limited liability so that no one can end up with a negative cash position.

3.1. The Real Sector

There is a continuum of small firms. All small firms have access to the same technology; the only difference among them is that they start out with different amounts of capital A. For simplicity, we assume that all initial capital is cash. Conversely, any type of asset could be pledged as collateral with first-period market value A. The distribution of assets across firms is delineated by the cumulative distribution function $F\left(A\right)$, measuring the fraction of firms with assets less than A. The aggregate amount of firm capital is $K_f={\int }_0^{\infty }AdF\left(A\right)$.

Assume that each firm has one economically viable project or idea. It costs $I>0$ (in Period 1) to undertake a project. If $I>A$, a firm needs at least $I-A$ in external funds to be able to invest. In Period 2, the investment generates a verifiable, financial return equaling either 0 (failure) or R (success).

Firms are run by entrepreneurs, who may have incentives to reduce the probability of success in order to enjoy a private benefit if there are no proper incentives or outside monitoring in place. This is a typical moral hazard problem. We follow Holmstrom and Tirole [16], formalizing this problem by assuming that the entrepreneur can privately choose between three versions of the project as described in

Table 1. This table describes three versions of the project that can be privately chosen by the entrepreneur.

Project	Good	Bad (Low Private Benefit)	Bad (High Private Benefit)
Private benefit	0	$b\left(z\right)$	$B\left(z\right)$
Probability of success	$p_H$	$p_L$	$p_L$

.

The probability of success is denoted by p. The project is subject to moral hazard. The entrepreneur (also called “borrower”, or “she”) can exert hidden effort to influence the probability of success. The entrepreneur can “behave” (“exert effort” or “take no private benefit”), which yields a high probability of success $p=p_H$, or “shirk" (“misbehave” or “take a private benefit”), which yields a lower probability of success $p=p_L<p_H$. The entrepreneur ceteris paribus prefers to shirk, because shirking can generate a private benefit $B\left(z\right)$ or $b\left(z\right)$ to her. (Undertaking a project with no exerting effort may be easier to implement, be more fun, have greater spinoffs in the future for the entrepreneur, benefit a friend, deliver perks, be more “glamorous”, etc.).

Define $\Delta p=p_H-p_L$. Assume that, in the relevant range of the rate of return on investor capital, denoted by $\tau$, only the good project is economically viable; that is,

$$p_{H}R-\tau I>0>p_{L}R-\tau I+B\left(1\right)$$

(1)

We introduce two levels of shirking (two bad projects), as did in Holmstrom and Tirole [16], to have a rich enough way of modeling monitoring (see below). Private benefits are ordered $B\left(z\right)>b\left(z\right)>0$ and may alternatively be regarded as opportunity costs from managing the project assiduously. Note that either level of shirking produces the same probability of success. This implies that the entrepreneur will prefer the high private benefit project (B-project) over the low private benefit project (b-project) irrespectively of the financial contract. We employ z to measure the degree (depth) of access to the Internet of the entrepreneur. (We do not model financial intermediaries’ or other investors’ access to the Internet for simplicity, or one may assume that financial intermediaries can use the Internet to expand their business timely. This is not a strong assumption, allowing for the reality that financial intermediaries have kept abreast of the development of Internet finance. A good case in point here is the various finance related app programs that are widely applied into life.)

Larger value of z means that the entrepreneur has access to the Internet in depth, or has high ability to make use of the Internet. Put differently, there are at least three channels through which access to the Internet may act on access to finance. First, if the entrepreneur is able to use the Internet to handle lots of things in daily life, such as collecting economic information, shopping online, making E-payment, releasing business project information, accessing online peer-to-peer loans, etc., then her personal credit record may be tracked or captured by a third party, even a bank-like financial intermediary, which in turn may be used as a basis for evaluating the possible loan demand from her. Secondly, if keeping track of entrepreneur’s personal credit record is not true and illegal by a third party, say a financial intermediary, then the above argument may not be true. However, if potential borrowers have access to online peer-to-peer platforms, they can conveniently disclose their personal information to attract lenders’ interest. On the other hand, for an online peer-to-peer lending contract to be entered into, potential lenders need to access to this sort of borrowers’ information for their credit risk evaluation. In fact, lenders do have right to access to such information on online peer-to-peer platforms, which is common both in (e.g., Paipaidai.com and Yirendai.com) and outside China (e.g., Prosper.com and Lendingclub.com). Moreover, such platforms provide borrowers’ credit record and also their credit scores to potential lenders for reference. Thirdly, lenders can also access borrowers’ information through their websites (we thank the referee for pointing out this possibility), which can further reinforce our assumption. Moreover, borrowers have incentives to show their credit quality via the Internet to have access to external finance to relax their financial constraints. Without access to the Internet, financial intermediaries or non-financial intermediary investors would not have access to such personal credit information so that it is hard or at least more costly to do due-diligence-like work in order to evaluating a potential loan demand from an entrepreneur. More importantly, more personal information disclosed on the Internet makes shirking more expensive for the entrepreneur, which in turn will make shirking less attractive. (Of course, deeply accessing the Internet, on the other hand, also implies that the entrepreneur could capture more information on potential lenders, which will benefit her in the form of expanded choice set of lenders.)

Based on the above analysis, we assume that both $B\left(z\right)$ and $b\left(z\right)$ are decreasing functions of z, while $B\left(z\right)-b\left(z\right)$ is increasing in z, where z is normalized to be in $[0,1]$ without loss of generality. $z=0$ stands for no access to the Internet, while $z=1$ means the highest degree of access to the Internet. Conversely, we may assume that both $B\left(z\right)$ and $b\left(z\right)$ are sufficiently smooth and well-behaved in the sense that they have continuous first order derivatives $B^{\prime }\left(z\right)<0$ and $b^{\prime }\left(z\right)<0$ with $B^{\prime }\left(z\right)-b^{\prime }\left(z\right)>0$. (This assumption is not artificially imposed. We define $g_B\left(z\right)=-B\left(z\right)$ and $g_b\left(z\right)=-b\left(z\right)$, and then regard $g_i(·),i=B,b$ as a revenue function of access to the Internet, z. We may assume that $B^{\prime }\left(0\right)\ge b^{\prime }\left(0\right)$. Now, the assumption $B^{\prime }\left(z\right)-b^{\prime }\left(z\right)>0$ is equivalent to $g_B^{\prime }\left(z\right)-g_b^{\prime }\left(z\right)<0$. According to the law of diminishing marginal revenue, this implies that the rate of decrease of $g_B^{\prime }\left(z\right)$ with z is faster than that of $g_b^{\prime }\left(z\right)$. This is natural, because there is a big difference between $B\left(z\right)$ and $b\left(z\right)$, or between $g_b\left(z\right)$ and $g_B\left(z\right)$.)

3.2. The Financial Sector

The financial sector consists of lots of intermediaries (also called “lenders”, “financiers”, or “creditors”). (Here, the concept of intermediaries is quite general. Intermediaries not only include banks, but also trust companies, pawn shops, micro-credit companies, etc.)

The function of intermediaries is to monitor firms and thereby mitigate the moral hazard problem. Monitoring may take many forms, such as inspection of a business’s potential cash flow, its balance sheet position, its management, its firm-level governance, and so on. More often than not, monitoring merely amounts to verifying that the firm abides by covenants of the loan contract, such as a minimum solvency ratio or a minimum cash balance. In terms of loan contract, covenants are particularly common and extensive. The purpose of covenants is to reduce the firm’s opportunity cost of being sedulous. Based on this line of reasoning, we assume that the monitor can prevent a firm from undertaking the B-project, which contracts the firm’s opportunity cost of being diligent from $B\left(z\right)$ to $b\left(z\right)$.

Monitoring is not without toll. One important element in our model is the assumption that monitoring is privately costly; the intermediary will have to pay a non-verifiable amount $c\left(z\right)>0$ to kill the B-project, where we also assume that $c^{\prime }\left(z\right)<0$, or monitoring cost is a decreasing function of the degree (depth) of access to the Internet of the entrepreneur, z. Videlicet, more relevant record of a potential borrower that can be tracked via the Internet makes monitoring more easier and thereby less costly. This setup implies that intermediaries also face a potential moral hazard problem. Although we assume that each intermediary has sufficient physical capacity to monitor an arbitrary number of potential borrowers, the moral hazard problem exerts a upper bound on the actual amount of monitoring that will take place. Moral hazard forces intermediaries to inject some of their own capital into the firms that they monitor, which makes the aggregate amount of intermediary (or “informed”) capital $K_m$ one of the important constraints on aggregate investment.

Following Holmstrom and Tirole [16], we assume that all projects financed by an intermediary are perfectly correlated and that the capital of each intermediary is large enough relative to the scale of a project, which is particularly appropriate as far as small firms being concerned. Therefore, the exact distribution of assets among intermediaries is irrelevant. In practice, projects may be correlated since intermediaries have an incentive to choose them so, or alternatively because monitoring requires specialized expertise in a given market or industry. Perfect correlation assumption is nevertheless manifestly unrealistic. The reason that we may still insist on this assumption lies in that, without some extent of correlation, intermediaries would not have an incentive to put up any capital for monitoring (see [56]).

3.3. Individual Investors

Individual investors (uninformed investors) are small. We differentiate them from intermediaries on the basis of who monitor the firms that they invest in. Uninformed investors demand an expected rate of return $\tau$. We assume that the open market (the market in which everybody can earn the same competitive rate of return) is perfectly competitive. When considering a partial equilibrium, $\tau$ is exogenous. The aggregate amount of uninformed capital invested in firms is determined by a standard, increasing supply function $S\left(\tau \right)$ when we study a general equilibrium. The determination of the equilibrium rate of return on intermediary capital, $\iota$, will be delineated when solving the model.

We assume that firms cannot monitor each other, which is rather realistic in terms of small and micro firms. One plausible reason is that each individual small firm neither has sufficient capital nor informational expertise to be a credible monitor. Thus, firms with excess capital will have to invest their surplus cash in the open market, earning the uninformed rate of return, $\tau$.

