Credit-scores in lending-based crowdfunding : crowdfunding platform credit-scores vs. third-party credit-scores and the effects of the interest component and loan term

Credit-scores in lending-based crowdfunding

Crowdfunding p latform credit-scores vs. third -party credit-scores and the effects of the interest component and loan term

Jeroen Snellen (10070389) 19/8/2016

MSc. in Business Administration – Entrepreneurship and Innovation Amsterdam Business School - University of Amsterdam

Table of contents STATEMENT OF ORIGINALITY ... 3 ABSTRACT... 4 1. INTRODUCTION ... 5 2. THEORETICAL FRAMEWORK... 7 CROWDFUNDING ... 7

FOUR FORMS OF CROWDFUNDING ... 8

INVESTMENT CROWDFUNDING ... 9

EQUITY CROWDFUNDING ... 9

LENDING-BASED CROWDFUNDING ... 10

TRADITIONAL BUSINESS LENDING ... 11

INFORMATION ASYMMETRY ... 12

SOLUTIONS TO INFORMATION ASYMMETRY... 13

CREDIT-SCORES... 14

THE LOAN TERM AND THE INTEREST COMPONENT ... 15

HYPOTHESES AND CONCEPTUAL MODEL ... 16

3. DATA AND METHOD ... 18

DATA COLLECTION ... 18 SAMPLE ... 18 VARIABLES ... 19 DATA CHANGES ... 20 STATISTICAL APPROACH ... 21 4. RESULTS ... 22 CORRELATIONS ... 22 MULTIPLE REGRESSION ... 23 5. DISCUSSION ... 25 6. CONCLUSION... 28 7. REFERENCES ... 29

Statement of originality

This document is written by Jeroen Snellen who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Abstract

As crowdfunding is becoming a mainstream way of financing businesses, academic research on how investors make a risk assessment of their investment decision is still limited. Following the academic literature, crowdfunding challenges issues related to information asymmetry. Quality signals are described as a solution to reduce information asymmetry. This research is concerned with quality signals as predictors of crowdfunding success within a lending-based crowdfunding context for businesses. Using multiple regression, the effects of third-party credit-scores, crowdfunding platform credit-scores, loan terms and interest rates are tested. Crowdfunding success is measured as the time it took for a crowdfunding campaign to be fully funded. There are no fully significant results for one of the hypotheses. Only the effect of third-party credit-scores on the speed of funding is

1. Introduction

Crowdfunding gained momentum in recent years. In the beginning, crowdfunding was mainly used in the creativity-based industry, now it is becoming mainstream (Agrawal, Catalini & Goldfarb, 2013). It has given businesses a new opportunity for (early-stage) finance. With the commercialization of the Internet, crowdfunding became a meaningful method. The Internet has lowered search costs and made it possible for people to fund from a larger distance (Agrawal et al., 2013). There are different forms of crowdfunding, this research will focus on lending-based crowdfunding as a way to fund businesses. This form has developed recently and is growing at a fast pace (Belleflamme et al., 2013). Like with all forms of crowdfunding, searching for potential new ventures for investment becomes more efficient and effective. Through the Internet, it is also possible to target a large audience, which makes it possible for a relatively big group to make smaller investments (Agrawal et al., 2013; Ahlers et al., 2012). As with every form of investing, there are potential risks associated with this form of crowdfunding. The main concern is associated with the concept of information asymmetry (Agrawal et al., 2013; Ahlers et al., 2012; Belleflamme et al., 2013). Entrepreneurs looking for investments can offer a limited amount of information on a crowdfunding platform and for small investors the costs of getting quality information are high (Ahlers et al., 2012; Belleflamme et al., 2013). For this reason, it is important that entrepreneurs signal the right information that reflects the quality of the offering (Ahlers et al., 2012; Connelly et al., 2011).

Crowdfunding is a recent phenomenon that is developing at a quick pace. The amount of money raised by crowdfunding platforms around the world has increased fast and steadily (Burtch, Ghose & Wattal, 2013). Because it is a recent development there is limited academic research considering the topic. Lending-based crowdfunding is an even younger field of study and because of stricter regulation even more complex. This research is concerned with crowdfunding as a way for businesses to raise capital. Crowdfunding did not only give businesses the opportunity to raise capital in a new way, it also gave access to the public to make these investments. Financial authorities question if the public is able to qualify these risks to make a considered investment decision. Crowdfunding has great potential but needs to be regulated in a way that investors are well informed about their investment decision.

This research focuses on how investors build their decision to engage in lending-based crowdfunding. From the literature is evident that certain information is important for the investment decision. Essentially, the investment decision for issuing a loan is based on the creditworthiness of a company. This creditworthiness is normally assessed through a company’s credit score. The

theoretical framework of this research will go in-depth on how credit scores are assessed by investors to make a grounded investment decision.