3.4. A World without Financial Intermediation

We first consider the possibility of financing a project in a world without intermediation. Imagine a small firm that can have access to external finance only from individual investors, Now, turn to a loan contract. A contract first stipulates whether the project is financed. If so, it further specifies how the profit is split between the financiers and the borrower. Suppose that the borrower receives $R_b^S$ in the case of success and $R_b^F$ in the case of failure. This is equivalent to stipulating that the lenders receives $R_u^S=R-R_b^S$ if the project succeeds, and $R_u^F=-R_b^F$ if it fails. The borrower’s limited liability implies that both $R_b^S$ and $R_b^F$ are nonnegative. Thus, the contract space is defined by

$$\mathcal{C}=\{(R_b^S,R_b^F):(R_b^S,R_b^F)\in {\mathbf{R}}_{+}^{\mathbf{2}}\}$$

(2)

where ${\mathbf{R}}_{+}^{\mathbf{2}}$ denotes the nonnegative two-dimensional Euclidean space, and both $R_b^S$ and $R_b^F$ are observable and verifiable by a third party such as a benevolent court of law. The timing of the contracting process is defined in Figure 3.

Given Equation (1), a necessary condition for the borrower to obtain external finance is that she behaves. The borrower will behave if the following “incentive compatibility constraint (IC)” is satisfied:

$$p_H{R}_b^S+(1-p_H)R_b^F\ge p_L{R}_b^S+(1-p_L)R_b^F+B\left(z\right)$$

(3)

On the other hand, the individual lenders must be compensated by at least the amount of expected reward they can earn in the open market by investing $I-A$. This will be true if the following “individual rationality constraint (IR)” or “participation constraint” is satisfied:

$$p_H{R}_u^S+(1-p_H)R_u^F\ge \tau (I-A)$$

(4)

Now, a firm with initial net worth (assets) $A_0$ will choose its own capital contribution A, and the loan contract menu $(R_b^S,R_b^F)$ to solve Program $\mathbb{P}1$:

$$\mathbb{P}1:\underset{A,R_b^S,R_b^F}{max}{p}_H{R}_b^S+(1-p_H)R_b^F+\tau (A_0-A)-\tau A$$

subject to Equations (3) and (4),

$$A\le A_0,$$

(5)

$$R_b^S\ge 0,\mathrm{and}$$

(6)

$$R_b^F\ge 0$$

(7)

To simplify Program $\mathbb{P}1$, we first introduce the following result, essentially following [57]. See Appendix A for a formal proof.

Lemma 1.

The entrepreneur’s risk neutrality implies that the only relevant loan contract menu is to reward her only if the project succeeds. Namely, the relevant contract menu is $(R_b^S,R_b^F)=(R_b^S,0)$.

By Lemma 1, the IC constraint in Equation (3) can be simplified as

$$p_H{R}_b^S\ge p_L{R}_b^S+B\left(z\right) \mathrm{or} \Delta p{R}_b^S\ge B\left(z\right)$$

(8)

This simplified IC constraint implies that the maximum payoff that can be pledged to the lenders without violating the entrepreneur’s incentives, provided that the project succeeds, is

$$R-\frac{B\left(z\right)}{\Delta p}.$$

The (expected) pledgeable payoff to the lenders is defined as

$$\mathcal{P}\left(z\right)=p_H\left(R-\frac{B\left(z\right)}{\Delta p}\right).$$

Lemma 1 also suggests that the IR constraint in Equation (4) can be rewritten as

$$p_H(R-R_b^S)\ge \tau (I-A).$$

Define

$$\overline{A}(\tau ,z)=\frac{1}{\tau }\left[p_H\frac{B\left(z\right)}{\Delta p}-(p_{H}R-\tau I)\right]$$

(9)

This means that only borrowers with $A\ge \overline{A}(\tau ,z)$ can be financed by individual investors. We exclude the uninteresting case that $\overline{A}(\tau ,z)\le 0$ by assuming that

$$p_H\frac{B\left(z\right)}{\Delta p}<p_{H}R-\tau I$$

(10)

The condition in Equation (10) simply states that the net present value (expected profit) or the total surplus from a project is smaller than the minimum expected rent that must be left to the entrepreneur for her with appropriate incentive to be diligent. To have access to external finance, thus, the expected profit must be redistributed between the two parties. Allowing for the limited liability, the only manner that a borrower can redistribute some of the surplus to the lenders is by staking her own assets. One point that one may pay attention to here is that, if a borrower’s net worth $A<\overline{A}(\tau ,z)$, her project is not funded, although it has positive expected profit. Without strong balance sheet or sufficient cash, the borrower must finance a large amount and so pledge a large proportion of the profit under the circumstance of success. The borrower then share only a small proportion of the project gain and thus could not be well behaved after entering into a loan contract. There is no feasible loan contract that both induces the borrower’s behaving and the lenders’ breaking even. This is so-called credit rationing. Conversely, if $A\ge \overline{A}(\tau ,z)$, then the borrower can guarantee that the project can be financed. Therefore, the necessary and sufficient condition for financing is

$$A\ge \overline{A}(\tau ,z)=\frac{1}{\tau }\left[p_H\frac{B\left(z\right)}{\Delta p}-(p_{H}R-\tau I)\right]$$

(11)

Now, consider the IR constraint $p_H(R-R_b^S)\ge \tau (I-A)$. We claim that the constraint must be binding at the optimum of $\mathbb{P}1$. Suppose first that $p_H(R-R_b^S)>\tau (I-A)$. Then, take $\epsilon$ with $0<\epsilon <p_H(R-R_b^S)-\tau (I-A)$. Then, the borrower can increase $R_b^S$ by $\epsilon$ without violating any constraint in $\mathbb{P}1$. Therefore, at the optimum, the IR inequality cannot be strict.

The optimal choice of A invested in the project by the borrower is also all her wealth $A_0$. This is because investing less than her endowment $A_0$ cannot improve her welfare. Suppose that the entrepreneur invests only the amount $a<A_0$ in the project. Then, if $a\ge \overline{A}(\tau ,z)$, she is able to access to external finance, and then she can still capture all the expected profit of the project. (Note that we have already shown that, at the optimum, the lenders can only break even, and thus the borrower can extract all social surplus.)

On the other hand, if $a<\overline{A}(\tau ,z)$, then the project can not be funded since the lenders cannot break even but would get a loss if financing the project. Put differently, in this case, it becomes more difficult for the project to be funded. We are done then.

In the light of the above analysis, we have proven the following lemma. (For a similar analysis, refer to [57]).

Lemma 2.

The lenders’ participation constraint must be binding at the optimum of $\mathbb{P}1$, while the optimal investment level of the borrower is to stake all her own capital in the project.

Based on Lemmas 1 and 2, as expected, the entrepreneur obtains the entire expected profit or social surplus generated by the project if it is undertaken. Therefore, the solution of Program $\mathbb{P}1$ has the following characteristics, as summarized in Proposition 1. See Appendix A for a formal proof.

Proposition 1.

When all agents are risk-neutral, at the optimum of Program $\mathbb{P}1$, the borrower receives the entire social surplus generated by the project if it is financed. The optimal menu of the loan contract entails:

• The stake shared by the borrower in the case of success is

  $$R_b^S=R-\frac{\tau (I-A_0)}{p_H},$$

  and zero otherwise.
• The borrower receives the entire expected profit of the project, $p_{H}R-\tau I$.
• The lenders’ stake in the case of success is

  $$\frac{\tau (I-A_0)}{p_H},$$

  and zero otherwise.

Thus far, we have demonstrated that there are two factors that may make a firm credit-constrained, which are lower level of firm net worth $A_0$ and high agency cost measured by combination of the private benefit $B\left(z\right)$ and the likelihood ratio $\Delta p/p_H$, given the project’s expected profit. We have shown the characteristics of solution of Program $\mathbb{P}1$, which delineates the equilibrium under the condition of the project being financed. However, we need to carefully analyze potential mechanisms that may result in credit rationing in depth. Specifically, we need to identify the possible role played by access to the Internet in the process of small and micro firm financing. A simple check suggests that the channel through which access to the Internet affects firm financing is the private benefit $B\left(z\right)$ in our model in the scenario of no financial intermediation. By the assumption, we know that $B^{\prime }\left(z\right)<0$, which implies that access to the Internet will decrease the private benefit $B\left(z\right)$. Intuitively, as aforementioned, there are at least three channels through which access to the Internet may act on access to finance. First, if the entrepreneur is able to use the Internet to handle lots of things in daily life, such as collecting economic information, shopping online, making E-payment, releasing business project information, even accessing to online peer-to-peer loans, etc., then her personal credit record may be tracked or captured by a third party, even a bank-like financial intermediary, which in turn may be used as a basis for evaluating the possible loan demand from her. Secondly, if keeping track of entrepreneur’s personal credit record is not true and illegal by a third party, say a financial intermediary, then the above argument may not be true. However, if potential borrowers have access to online peer-to-peer platforms, they can conveniently disclose their personal information to attract lenders’ interest. On the other hand, for an online peer-to-peer lending contract to be entered into, potential lenders need to access to this sort of borrowers’ information for their credit risk evaluation. In fact, lenders do have right to access to such information on online peer-to-peer platforms, which is common both in (e.g., Paipaidai.com and Yirendai.com) and outside China (e.g., Prosper.com and Lendingclub.com). Moreover, such platforms provide borrowers’ credit record and also their credit scores to potential lenders for reference. Thirdly, lenders can also access borrowers’ information through their websites, which can further reinforce our assumption. Moreover, borrowers have incentives to show their credit quality via the Internet to have access to external finance to relax their financial constraints. This kind of information in turn may be used as a basis for evaluating the possible loan demand from borrowers. Without access to the Internet, individual investors would not have access to such personal credit information so that it is hard or at least more costly to identify relevant credit risk. More importantly, more personal information disclosed on the Internet makes shirking more expensive for the entrepreneur, which in turn will make shirking less attractive. To further uncover the function of access to the Internet, we rearrange Equation (9) as

$$\overline{A}(\tau ,z)=\frac{1}{\tau }\left[\tau I-p_H\left(R-\frac{B\left(z\right)}{\Delta p}\right)\right]=\frac{1}{\tau }[\tau I-\mathcal{P}\left(z\right)]$$

(12)

By Equation (12), $\overline{A}(\tau ,z)$ is decreasing in z since $\partial \overline{A}(\tau ,z)/\partial z=-\mathcal{P}{\left(z\right)}^{\prime }/\tau =p_H{B}^{\prime }\left(z\right)/\tau \Delta p<0$. This implies that access to the Internet can lower the net worth threshold at or above which the entrepreneur can have access to external finance through raising the pledgeable income $\mathcal{P}\left(z\right)$ via reducing the private benefit $B\left(z\right)$. Because only firms with net worth at or above the threshold $\overline{A}(\tau ,z)$ can be funded, access to the Internet can enhance the likelihood of a project being financed. Precisely, since the probability that a project can be financed can be formulated as $Prob(A_0\ge \overline{A}(\tau ,z))=1-F\left(\overline{A}(\tau ,z)\right)$, taking the partial derivative of the probability with respect to z yields that $\partial Prob(A_0\ge \overline{A}(\tau ,z))/\partial z=-(\partial F/\partial \overline{A})(\partial \overline{A}(\tau ,z)/\partial z)>0$, where the last inequality comes from $\overline{A}(\tau ,z)/\partial z)<0$ and $\partial F/\partial \overline{A}>0$. Now, we have shown one of our main results, as summarized in Proposition 2.

Proposition 2.

The entrepreneur’s access to the Internet can raise the probability ($1-F\left(\overline{A}(\tau ,z)\right)$) that the project is funded. The deeper the degree of the entrepreneur’s access to the Internet, the higher the probability. Therefore, access to the Internet leads to more projects with positive expected profit to be undertaken, which improves investment efficiency and social welfare.

Even if in this very simple partial equilibrium framework, we found that access to the Internet has a positive effect on firms’ access to external finance. The main economic mechanism is that a potential borrower’s shirking incentive can be reduced via access to the Internet. Now, we switch to a more complicated world by introducing one more economic agent, financial intermediation.

3.5. A World with Financial Intermediation

In a world with financial intermediation, an intermediary can help alleviate financial constraint of a firm by monitoring. Monitoring debases the entrepreneur’s opportunity of shirking by eliminating the B-project with high agency cost (high private benefit), which contributes to attracting more investors to come in. Now, borrowers face two types of investors, individual (uniformed) investors and financial intermediaries. Therefore, in this case, there are three parties to the loan contract: the entrepreneur, the intermediary, and the individual investors.

Based on the discussion in Section 3.4, an optimal three-party contract can be characterized by a form analogous to the one analyzed earlier: everyone gets nothing if the project fails; and the total return from the project is divided up so that

$$R_b^S+R_u^S+R_m^S=R,$$

if the project succeeds, where the newly added notation, $R_m^S$, stands for the intermediary’s share, while the remaining notations are defined as before.

Now, under the circumstance of monitoring, the entrepreneur is left to choose between the good project and the b-project with lower private benefit since the B-project is eliminated. The entrepreneur’s incentive-compatible constraint (IC) is changed to

$$R_b^S\ge b\left(z\right)/\Delta p$$

(13)

Since $b\left(z\right)<B\left(z\right)$, firms that can not be funded due to too high agency cost measured by $B\left(z\right)$ now may be financed on account of reduced agency cost. We pay more attention to the firms with $R_b^S<B\left(z\right)/\Delta p$, for otherwise the entrepreneur would well behave without monitoring. For the intermediary to monitor, allowing for the monitoring cost, there is one more IC constraint that must be satisfied,

$$R_m^S\ge c\left(z\right)/\Delta p$$

(14)

The two IC constraints in Equations (13) and (14) give the minimum shares that need to be distributed to the entrepreneur and the intermediary, respectively. The pledgeable (expected) income is defined as the maximum expected income that can be promised to the individual investors as before, which is

$$\mathcal{P}\left(z\right)=p_H\left[R-\frac{b\left(z\right)+c\left(z\right)}{\Delta p}\right]$$

(15)

Denote by $I_m$ the amount of capital that an intermediary invests in a firm that it monitors. The rate of return on intermediary capital is then $\iota =p_H{R}_m^S/I_m$. We know $\iota$ must exceed $\tau$, taking the monitoring cost into account. This implies that firms prefer individual capital to intermediary capital whenever the former is available. Given Equation (14), the IC constraint of the intermediary requires that it be paid at least $R_m^S=c\left(z\right)/\Delta p$, hence it will invest at least

$$I_m\left(\iota \right)=\frac{p_{H}c\left(z\right)}{\Delta p\iota }$$

(16)

in each project that it monitors. On the other hand, all firms that are monitored must demand exactly this level of intermediary capital since it is more expensive than the capital from individual investors. In fact, the motivation of being a monitor for an intermediary comes from the return $R_m$. The required investment $I_m\left(\tau \right)$ stipulates the rate of return on the intermediary capital so that the market for that class of capital clears.

For a project to be undertaken, individual investors must provide the balance $I_u=I-A-I_m\left(\tau \right)$ whenever there exists a positive gap between I and $A+I_m\left(\iota \right)$. Similar to the reasoning in Section 3.4, a necessary and sufficient condition for a firm to be funded thus is

$$\tau [I-A-I_m\left(\iota \right)]\le p_H\left[R-\frac{b\left(z\right)+c\left(z\right)}{\Delta p}\right]$$

Rearranging terms yields the following equivalent condition

$$A\ge I-I_m\left(\iota \right)-\frac{p_H}{\tau }\left[R-\frac{b\left(z\right)+c\left(z\right)}{\Delta p}\right]=\underline{A}(\tau ,\iota ,z)$$

(17)

To make the story interesting, we assume that $c\left(z\right)\Delta p<p_H(B\left(z\right)-b\left(z\right))$ for all $z\in [0,1]$ (this assumption can be satisfied for a sufficiently small c(0).) which guarantees that $\overline{A}(\tau ,z)>\underline{A}(\tau ,\underline{\iota },z)$ for a lower limit of $\iota$ that will be defined later on and for all possible z. Note that a firm with net worth $A<\underline{A}(\tau ,\iota ,z)$ cannot be financed. It is easy to see that the minimum acceptable rate of return $\underline{\iota }$ (if there were a upper limit of $\iota$ set by a regulation authority, $\overline{\iota }$ with $\overline{\iota }>\underline{\iota }$, then the intermediary market would shrink to be zero, or no intermediary capital would be supplied in this case) is pinned down by

$$\frac{p_{H}c\left(z\right)}{\Delta p}-c\left(z\right)=\tau I_m\left(\iota \right)=\tau \frac{p_{H}c\left(z\right)}{\Delta p\iota },$$

or

$$\underline{\iota }=\frac{p_H\tau }{p_L}>\tau$$

Now, in the economy, there are three types of firms according to their demand for intermediary capital. One category with sufficiently strong balance sheet (with net worth $A_0>\overline{A}(\tau ,z)$) can finance their projects without reckoning on intermediaries. At the other extreme are the firms with sufficiently weak balance sheet (with net worth $A_0<\underline{A}(\tau ,\iota ,z)$), whose projects cannot be funded even though in a world with financial intermediation. The firms with net worth $A_0\in [\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$ fall in between. These firms can access to external finance but only with the help of monitoring.

The preceding analysis shows that we have almost solved the following program for a firm falling into the moderate category with net worth $A_0\in [\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$.

$$\mathbb{P}2:\underset{A,R_b^S,R_m^S,R_u^S,I_m,I_u}{max}{p}_H{R}_b^S-p_H{R}_m^S-p_H{R}_u^S+\tau (A_0-A)-\tau A$$

subject to Equations (13), (14) and (17),

$$A\le A_0,$$

(18)

$$(R_b^S,R_m^S,R_u^S)\ge 0,$$

(19)

$$R_b^S+R_m^S+R_u^S\le R,$$

(20)

$$A+I_m+I_u\ge I,$$

(21)

$$p_H{R}_m^S\ge \iota I_m, \mathrm{and}$$

(22)

$$p_H{R}_u^S\ge \tau I_u$$

(23)

Similar to the reasoning of Proposition 1, the optimal investment level for the entrepreneur is to invest all her own capital in the project. Combing this and the preceding analysis, we obtain the following results.

Proposition 3.

When all the agents are risk-neutral, at the optimum of Program $\mathbb{P}2$, the borrower receives the entire social surplus generated by the project if it is financed. The solution of Program $\mathbb{P}2$ is characterized by

• The stakes shared by the borrower, the intermediary, and the individual investors in the case of success are

  $$R_b^S=R-\frac{c\left(z\right)}{\Delta p}-\frac{\tau [I-A_0-p_{H}c\left(z\right)/\Delta p\iota ]}{p_H},R_m^S=\frac{c\left(z\right)}{\Delta p}, \mathrm{and} R_u^S=\frac{\tau [I-A_0-p_{H}c\left(z\right)/\Delta p\iota ]}{p_H},$$

  respectively, and zero otherwise.
• The optimal investment for the borrower, the intermediary, and the individual investors are

  $$A=A_0,I_m=\frac{p_{H}c\left(z\right)}{\Delta p\iota }, \mathrm{and} I_u=I-A_0-\frac{p_{H}c\left(z\right)}{\Delta p\iota },$$

  respectively.

Proposition 3 describes the optimal loan contract menu for the firms with net worth $A_0\in [\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$. What we care more about is the possible effect of the degree of access to the Internet z on the optimal contract. First, the degree of access to the Internet has a negative effect on the demand for intermediary capital since $\partial I_m/\partial z=p_H{c}^{\prime }\left(z\right)/\Delta p\iota <0$. This reduces the cost of capital of the firm provided that the project can be funded, because intermediary capital is more expensive than individual investor capital. The channel through which the degree of access to the Internet z has a negative impact on the demand for intermediary capital is that z first reduces the cost of monitoring, which in turn lowers the agency rent extracted by the intermediary, and the demand for intermediary fund is decreased then.

The other side of the same coin is that the degree of access to the Internet z has a positive effect on the demand for individual or uninformed capital (it is easy to see that $\partial I_u/\partial z=-p_H{c}^{\prime }\left(z\right)/\Delta p\iota >0$), because the total investment of the project is given by a constant number I and the optimal investment level of the borrower is to invest all her own capital (noting that there is no correlation between the borrower’s own investment A and the degree of access to the Internet z). This benefits the firm since individual investors require a lower rate of return than the intermediary. More precisely, individual investors require no agency rent. This leads the external capital with lowest cost provided by them to the firm.

As far as the profit shared by the three parties are concerned, in the case of success, since z plays a role in reducing monitoring cost and thus the intermediary’s agency rent, the stake owned by the intermediary must decrease as z increases ($\partial R_m^S/\partial z=c^{\prime }\left(z\right)/\Delta p<0$). The preceding analysis shows that the degree of access to the Internet has a negative effect on the demand for intermediary capital and a positive effect on the demand for individual or uninformed capital, given the total investment I, hence the stake shared by individual investors must be increased as z increases ($\partial R_u^S/\partial z=-\tau c^{\prime }\left(z\right)/\Delta p\iota >0$). Whether the borrower’s stake increases in z is not so manifest. Taking the partial derivative of $R_b^S$ with respect to z at the optimum gives

$$\frac{\partial R_b^S}{\partial z}=\frac{c^{\prime }\left(z\right)}{\Delta p}\left(\frac{\tau }{\iota }-1\right)>0$$

since $c^{\prime }\left(z\right)<0$ and $\tau <\iota$. (Since the intermediary has a higher opportunity cost that individual investors, allowing for monitoring cost, $\tau$ must be less than $\iota$.)