As crowdfunding is becoming a more important form for raising (seed) capital, it is important to research its specific characteristics. Crowdfunding involves a large group of people contributing relatively small amounts through an online platform. This often means that the two entities involved do not know each other and have limited information. This makes crowdfunding sensitive to illegal practices and fraud and raises questions related to asymmetric information. It becomes more relevant when it comes to crowdfunding as a way to invest capital because investing involves certain risks of losing the investment. The complexity associated with lending-based crowdfunding makes it an interesting research topic. Because it is a growing phenomenon it is also important to conduct academic research and fully understand its dynamics. Crowdfunding has already raised billions of dollars throughout recent years. A large amount of projects with millions of people backing them proves crowdfunding is becoming mainstream. Academic literature on crowdfunding is still in its infancy besides few efforts to fathom the phenomenon (Agrawal, Catalini & Goldfarb, 2011; Burtch et al., 2013).

2. Theoretical framework

Crowdfunding

Crowdfunding is described as a form of fundraising where a group raises money to support a particular goal (Ahlers et al., 2015). Crowdfunding is a relatively new phenomenon that has been growing explosively in recent years. According to www.crowdsourcing.org (2016), crowdfunding raised more money each year since its introduction and in 2014 $ 16.2 billion was raised with successful crowdfunding campaigns. The dynamics of crowdfunding allow for small investors to engage in funding that was otherwise not available to them (Ahlers, 2015). There are various forms of crowdfunding and crowdfunding campaigns can vary in the goals that are set to reach (Ahlers et al., 2015). Goals to be achieved can range from political campaigns to art projects to software design. Funding can be raised using multiple crowdfunding alternatives including donations, reward-based methods, lending and equity selling. Funding goals can differ from funding for small projects to funding of large amounts of capital as an alternative to bank loans or venture capital financing (Schwienbacher & Larralde, 2010).

Crowdfunding stems from the broader concept of crowdsourcing. Crowdsourcing involves using the crowd to come up with ideas, solutions, and feedback for corporate activities

(Belleflamme, Lambert & Schwienbacher, 2013; Poetz & Schreier, 2012). The main difference between crowdsourcing and crowdfunding is that crowdsourcing involves ideas and time as input and crowdfunding involves capital as input (Belleflamme et al., 2013). Because crowdfunding involves capital as input, the legal limitations differ compared to crowdsourcing (Bradford, 2012). Belleflamme et al. (2013) give the following definition of crowdfunding: “Crowdfunding involves an open call, mostly through the Internet, for the provision of financial resources either in form of donation or in exchange for the future product or some form of reward to support initiatives for specific purposes.” The reward can be either monetary or non-monetary, non-monetary rewards are used in many different forms of crowdfunding (Belleflamme et al., 2013).

Because crowdfunding is used for many different causes in a great variety of ways, according to Mollick (2014), a broad definition is illusive. Mollick (2014) argues that a narrow definition is more salient in the context of new ventures and entrepreneurial finance. Mollick states: “Crowdfunding refers to the efforts by entrepreneurial individuals and groups – cultural, social, and for-profit – to fund their ventures by drawing on relatively small contributions from a relatively large number of individuals using the internet, without standard financial intermediaries.” This definition is more specific about the fundraising entity but doesn’t address the goal of the crow dfunding effort and the goal of the investors (Mollick, 2014). Mollick (2014) explains the many different purposes fundraisers and investors have to engage in crowdfunding.

In the beginning, crowdfunding was most often used to fund one-time projects. Friends & family were the most important source of capital to fund events, charity or to initiate a small project (Mollick, 2014). Crowdfunding is becoming increasingly popular as a source of (seed) capital for businesses (Schwienbacher & Larralde, 2010). Because crowdfunding is still in its infancy, it is not clear to what extent crowdfunding will be a substitute for traditional forms of venture funding. Rules and regulations around especially lending-based and equity crowdfunding are different between countries and do not always allow for crowdfunding to reach its full potential (Bradford, 2012; Mollick, 2014).

The dynamics of crowdfunding do not only allow to raise capital; it can be used for several different purposes. Through reward-based crowdfunding, entrepreneurs can test the demand for a product which can save a lot of additional capital and effort (Mollick, 2014). Crowdfunding is also described as an effective marketing tool. The large extent of people reached with a crowdfunding campaign have been exposed to the company and can become future customers (Mollick, 2014). Compared to other early stage crowdfunding usually doesn’t offer benefits like expertise, governance and network (Ferrary & Granovetter, 2009).