This shows that the borrower’s payoff will be enhanced as z increases. Intuitively, access to the Internet lowers the demand for the intermediary capital and meanwhile raises the demand for individual investor capital. The former has a relatively high cost but the latter has a relatively low cost, which implies that the borrower must benefit from access to the Internet in some way. However, the only way in which the borrower’s welfare can be improved is to share more profit. This is the potential mechanism through which the borrower’s welfare could be bettered.

Finally, we try to explore whether or not access to the Internet has an effect on the whole social welfare. The total welfare amounts to the project’s expected profit, expressed as follows.

$$\pi =p_{H}R-\tau (A_0+I_u)-\iota I_m$$

Taking the first order partial derivative of $\pi$ with respect to z at the optimum yields

$$\frac{\partial \pi }{\partial z}=-\tau \frac{\partial I_u}{\partial z}-\iota \frac{\partial I_m}{\partial z}=-\frac{\partial I_m}{\partial z}(\iota -\tau )>0$$

by Proposition 3 and the fact that the cost of intermediary capital is higher than that of individual investor capital. (Note that we have already known $\partial I_u/\partial z=-\partial I_m/\partial z$ at the optimum according to our previous derivation. We also make use of this fact here.) Therefore, the total social welfare increases in the degree of access to the Internet z. It is not hard to find that the economic mechanism for this is that the entrepreneur with deeper access to the Internet takes advantage of attracting more cheap capital (from individual investors) and demanding less expensive capital (from the intermediary) so that the expected profit from the project is enhanced.

We summarize the above analysis as a corollary of Proposition 3 as follows.

Corollary 1.

When all the agents are risk-neutral, the optimal solution of Program $\mathbb{P}2$ has the following comparative static properties:

• The optimal demand for the intermediary capital $I_m$ decreases as the degree of access to the Internet of the entrepreneur, z, increases, while the optimal demand for individual investors’ capital $I_u$ increases in the the degree of access to the Internet of the entrepreneur, z.
• The optimal stake of the intermediary in the profit, $R_m^S$, becomes lower when the degree of access to the Internet of the entrepreneur is deeper, and the optimal stake of individual investors in the profit, $R_u^S$, turns out to be higher when the degree of access to the Internet of the entrepreneur is deeper.
• Finally, the optimal stake of the entrepreneur in the profit, $R_b^S$, becomes larger when she has deeper access to the Internet. Furthermore, the expected profit on the project or the total social welfare, π, improves as the degree of access to the Internet of the entrepreneur goes deeper.

Our analysis in this subsection so far has exclusively focused on the firms falling into the moderate category with net worth $A_0\in [\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$. However, another important question is whether access to the Internet can raise the probability that a project is funded, which has not been investigated. We first consider the effect of access to the Internet on the threshold ($\underline{A}(\tau ,\iota ,z)$) of net worth at or above which a project can be funded. Simple calculation shows that

$$\frac{\partial \underline{A}(\tau ,\iota ,z)}{\partial z}=\frac{p_H}{\Delta p}\left[\frac{b^{\prime }\left(z\right)}{\tau }+c^{\prime }\left(z\right)\left(\frac{1}{\tau }-\frac{1}{\iota }\right)\right]<0$$

This implies that access to the Internet can lower the threshold of net worth, and thus allows more projects to be financed. The rationale for this is that access to the Internet can reduce the entrepreneur private benefit, which in turn relaxes the minimum requirement of net worth and allow more firms to have access to external finance. In fact, access to the Internet can inflate the expected pledgeable income ($\partial \mathcal{P}\left(z\right)/\partial z=-p_H[b^{\prime }\left(z\right)+c^{\prime }\left(z\right)]/\Delta p>0$), and therefore boost the ability to borrow then. That borrowing capacity becomes strong directly means that the probability ($1-F\left(\underline{A}(\tau ,\iota ,z)\right)$) that a project is funded goes up. (Mathematically, it is easy to see $\partial [1-F\left(\underline{A}(\tau ,\iota ,z)\right)]/\partial z=-(\partial F/\partial \underline{A})(\partial \underline{A}/\partial z)>0$.)

Now, we have shown the following results.

Proposition 4.

In a world with financial intermediation, access to the Internet can boost the ability of the entrepreneur to borrow by reducing the agency rents ($c\left(z\right)$ and $b\left(z\right)$) of the intermediary, lowering the minimum requirement of net worth, and raising the expected peldgeable profit. Therefore, The entrepreneur’s access to the Internet can heighten the probability ($1-F\left(\underline{A}(\tau ,\iota ,z)\right)$) that her project is funded. Since more projects can be undertaken because of access to the Internet, both the investment efficiency and social welfare are improved.

Another interesting question we may investigate is whether there exists a certain effect of access to the Internet on the distance between the two thresholds $\underline{A}(\tau ,\iota ,z)$ and $\overline{A}(\tau ,z)$. Let $d\left(z\right)$ denote the distance. Taking the derivative of $d\left(z\right)$ with respect to z yields

$$d^{\prime }\left(z\right)=\frac{p_H}{\Delta p}\left[\frac{B^{\prime }\left(z\right)-b^{\prime }\left(z\right)}{\tau }-c^{\prime }\left(z\right)\left(\frac{1}{\tau }-\frac{1}{\iota }\right)\right]>0$$

where the last inequality comes from our assumptions that $B^{\prime }\left(z\right)-b^{\prime }\left(z\right)>0$ and $c^{\prime }\left(z\right)<0$ and the fact that $\tau <\iota$. This result implies that access to the Internet acts on both of the thresholds, but the force on the threshold $\underline{A}(\tau ,\iota ,z)$ dominates the one on the threshold $\overline{A}(\tau ,z)$. Therefore, in a world with financial intermediation, access to the Internet widens the distance between the two thresholds, and further incorporates more firms into the middle class of firms in the sense that there are more firms whose net worth falls into the interval $[\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$. This results in alleviation of financing difficulty of firms. Formally, we formulate this line of reasoning as follows.

Proposition 5.

In a world with financial intermediation, under the assumptions that $B^{\prime }\left(z\right)-b^{\prime }\left(z\right)>0$ and $c^{\prime }\left(z\right)<0$, access to the Internet enlarges the set of firms whose net worth falls into the interval $[\underline{A}(\tau ,\iota ,z),\overline{A}(\tau ,z))$ by prolonging the length of the interval (both ends of the interval are reduced, while the left one is reduced more). This alleviates financing difficulty of firms.

3.6. Equilibrium in the Credit Market

Similar to Holmstrom and Tirole [16], since each firm demands the minimum amount of intermediary capital $I_m\left(\iota \right)$, the aggregate demand for intermediary capital is $D_m(\tau ,\iota ,z)=[F\left(\overline{A}(\tau ,z)\right)-F\left(\underline{A}(\tau ,\iota ,z)\right)]I_m\left(\iota \right)$. We only consider the case in which there is no excess supply of intermediary capital at the minimum acceptable rate of return $\underline{\iota }$, the equilibrium $\iota$ can obtained by making the monitoring market clear.

$$K_m=D_m(\tau ,\iota ,z)=[F\left(\overline{A}(\tau ,z)\right)-F\left(\underline{A}(\tau ,\iota ,z)\right)]I_m(\iota ,z)$$

(24)

What is interesting in Equation (24) is the influence of the degree of access to the Internet, z, on the equilibrium demand for intermediary capital. By Corollary 1, $I_m(\iota ,z)$ is a decreasing function of z, which inclines to lower the demand for intermediary capital. On the other hand, by Proposition 5, the distance between the two thresholds, $d\left(z\right)$, is increasing in z, which allows more firms to be successfully financed and thus is a force to potentially increase the demand for intermediary capital. Without knowing the distribution of A in the economy, the effect of z on the demand for intermediary capital is ambiguous ex ante. If we further assume that A follows uniform distribution, then we can reach a clear conclusion about the effect of access to the Internet upon the demand for intermediary capital. Specifically, assume $A\sim U(0,W)$ for some upper bound W. Rewrite the demand for intermediary capital $D_m(\tau ,\iota ,z)$ as

$$D_m(\tau ,\iota ,z)=[\overline{A}(\tau ,z)-\underline{A}(\tau ,\iota ,z)]I_m(\iota ,z)$$

(25)

We claim that access to the Internet pushes up the demand for the intermediary capital. See Appendix A for a formal proof.

Proposition 6.

Assume that firm net worth A follows uniform distribution $U(0,W)$ for some large enough positive number W. Then, the degree of access to the Internet, z, helps push up the aggregate demand for intermediary capital $D_m(\tau ,\iota ,z)$ in a world with financial intermediation.

Proposition 6 states that access to the Internet plays a part in increasing in aggregate demand for intermediary capital. A natural question from this Proposition is how the increase in aggregate demand for intermediary capital affects the cost of intermediary capital, $\iota$, in equilibrium. The equilibrium condition in Equations (24) suggests that the demand $D_m(\tau ,\iota ,z)$ is decreasing in $\iota$ since $I_m(\iota ,z)$ is decreasing and $\underline{A}(\tau ,\iota ,z)$ is increasing in $\iota$. Therefore, for each pair of $(\tau ,z)$, there is a unique $\iota$ that clears the market for intermediary capital. Given the supply of intermediary capital $K_m$ and the rate of return, $\tau$, demanded by individual investors, an increase in the demand $D_m(\tau ,\iota ,z)$ must push up the equilibrium cost of intermediary capital $\iota$. Equation (24) fully describes the equilibrium if the rate of return, $\tau$, required by individual investors is exogenous. Hence, by Proposition 6, an increase in the degree of access to the Internet will force the cost of intermediary capital, $\iota$, higher. Summarizing this strand of analysis yields the following Corollary.

Corollary 2.

Given the supply of intermediary capital $K_m$, the rate of return, τ, required by individual investors, being exogenous, and the assumption that firm net worth A follows uniform distribution $U(0,W)$ for some large enough positive number W, access to the Internet makes the equilibrium cost of capital, ι, demanded by intermediaries, pinned down by Equation (24), more expensive.

Consider the case in which the cost of capital, $\tau$, required by individual investors is endogenous, or assume that the supply of individual investors’ capital $S\left(\tau \right)$ is imperfectly elastic. We must add one more equilibrium condition for individual capital then. Let

$$D_u(\tau ,\iota ,z)={\int }_{\underline{A}(\tau ,\iota ,z)}^{\overline{A}(\tau ,z)}[I-A-I_m(\iota ,z)]dF\left(A\right)+{\int }_{\overline{A}(\tau ,z)}^{\infty }[I-A]dF\left(A\right)$$

(26)

denote the demand for individual capital. (We have eliminated the capital that firms with $I_m(\iota ,z)+A>I$ or $I>A$ will invest in the market in the demand function.)

The market clearing condition for individual capital is

$$D_u(\tau ,\iota ,z)=S\left(\tau \right)$$

(27)

For each $\iota$ and z, the demand $D_u(\tau ,\iota ,z)$ is decreasing in $\tau$. To see this, noting that firms with net worth just above $\underline{A}(\tau ,\iota ,z)$ are squeezed out by an increase in $\tau$, while firms with net worth just above $\overline{A}(\tau ,z)$ move from financing only from individual investors to from both intermediaries and individual investors, which lowers the demand for individual capital, $D_u(\tau ,\iota ,z)$ must go down when $\tau$ goes up. In equilibrium for the individual capital market, there is a unique $\tau$ that solves Equation (27) for each pair of $(\iota ,z)$. It is not clear how the change in z impacts on the demand $D_u(\tau ,\iota ,z)$ given in Equations (26). We replace Equation (27) with the following condition to determine $\tau$ and $\iota$, by plugging Equation (24) into Equation (27),

$${\int }_{\underline{A}(\tau ,\iota ,z)}^{\infty }[I-A]dF\left(A\right)=S\left(\tau \right)+K_m$$

(28)

Our main interest is with the effects that changes in the degree of access to the Internet have on the equilibrium outcome. An increase in z affects $\underline{A}(\tau ,\iota ,z)$ negatively for each pair of $(\tau ,\iota )$ by Proposition 5. This directly implies an increase in aggregate investment. Since an increase in investment must be funded by either individual or intermediary capital, noting the fact that each firm demands the minimum amount of intermediary capital, S will have to go up by Equation (28) to satisfy firms’ financing demand regardless of increase or decrease in aggregate demand for intermediary capital. Therefore, given any distribution of A, the cost of individual capital will be higher for a deeper access to the Internet. If we assume as before that A follows uniform distribution, then we already know that access to the Internet also has a positive effect on the cost of intermediary capital by Corollary 2. Now, we finish this section by summarizing the above analysis as the following Proposition.

Proposition 7.

If there is an increase in the degree of access to the Internet, there will be an increase in aggregate investment and the cost of individual investor capital τ. If we further assume that firm net worth A follows uniform distribution $U(0,W)$ for some large enough positive number W. Then, the degree of access to the Internet, z, contributes to push up the aggregate demand for intermediary capital, $D_m(\tau ,\iota ,z)$.

3.7. Empirical Implications

An immediate testable implication from the model is that access to the Internet is conducive to access to external finance via reducing agency cost. This result holds regardless of whether the economy has financial intermediation or not by Propositions 2 and 4. Proposition 6 and Corollary 2 show that the equilibrium of interest rate or the cost of debt has been pushed up due to more demand from more firms that can have access to external finance because of lower agency cost led to by access to the Internet. Of course, the model also has other important empirical implications, such as the enhancement of investment efficiency due to access to the Internet, the improvement of social welfare generated by access to the Internet, etc. However, in consideration of the data availability, in this study, we mainly focus on the first empirical prediction. Formally, based on the above analysis, we formulate our empirical hypothesis as follows.

Hypothesis 1.

All else equal, access to the Internet helps alleviate small and micro businesses’ financing difficulty by raising their ability to access to the external finance.

4. Data and Empirical Design

4.1. Data

We obtained detailed China household finance data from China Household Finance Survey (CHFS) released by Survey and Research Center for China Household Finance (SRCCHF), Southwestern University of Finance and Economics, China. CHFS is a national wide statistical investigation, which is performed every two years starting from 2011. In 2011, SRCCHF randomly sampled 8438 households distributed in 320 districts, 80 counties, and 25 provinces to visit, obtaining the first household micro finance survey data in China. Household finance information included in CHFS is mainly comprised of housing asset, financial asset, debt, payment practices, credit constraint, demographics, employment status, etc. Because SRCCHF has not yet publicly released the second round and the third round survey data, in this paper we have to rely on the 2011 CHFS data to explore the impact of the use of Internet on the alleviation of the difficulty of access to external finance of households engaging in agriculture or industry and commerce. Even though we only have the 2011 CHFS data, the data still provide a good opportunity to investigate how access to the Internet can make a significant difference in households’ access to external finance.

Based on the CHFS data, we define two key variables to measure households’ access to the Internet (those households engaging in industry, commerce or agriculture). The first one is $Z_1$, which is defined as a dummy variable that equals 1 if the head of a household that runs an individual-owned firm or a small and micro business has access to the Internet, and zero otherwise. Admittedly, this variable cannot perfectly measure personal information dissemination through the Internet. However, due to data availability problem, we do not have finer information or data to measure the process of personal information dissemination through the Internet. Put differently, precisely verifying the model implication is somewhat beyond the scope of the paper because of unavailability of finer data. However, even so, we can still gauge the effect of access to the Internet in some degree based on the current data. Depending on data availability, we plan to address this issue in our future research (we thank the referee and the academic editor for pointing out this issue).

As discussed above, the access to the Internet measure $Z_1$ may contain too much noise in terms of gauging personal information dissemination through the Internet. For instance, one may concern that various information can be accessed via the Internet, which could have all kinds of uses, and also that the head of a household’s access to the Internet is not necessary to disclose or disseminate personal information to potential lenders. To mitigate such concern, we define the second variable of interest $Z_2$, which is a dummy variable that equals 1 if the head of a household that runs an individual-owned firm or a small and micro business not only has access to the Internet but also pays attention to and captures economic-related information via the Internet, and zero otherwise. This is a finer measure of access to the Internet, which can at least eliminate part of noise contained in $Z_1$. If an owner of small and micro firm focuses on economic-related information through the Internet, intuitively, she may have stronger incentive to disclose personal information through the Internet if her firm is constrained by internal finance. Therefore, this is a somewhat more convincing measure than $Z_1$, although it is still far from perfection.

To visualize the spatial distribution of the number of households with $Z_1=1$ or $Z_2=1$ across provinces, we draw two pictures as follows. Figure 4 shows that access to the Internet is relatively deep (more households have access to the Internet) for those economically developed or wealthy provinces (with darker color in the figure), such as Beijing, Shanghai, Guangdong, etc. This is consistent with our intuition and our expectation. Similarly, Figure 5 also indicates that more heads of households pay attention to and capture economic-related information via the Internet in those developed provinces. This also makes sense, because usually people living in those developed area are more sensitive to economic-related information.

We define the main variables used in the paper in

Table 2. Variable definitions.

Variable Name	Brief Definition
Loan	An indicator equal to 1 if a household has loans, and 0 otherwise.
Credit Loan	An indicator equal to 1 if a household has credit loans, and 0 otherwise.
Informal Loan	An indicator equal to 1 if a household has informal loans, and 0 otherwise.
Loan Rate	Interest rate of loan.
$Z_1$	An indicator equal to 1 if the head of a household browses or searches information via the Internet, and 0 otherwise.
$Z_2$	An indicator equal to 1 if the head of a household pays attention to and captures economic-related information via the Internet, and 0 otherwise.
Head Age	The head age of a household.
Head Edu	The schooling years of the head of a household.
Female Head	An indicator equal to 1 if the head of a household is female, and 0 otherwise.
Household Size	The number of members of a household.
Home Ownership	An indicator equal to 1 if a household has the home ownership, and 0 otherwise.
Housing Area	Housing area size in meter squared.
Rural Area	An indicator equal to 1 if a household lives in the countryside, and 0 otherwise.
East	An indicator equal to 1 if a household lives in the eastern China, and 0 otherwise.
West	An indicator equal to 1 if a household lives in the western China, and 0 otherwise.
Risk Preference	The subjective risk preference of a household.

. Then, we present the basic summary statistics in

Table 3. Summary statistics.

Variable	Observations	Mean	Std	Min	Max
Loan	2762	0.08	0.53	0	1
Credit Loan	227	0.54	0.71	0	1
Informal Loan	2763	0.26	0.66	0	1
Loan Rate	187	0.07	1.93	0	0.156
$Z_1$	6214	0.26	0.66	0	1
$Z_2$	6195	0.18	0.62	0	1
Head Age	8432	49.93	3.75	4	111
Head Edu	8354	9.34	2.06	0	21
Female Head	8433	0.27	0.66	0	1
Household Size	8433	3.48	1.24	1	18
Home Ownership	6223	0.83	0.61	0	1
Housing Area	5910	103.25	8.70	1	902
Rural Area	8433	0.38	0.70	0	1
East	8433	0.47	0.71	0	1
West	8433	0.23	0.65	0	1
Risk Preference	6112	0.14	0.59	0	1

. The simple descriptive statistics show that among the households engaging in industry, commerce or agriculture, the percentage of those with loans is $8.3\%$, and the average rate of interest on loans is $6.73\%$. About one half of them is in the form of credit loan, and $26\%$ of them are in the form of informal finance, among all the households with loans. This implies that informal finance is an important financing source of the households engaging in industry, commerce or agriculture. Table 3 and

Table 4. The distribution of penetration rate of the Internet in rural and urban areas.

Variable	Urban Area	Rural Area	Whole Sample	Observations
$Z_1$	37%	7%	26.31%	6214
$Z_2$	26%	5%	18.18%	6195

also show that in our sample there are around $26\%$ of heads of households who capture information or news via the Internet, while there are about $18\%$ of heads of households who pay attention to and capture economic-related information via the Internet. Furthermore, we found that there is a big difference in the use of the Internet between urban and rural areas in China. As of 2011, there are about $7.2\%$ of heads of households who capture information or news via the Internet in rural area, while at the same time there are only around $5.1\%$ of those concerning and capturing economic-related information via the Internet. In contrast, in urban area, the proportion of heads of households capturing information or news via the Internet is $37.4\%$, and the number of heads of households paying attention to and capturing economic-related information via the Internet accounts for about $26\%$. This implies that there is a big gap in the rates of individuals using the Internet between rural and urban areas.

What the empirical part of the study focused on is whether access to the Internet has a positive effect on access to external finance of households engaging in industry, commerce or agriculture. Therefore, we first investigated the relationship between access to the Internet and access to finance via simple descriptive statistics.

Table 5. Access to the Internet and access to finance.

Variable	Mean	Std	Observations
Access to the Internet($Z_1$)	0.15	0.60	421
No Access to the Internet	0.07	0.51	2339
Access to the Internet($Z_2$)	0.18	0.62	276
No Access to Economic-Related Information via the Internet	0.07	0.51	2471

suggests that there are about $15\%$ of households who can have access to loans among the households with access to the Internet, while there are only about $7\%$ of those who can have access to loans among the households with no access to the Internet. Moreover, if we restrict the set of households to those who pay attention to and search economic-related information via the Internet, then the rate of access to external fund is $18\%$, and otherwise it is only $7\%$. This intuitive observation impresses us with a significant positive effect of access to the Internet on the access to external finance of households engaging in industry, commerce or agriculture.

4.2. Empirical Methodology

The empirical part of this study attempted to disentangle whether access to the Internet has an important effect on access to external finance. The main potential outcome variable is an dummy variable Loan that equals 1 if a household has loans, and 0 otherwise. The variable of interest is an dummy variable $Z_1$ that equals 1 if the head of a household browses or searches information via the Internet, and 0 otherwise. Allowing for the characteristics of main explained and explanatory variables, we applied the two-valued logit model. Put differently, if X is a vector of control variables that may potentially determine the external fund availability for a household engaging in industry, commerce or agriculture, we estimate

$$\mathrm{Prob}(Y=1|X)=\frac{e^{\alpha Z+{\beta }^{\prime }X}}{1+e^{\alpha Z+{\beta }^{\prime }X}}$$

(29)

where Y is our outcome variable, which can be Loan or Credit Loan, and Z is the explanatory variable, which can be $Z_1$ or $Z_2$. Parameter $\alpha$ is the estimand we are interested in and tried to estimate, and $\beta$ is a vector of parameters on control variables to be estimated.

The set of control variables mainly consist of demographic characteristics of households. For example, female may be more conservative when making financial decisions, the schooling years of the head of a household may positively reflect its debt-paying ability [20]. We also control Head Age. Intuitively, relatively older heads of households may hate risk more, but this influence may not be linear and so its quadratic term is also incorporated into the empirical model in Equation (29). Admittedly, we vould use variables such as Female Head and Head Age to control for risk preference of each household, but these two variables may not be sufficient to reflect the risk preference of a household. Fortunately, the questionnaire of CHFS sets up a risk preference related question to survey the subjective attitude toward risk. The question is that If you had an asset, which of the following projects would you invest in? The options include “The project with both high return and high risk”, “The project with both relatively high return and relatively high risk”, “The project with both average return and average risk”, “The project with both relatively low return and relatively low risk”, and “Unwilling to take any risk”. We use “The project with both average return and average risk” as a benchmark to define a dummy variable Risk Preference equal to 1 if the risk preference level is higher than the benchmark, and 0 otherwise. Then, we incorporate this variable into our regression Equation (29). In addition, we also control the number of members of each household, housing conditions of households, residential zones (rural area or urban area) of households, and province fixed effects.

One may expect that household income is a paramount factor that should be considered by potential external fund suppliers when making lending policy. However, loan availability or access to finance itself also has an important effect on household income. Therefore, if we directly incorporate household income into the regression to control, then this obviously causes reverse causality problem. To avoid such endogeneity problem, allowing for the great importance of houses in Chinese family [58], we control Home ownership and Housing Area for wealth effect instead of directly controlling household income. Moreover, taking the striking imbalance of economic development among different regions in China into account, we also control eastern China, western China, and middle China fixed effects for the impacts of residential areas of households on their access to external finance. Finally, to reduce the effect of possibly outliers, we winsorize all continuous variables at the top $99\%$ and the bottom $1\%$ levels.

5. Empirical Results

In this Section, we mainly examine the effects of the use of the Internet on loan availability of households who run individual-owned firms or small and micro businesses in industry, commerce or agriculture. To address or mitigate possible endogeneity concern, we carefully conducted robustness analysis for our main empirical results.

5.1. Main Results

CHFS database provides the data on whether households have loans for those who engage in industry, commerce or agriculture. We measured this by variable Loan defined in Table 2. We uses access to the Internet variable $Z_1$ as the variable of interest to estimate its effect on outcome variable Loan. The regression results are reported in

Table 6. The effect of access to the Internet on households’ loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Loan
$Z_1$	0.60 ***	0.60 ***	0.48 **	0.54 **
	(0.23)	(0.23)	(0.22)	(0.22)
Head Age	0.29	0.03	0.03	0.03
	(0.04)	(0.04)	(0.04)	(0.04)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.06	−0.06	−0.06	−0.06
	(0.05)	(0.05)	(0.04)	(0.04)
Head Edu	0.05 **	0.06 **	0.04 **	0.06 **
	(0.02)	(0.02)	(0.02)	(0.02)
Female Head	−0.24	−0.26	−0.26	−0.29
	(0.19)	(0.19)	(0.19)	(0.19)
Household Size	0.09 *	0.09 **	0.08 *	0.09 *
	(0.05)	(0.05)	(0.04)	(0.05)
Home Ownership	0.71 **	0.71 **	0.79 **	0.80 **
	(0.32)	(0.32)	(0.32)	(0.32)
Housing Area	0.00	0.00	0.00	0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	0.29 *	0.29 *	0.40 **	0.39 **
	(0.17)	(0.17)	(0.17)	(0.17)
Risk Preference	0.56 ***		0.61 ***
	(0.17)		(0.17)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	8.57%	7.96%	5.82%	5.07%
Observations	2562	2609	2562	2609

Notes. This table presents the probability estimates of a two-valued logit model for the effect of access to the Internet on households’ loan availability. Specifically, the outcome variable Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. ***, **, and * indicates two-sided statistical significance at the $1\%$, $5\%$, and $10\%$ levels, respectively.

.

Columns (1)–(4) of Table 6 show that access to the Internet of the head of a household running an individual-owned firm or a small and micro business in industry, commerce or agriculture has a statistically significant positive effect on its access to external finance in the form of loans with statistical significance at the $1\%$ or $5\%$ level for any of the four different model specifications. Column (1) of Table 6 shows that the odds ratio from no access to the Internet ($Z_1=0$) to access to the Internet ($Z_1=1$) of loan availability is about $1.82$ ($e^{0.06}\approx 1.82$), which shows that the result is also economically significant. In other words, on average. the likelihood of obtaining loans for households with access to the Internet is about $1.82$ times that of households with no access to the Internet. This evidence strongly substantiates our prediction that access to the Internet helps alleviate small and micro businesses’ financing difficulty by raising their ability to access to the external finance. The mechanism through which access to the Internet acts on access to external fund is that more relevant information available through the Internet can effectively mitigate the information asymmetry between lenders and borrowers and reduce moral hazard or agency problem of borrowers as shown in our theoretical model.

As aforementioned, the access to the Internet variable $Z_1$ may contain too much noise in terms of measuring personal information dissemination through the Internet. For instance, one may concern that various information can be accessed via the Internet, which could have all kinds of uses, and also that the head of a household’s access to the Internet is not necessary to disclose or disseminate personal information to potential lenders. This may potentially contaminate the association between individuals using the Internet and access to external finance. For the sake of robustness, therefore, in the part of robustness that will be discussed soon, we further examine the influence of the head of a household who concerns economic information provided by the Internet on loan availability.

For the control variables, we find that the head’s schooling years have a positive effect on loan availability with statistical significance at the $5\%$ level. That is, the better the educational background of a household is, the higher the likelihood of access to external finance will be. We also evidence that the probability of loan availability is positively associated with both the number of members of a household and home ownership with significance at the $10\%$ and $5\%$ levels, respectively. This indicates that the better a household’s economic condition, the greater the probability obtaining a loan, which is consistent with the net worth effect (A) in our theoretical model. Additionally, we find that risk preference of a household significantly and positively affects its probability of loan availability. This result seems to be contrary to economic intuition. A possible interpretation for this is that the risk and profit of a project itself should directly influence on loan availability, which is clearly reflected in our model as important elements of the model. Hence, the risk preference of a household can be only a proxy of its project’s risk to the extent, but cannot reflect its project’s potential profit. Unfortunately, however, CHFS does not provide loan project-related data, and thus we cannot further explore a clear rationale for this phenomenon.

The next related question we try to investigate is the effect of access to the Internet on loan types. Concretely, we are interested in the effects of access to the Internet on the availability of pure credit loans and informal loans. We report relationship between access to the Internet and pure credit loan availability in

Table 7. The effect of access to the Internet on households’ credit loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Credit Loan
$Z_1$	−1.13 ***	−1.16 ***	−1.04 **	−1.07 **
	(0.42)	(0.42)	(0.42)	(0.42)
Head Age	0.02	0.02	0.03	0.03
	(0.10)	(0.10)	(0.09)	(0.09)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.05	−0.05	−0.05	−0.05
	(0.10)	(0.10)	(0.10)	(0.10)
Head Edu	0.07	0.07	0.07	0.07
	(0.06)	(0.06)	(0.05)	(0.06)
Female Head	−0.28	−0.29	−0.19	−0.21
	(0.42)	(0.42)	(0.42)	(0.41)
Household Size	−0.01	−0.01	−0.02	−0.02
	(0.10)	(0.10)	(0.10)	(0.10)
Home Ownership	0.63	0.65	0.33	0.35
	(0.00)	(0.00)	(0.00)	(0.00)
Housing Area	0.00	0.00	0.00	0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	1.00 **	1.00 **	1.07 ***	1.08 ***
	(0.40)	(0.40)	(0.38)	(0.37)
Risk Preference	−0.13		−0.15
	(0.17)		(0.34)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	11.06%	11.32%	9.25%	9.42%
Observations	2562	2609	2562	2609

Notes. This table presents the probability estimates of a binary response logit model for the effect of access to the Internet on households’ credit loan availability. Specifically, the outcome variable Credit Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has credit loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. *** and ** indicates two-sided statistical significance at the $1\%$ and $5\%$ levels, respectively.

.

The regression results in Table 7 show that access to the Internet has a significantly negative impact on pure credit loan availability. One plausible explanation for this is that more credit relevant information on each potential borrower is available for potential fund providers through access to the Internet of potential borrowers. This can effectively lower the net worth requirement for each borrower to behave by reducing agency rent as predicted by our theoretical model, which may in turn increase the amount or proportion of secured loans. This may lead to a decrease in pure credit loans, allowing for the possible substitution effect from the possible increase in the proportion of secured loans. One minor point we also point out is that the regression coefficient on rural area dummy variable in Table 7 assumes a significant positive sign. The most possible reason for this mainly lies in the collective property institution extensively rooted in rural area of China, which leads households in rural area to less or no effective collateral or pledge and makes pure credit loan be only means by which they can reach formal external finance.

We present the estimates on the relationship between access to the Internet and informal loan availability in

Table 8. The effect of access to the Internet on households’ informal loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Informal Loan
$Z_1$	0.06	0.09	−0.06	−0.02
	(0.16)	(0.16)	(0.16)	(0.16)
Head Age	−0.00	−0.00	0.00	−0.00
	(0.03)	(0.03)	(0.03)	(0.03)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.02	−0.01	−0.02	−0.02
	(0.03)	(0.03)	(0.03)	(0.03)
Head Edu	−0.03 *	−0.02 *	−0.03 **	−0.03 **
	(0.01)	(0.01)	(0.01)	(0.01)
Female Head	0.10	0.07	0.09	0.05
	(0.11)	(0.11)	(0.11)	(0.11)
Household Size	0.11 ***	0.11 ***	0.11 ***	0.12 ***
	(0.03)	(0.03)	(0.03)	(0.03)
Home Ownership	−0.26	−0.27 *	−0.25	−0.25
	(0.16)	(0.16)	(0.16)	(0.16)
Housing Area	−0.00 ***	−0.00 ***	−0.00 ***	−0.00 ***
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	0.04	0.02	0.15	0.12
	(0.11)	(0.11)	(0.11)	(0.11)
Risk Preference	0.53 ***		0.55 ***
	(0.13)		(0.13)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	3.75%	3.09%	2.62%	1.94%
Observations	2563	2610	2563	2610

Notes. This table presents the probability estimates of a binary response logit model for the effect of access to the Internet on households’ informal loan availability. Specifically, the outcome variable Informal Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has informal loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. ***, **, and * indicates two-sided statistical significance at the $1\%$, $5\%$, and $10\%$ levels, respectively.

. Clearly, Table 8 shows that there is no significant relationship between access to the Internet and informal loan availability. This further corroborates the conjecture mentioned in preceding paragraph. Scilicet, households running individual-owned or small and micro businesses cannot get more informal finance through access to the Internet. This is consistent with what we have in Table 7 and indirectly supports our conjecture that there is a substitution effect between secured and pure credit loan.

In sum, we found that access to the Internet plays a significant role in access to external finance of households running individual-owned firms or small and micro businesses. However, the increase in the likelihood of loan availability does not stem from the increase in pure credit loan availability or informal loan availability. Put differently, the enhancement of loan availability mainly derives from the mitigation in moral hazard and information asymmetry (via firm net worth effect) as predicted in our theoretical model. However, one may concern that a variety of information can be reached via the Internet, which could have all kinds of uses, and thereby focusing only on the dependent variable, the use of Internet, $Z_1$, may include confounding factors that could potentially contaminate the association between individuals using the Internet and access to external finance. For instance, Internet users may pay attention to political, military, or entertaining information, which could not help to enhance loan availability. To overcome or alleviate such concern, we display a series of robustness tests in the following subsection.

The above empirical results imply that entrepreneurs or managers of micro and small firms should pay more attention to their online information disclosure. This can reduce information asymmetry between borrowers and lenders, and thus help them access to external finance so that their financial difficulty can be alleviated in this way. Furthermore, access to external finance can push forward their sustainable growth. Stated differently, appropriately disclosing personal credit-relevant information via the Internet can help micro and small firms raise external debt and thus invest in more profitable projects, which is vital not only for solving their current financing difficulty but also for breaking through the bottleneck of their sustainable development (we thank the referee for suggesting adding this managerial implication to improve our paper a lot).

5.2. Robustness Tests

To further reduce the possible adverse impact from the noise of our measure of access to the Internet, $Z_1$, on our main results, we restricted the form of access to the Internet to those households whose heads pay attention to and capture economic-related information via the Internet. We define variable $Z_2$ to be an indicator to denote this type of access to the Internet, and rerun all the regressions similar to those reported in Section 5.1. We present this line of empirical results in

Table 9. Access to economic-related information via the Internet and loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Loan
$Z_2$	0.76 ***	0.82 ***	0.62 ***	0.69 ***
	(0.23)	(0.23)	(0.22)	(0.22)
Head Age	0.02	0.02	0.03	0.02
	(0.04)	(0.04)	(0.04)	(0.04)
$\mathrm{Head}{\mathrm{Age}}^2/100$	-0.05	-0.05	-0.06	-0.06
	(0.05)	(0.05)	(0.04)	(0.04)
Head Edu	0.05 ***	0.06 ***	0.04 *	0.05 **
	(0.02)	(0.02)	(0.02)	(0.02)
Female Head	−0.24	−0.26	−0.26	−0.29
	(0.19)	(0.19)	(0.19)	(0.19)
Household Size	0.09 **	0.09 **	0.09 **	0.09 **
	(0.04)	(0.05)	(0.06)	(0.05)
Home Ownership	0.72 **	0.72 **	0.50 ***	0.81 ***
	(0.35)	(0.32)	(0.32)	(0.32)
Housing Area	0.00	0.00	0.00	0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	0.27	0.26	0.39 **	0.37 **
	(0.17)	(0.17)	(0.17)	(0.17)
Risk Preference	0.56 ***		0.64 ***
	(0.17)		(0.17)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	8.79%	8.15%	5.99%	5.22%
Observations	2555	2596	2555	2596

Notes. This table presents the probability estimates of a two-valued logit model for the effect of access to economic-related information via the Internet on households’ loan availability. Specifically, the outcome variable Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. ***, **, and * indicates two-sided statistical significance at the $1\%$, $5\%$, and $10\%$ levels, respectively.

,

Table 10. Access to economic-related information via the Internet and credit loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Credit Loan
$Z_2$	−0.70 *	−0.75 *	−0.63	−0.6 *
	(0.42)	(0.42)	(0.41)	(0.41)
Head Age	0.05	0.06	0.06	0.07
	(0.10)	(0.10)	(0.09)	(0.09)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.08	−0.08	−0.08	−0.09
	(0.10)	(0.10)	(0.10)	(0.10)
Head Edu	0.05	0.05	0.05	0.05
	(0.06)	(0.06)	(0.04)	(0.05)
Female Head	−0.30	−0.31	−0.21	−0.23
	(0.41)	(0.41)	(0.41)	(0.41)
Household Size	−0.02	−0.02	−0.02	−0.02
	(0.10)	(0.10)	(0.09)	(0.09)
Home Ownership	0.50	0.53	0.22	0.24
	(0.73)	(0.71)	(0.70)	(0.68)
Housing Area	−0.00	−0.00	0.00	0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	1.09 ***	1.08 ***	1.16 ***	1.16 ***
	(0.40)	(0.40)	(0.37)	(0.37)
Risk Preference	−0.19		−0.22
	(0.35)		(0.34)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	9.71%	9.93%	8.03%	8.16%
Observations	2555	2596	2555	2596

Notes. This table presents the probability estimates of a binary response logit model for the effect of access to economic-related information via the Internet on households’ credit loan availability. Specifically, the outcome variable Credit Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has credit loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. *** and * indicates two-sided statistical significance at the $1\%$ and $10\%$ levels, respectively.

and

Table 11. Access to economic-related information via the Internet and informal loan availability.

Dependent Variable	(1)	(2)	(3)	(4)
	Informal Loan
$Z_2$	0.08	0.13	−0.02	0.04
	(0.17)	(0.17)	(0.17)	(0.17)
Head Age	-0.00	-0.01	0.01	0.00
	(0.03)	(0.03)	(0.03)	(0.03)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.02	−0.02	−0.03	−0.02
	(0.03)	(0.03)	(0.03)	(0.03)
Head Edu	−0.03 **	−0.02 *	−0.04 **	−0.03 **
	(0.01)	(0.14)	(0.01)	(0.01)
Female Head	0.10	0.07	0.09	0.05
	(0.11)	(0.11)	(0.11)	(0.11)
Household Size	0.11 ***	0.11 ***	0.12 ***	0.12 ***
	(0.03)	(0.03)	(0.03)	(0.03)
Home Ownership	−0.26	−0.27 *	−0.25	−0.25
	(0.16)	(0.16)	(0.16)	(0.16)
Housing Area	0.00 ***	−0.00 ***	−0.00 ***	−0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	0.04	0.01	0.15	0.13
	(0.11)	(0.11)	(0.11)	(0.11)
Risk Preference	0.54 ***		0.55 ***
	(0.12)		(0.13)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	3.80%	3.18%	2.65%	1.99%
Observations	2556	2597	2556	2597

Notes. This table presents the probability estimates of a binary response logit model for the effect of access to economic-related information via the Internet on households’ informal loan availability. Specifically, the outcome variable Informal Loan is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business has informal loans. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. ***, **, and * indicates two-sided statistical significance at the $1\%$, $5\%$, and $10\%$ levels, respectively.

.

Table 9 shows that the effect of access to the Internet is even stronger when we replace the access to the Internet measure, $Z_1$, with a more accurate measure $Z_2$. In particular, Column (1) of Table 9 shows that the economic magnitude of the effect of $Z_2$ on access to loans is greater than what we have obtained in Table 6. The odds ratio from no access to the Internet ($Z_2=0$) to access to the Internet ($Z_2=1$) of loan availability is about $2.14$ ($e^{0.76}\approx 2.14$), which is higher than that of reported in Column (1) of Table 6. This documents that the results in Table 9 is not only highly consistent with but also even stronger than that of in Table 6.

When we look at the results presented in Table 10 and Table 11, we found that they are highly consistent with those reported in Table 7 and Table 8. Therefore, this line of robustness analysis strongly reinforces our main results. In other words, our empirical evidence that access to the Internet has a significantly positive effect on access to external finance of small and micro businesses is highly robust. Clearly, our empirical evidence not only supports our main hypothesis, but also verifies the fundamental implications of our theoretical model. Thus, we have positively developed a theory on access to the Internet and access to finance, and evidenced the positive role played by access to the Internet.

5.3. Additional Analysis

To fully uncover the effect of access to the Internet on economic behavior of economic agents, we further explore whether access to the Internet can change households’ bank choice preference when they decide to apply loans. Traditionally, one may prefer a bank with close quarters to apply a loan, and therefore economic agents rely more on physical branches of banks. In contrast, since access to the Internet can relax the constraints of time and space, one may expect that access to the Internet can change this sort of bank choice preference. Fortunately, CHFS database provides the data on the main reason that a household chooses a specific bank to apply for a loan, where “time and bank location convenience” is one of the choices in the relevant question of the questionnaire. To examine this issue based on the data from CHFS, we define an indicator Convenience that equals 1 if a household chooses a bank to apply a loan according to convenience of time and distance, and 0 otherwise. We use $Z_1$ as a proxy of access to the Internet and specify $Convenience$ as the explanatory variable. We run a logit regression and present the results in

Table 12. Access to the Internet and households’ bank choice.

Dependent Variable	(1)	(2)	(3)	(4)
	Convenience
$Z_1$	−1.11 *	−0.84	−1.07 *	−0.81
	(0.59)	(0.54)	(0.59)	(0.55)
Head Age	−0.10	−0.12	−0.10	0.11
	(0.10)	(0.10)	(0.10)	(0.10)
$\mathrm{Head}{\mathrm{Age}}^2/100$	0.09	0.11	0.09	0.11
	(0.11)	(0.11)	(0.11)	(0.11)
Head Edu	0.04	0.04	0.05	0.05
	(0.07)	(0.07)	(0.07)	(0.06)
Female Head	−0.23	−0.32	−0.21	−0.29
	(0.55)	(0.54)	(0.54)	(0.53)
Household Size	−0.24 **	−0.22 **	−0.25 **	−0.23 **
	(0.11)	(0.11)	(0.11)	(0.11)
Home Ownership	−0.11	−0.18	0.00	−0.04
	(0.71)	(0.76)	(0.66)	(0.71)
Housing Area	0.00 ***	0.00 ***	0.00 ***	0.00 ***
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	0.69	0.75	0.57	0.65
	(0.53)	(0.52)	(0.48)	(0.47)
Risk Preference	0.76 *		0.73
	(0.40)		(0.40)
Province FE	Yes	Yes	No	No
Pseudo $R^2$	7.46%	5.93%	7.01%	5.60%
Observations	2555	2596	2555	2596

Notes. This table presents the probability estimates of a binary response logit model for the effect of access to the Internet on households’ bank choice. Specifically, the outcome variable Convenience is a 0–1 variable to denote whether a household running an individual-owned firm or a small and micro business chooses a bank to apply a loan on the basis of convenience of time and distance. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. ***, **, and * indicates two-sided statistical significance at the $1\%$, $5\%$, and $10\%$ levels, respectively.

.

Table 12 suggests that access to the Internet may result in the change in households’ bank choice for their loan applications. Columns (1) and (3) of Table 12 indicate that access to the Internet reduces the dependence of houses’ bank choice on time and distance constraints with statistical significance at least at the $10\%$ level. The signs of corresponding regression coefficients in Columns (2) and (4) of Table 12 are both negative, as expected, despite insignificance. Consequently, we found weak evidence that access to the Internet effectively lower the dependence of households on physical branches of banks when making decision on choosing a bank to apply loans. This once again convincingly supports our main point that access to the Internet has a positive effect on households’ access to external finance. More precisely, access to the Internet to some degree has expanded the households’ choice set of external financing sources.

Thus far, we have analyzed one aspect of the roles of access to the Internet in alleviating financing difficulty faced by small and micro businesses, that is, loan availability. One may also concern the other dimension, the price of a loan, or the interest rate charged on a loan. In our theoretical model, Corollary 2 shows that access to the Internet may push up the cost of debt under some distribution of firm net worth. Since CHFS database provides the data on loan rate, we run simple OLS regression to see the possible impact of access to the Internet on loan price and present the results in

Table 13. Access to the Internet and Loan Rate.

Dependent Variable	(1)	(2)	(3)	(4)
		Loan Rate
$Z_1$	0.20	0.10	0.17	0.07
	(0.80)	(0.79)	(0.77)	(0.78)
Head Age	0.04	0.05	0.03	0.04
	(0.21)	(0.21)	(0.21)	(0.21)
$\mathrm{Head}{\mathrm{Age}}^2/100$	−0.07	−0.08	−0.06	−0.06
	(0.21)	(0.21)	(0.21)	(0.21)
Head Edu	−0.06	−0.06	−0.05	−0.06
	(0.09)	(0.09)	(0.09)	(0.09)
Female Head	−1.08	−1.02	−1.15	−1.07
	(0.94)	(0.91)	(0.94)	(0.91)
Household Size	0.11	0.10	0.11	0.10
	(0.19)	(0.19)	(0.19)	(0.19)
Home Ownership	−0.95	−0.92	−0.70	−0.67
	(1.41)	(1.38)	(1.39)	(1.37)
Housing Area	0.00	0.00	0.00	0.00
	(0.00)	(0.00)	(0.00)	(0.00)
Rural Area	1.34 *	1.30 *	1.20 *	1.14 *
	(0.74)	(0.74)	(0.72)	(0.72)
Risk Preference	−0.31		−0.32
	(0.59)		(0.57)
Province FE	Yes	Yes	No	No
$R^2$	6.45%	6.29%	6.01%	5.80%
Observations	178	179	178	179

Notes. This table presents the OLS regression for the effect of access to the Internet on loan rate. Specifically, the outcome variable Loan Rate is the interest rate charged by banks on loans in percentage. Column (1) offers the estimate results with all control variables; Column (2) gives the estimate without controlling the subjective attitude toward risk of households; Column (3) reports the estimate with no province fixed effects; and Column (4) presents the estimate without controlling both the subjective attitude toward risk of households and province fixed effects. Robust standard errors are in parentheses. * indicates two-sided statistical significance at the $10\%$ levels, respectively.

. We found that access to the Internet does have a positive effect on the cost of debt (Loan Rate) in terms of the signs of relevant regression coefficients, but none of them is statistically significant. Of course, this result comes from a relatively small sample, which may somewhat impact the regression result. We plan to further investigate this issue in depth in future research when more data are available.

6. Conclusions

In this paper, we study the impact of using the Internet on individual or small and micro businesses’ access to finance or sustainable growth, and also social welfare. We also attempt to investigate the economic mechanisms through which such impact works. To this end, we first build a simple theoretical model based on information asymmetry literature, such as Holmstrom and Tirole [16], Tirole [57], etc., to analyze the economic mechanisms linking access to the Internet with access to external finance. One overriding implication of our model is that access to the Internet helps to mitigate the agency problem between creditors and borrowers caused by the information asymmetry between the two kinds. Our model predicts that access to the Internet will have a prominent positive effect on access to finance. Since more valuable investment projects can be taken on account of the positive effect, the use of the Internet will also shrink the gap between optimal social gross investment level and the actual one. Moreover, access to the Internet will promote the sustainable growth of individual-owned or small and micro businesses thanks to the alleviation of their financial constraints. Hence, a direct consequence of such effects will be the improvement of total social welfare.

To examine the main empirical implication of our model, we mainly apply binary response logit model to our setting. Consistent with our theoretical predictions, evidence from our empirical analysis shows that whether the head of a household running individual-owned or small and micro businesses browses or searches information or news via the Internet has a significant positive effect on its loan availability. This implies that the use of the Internet has broadened the breadth of credit availability insofar as households run small and micro businesses. This effect is still highly significant after controlling householders’ risk preference and province heterogeneity. Such effects have important implications but have not been previously documented.

Since a variety of information can be reached via the Internet, which could have all kinds of uses, focusing only on the dependent variable, the use of the Internet, may include confounding factors that could potentially contaminate the association between individuals using the Internet and access to external finance. To more thoroughly disentangle the relationship between access to the Internet and access to external finance, therefore, we further examine the influence of the head of a household who concerns economic information provided by the Internet on loan availability. Our evidence indicates that households whose heads pay attention to and capture economic-related information through the Internet are more likely to be funded. In other words, after purging much noise possibly included in the explanatory variable, individuals using the Internet, our empirical finding that Internet use has enhanced individual-owned or small and micro businesses’ credit availability is still true and even stronger, which at least has partially addressed the financing difficulty of households or those businesses run by them.

Our empirical evidence implies that entrepreneurs or managers of micro and small firms should pay more attention to their online information disclosure. This can reduce information asymmetry between borrowers and lenders, and thus help them access to external finance so that their financial difficulty can be alleviated in this manner. Furthermore, access to external finance can push forward their sustainable growth. Stated differently, appropriately disclosing personal credit-relevant information via the Internet can help micro and small firms raise external debt and thus invest in more profitable projects, which is vital not only for solving their current financing difficulty but also for breaking through the bottleneck of their sustainable development.

We also investigate the possible effects of access to the Internet on access to finance from alternative dimensions. For instance, we examine whether access to the Internet can change households’ bank choice preference when they decide to apply loans and find weak evidence that access to the Internet may result in the change in households’ bank choice for their loan applications. Consequently, we find weak evidence that access to the Internet effectively lower the dependence of households on physical branches of banks when making decision on choosing a bank to apply loans. This once again convincingly supports our main point that access to the Internet has a positive effect on households’ access to external finance. Put differently, access to the Internet to the extent has expanded the households’ choice set of external financing sources.

Our study is among the first to explore the relationship between access to the Internet and access to finance and its economic mechanisms and economic results both theoretically and empirically to understand their impact on economic agent behaviors, sustainable growth of individual-owned or small and micro enterprises, and social welfare. Both our theory and empirical evidence demonstrate that individuals using the Internet has been a paramount important driver for alleviating the financing difficulty of small and micro businesses in China. Allowing for the important role in economic sustainable development played by small and micro businesses in China, manifestly, alleviating financing difficulty of those firms may have important implications. First, alleviating financing difficulty of small and micro businesses via access to the Internet is conducive to their sustainable growth by breaking through their bottle necks of sustainable growth. Secondly, access to the Internet makes more profitable projects be financed through access to external finance, which no doubt can improve total social welfare. Finally, since there are more good projects that can be financed, and thereby the aggregate social investment efficiency must be enhanced.

Based on the above analysis, the main policy implication of our study is that policy makers may pay particular attention to formulate Internet finance supported relevant policies or attempt to encourage the use of the Internet. However, one caution we would like to emphasize is that Internet finance is still a newborn thing and thereby it must not only have a bright side but also a dark side, such as new Internet finance related types of risk. Therefore, policy authorities should also pay a special attention on this side of the use of the Internet. Admittedly, this is beyond the scope of this study, and we cannot say much on them. Our study also has some limitations. For example, our theoretical model fails to endogenize the choice of the degree of access to the Internet. We base our empirical analysis on cross-sectional data, which may not be able to identify the dynamic effects of access to the Internet on access to finance. Additionally, our variables of access to the Internet may not be precise measures of the process of personal information dissemination via the Internet due to data unavailability. We plan to overcome those limitations in our future research.