Four forms of crowdfunding

Because raisers can choose between several forms of crowdfunding, it is important to get an understanding of the reason behind choosing a certain form of crowdfunding (Belleflamme et al., 2013). Crowdfunding can be divided into four different forms; donations, reward-based, lending-based and equity crowdfunding (Cumming & Johan, 2016). Lending-lending-based and equity crowdfunding will be explained in-depth in a later section of the theoretical framework. Donations crowdfunding is straightforward; funders can donate money for, in most cases, a charitable project. The funders of the project do not expect any return other than a cautious spending of the proceeds. Examples are crowdfunding for people in disaster areas or helping individuals to get medical help or education (Cumming & Johan, 2016). Reward-based crowdfunding involves a non-monetary reward such a product, ticket or a token of appreciation (Cumming & Johan, 2016). The world’s largest and probably most famous crowdfunding platform is Kickstarter. To date, Kickstarter raised more than 3,2 billion USD. Mollick (2014) performed one of the most comprehensive crowdfunding studies up to date to understand to underlying dynamics of success and failure in Kickstarter-projects. Mollick (2014) found that both personal networks and underlying project quality are related to the success of crowdfunding efforts.

Investment crowdfunding

The way companies receive sources of financing can be divided into equity and debt. An equity investment is a direct investment into a company. Investors receive shares for their invested money and gain control over the company. At the same time, they bear the full risk for their investment (Schwienbacher & Larralde, 2010). Debt financers will not become owners of the company but engage in a contractual agreement in which capital is exchanged in return for an interest on the principal amount. The principal amount will be returned over a certain period of time. Debt holders have senior claims over equity holders and possibly require collateral (Schwienbacher & Larralde, 2010). Table 1 shows the common sources of equity and debt financing.

Table 1. Different types of entrepreneurial finance investors, grouped by debt and equity claims (Schwienbacher & Larralde, 2010)

Equity crowdfunding

Equity crowdfunding gives small investors the opportunity to fund businesses in exchange for shares in the company (Ahlers et al., 2015). The main difference between equity crowdfunding and more traditional forms of raising equity finance is the way the funding process takes place (Ahlers et al., 2015). Businesses use an online platform that allows everyone to see the investment opportunity. Because a larger group of investors is reached, equity crowdfunding allows smaller investors to make a relatively smaller contribution (Ahlers et al., 2015). Ahlers et al. (2015) give the following definition of equity crowdfunding: “Equity crowdfunding is a method of financing whereby an entrepreneur sells equity or equity-like shares in a company to a group of (small) investors through an open call for funding on Internet-based platforms.”

Because there are various risks associated with equity crowdfunding, the market for equity crowdfunding is subject to regulations (Ahlers et al. 2015, Bradford, 2012). The legislation on

crowdfunding differs between countries and equity crowdfunding is restricted in many countries. Up until the beginning of 2013, equity crowdfunding was restricted in the United States. By signing the JOBS act equity crowdfunding became possible. This means unregistered securities are available to non-accredited investors. OECD countries where equity crowdfunding is already possible for a longer period of time are the United Kingdom, Ireland, France, the Netherlands, Switzerland and Australia (Ahlers et al., 2015). For certain OECD countries, like Germany, financial constructions within legislation frameworks allow for revenue and profit sharing arrangements (Ahlers et al., 2015).

Ahlers et al. (2015) are one of the first to conduct empirical research on equity

crowdfunding. Using data obtained from ASSOB, an Australian-based equity crowdfunding platform, they find evidence that signaling is important for investors1_{. Especially potential risk factors, financial} roadmaps and board size and structure are important (Ahlers et al., 2015). Another interesting finding is the fact that 53% of the investors invest in projects within their own state of residence (Ahlers et al., 2015). This is in contrast with the statement that equity crowdfunding can eliminate distance-related risk factors that are associated with investing. As the equity crowdfunding market is supposed to operate in a rational manner, financial and governance material provided are important for investors (Ahlers et al., 2015). For investment-seekers is important to proficiently present

financial projections and roadmaps (Ahlers et al., 2015).

Lending-based crowdfunding

Lending-based crowdfunding, also referred to as crowdlending or peer-2-peer lending, exists in many different forms ranging from (small) consumer lending to developing world micro credit to (larger) business loans (Mitra, 2012). According to www.crowdsourcing.org (2016), lending-based crowdfunding dominated the industry with over $ 11.8 billion raised in 2014. There is a distinction between lending-based crowdfunding with a more charitable character, where only the principal or the principal and a marginal amount of interest needs to be repaid, and lending-based crowdfunding that becomes interesting from an investor’s perspective, with larger interest rates (Bradford, 2012). This research has a focus on the latter.

One of the first crowdfunding platforms that offered loans with a serious amount of interest to be repaid are U.S. based Prosper and Lending Club, focusing on consumer loans. Interest usually depends on the borrowers “credit-risk”. Because there is an interest involved these offerings are

1_{It can be questioned to what extent this platform really fits for their research within their own definition.}

ASSOB is only available for accredited (sophisticated) investors. This way the ‘call’ for funding is not really open to everyone. Still, it has to be acknowledged that in this online environment larger syndicates are formed to fund businesses with an equity transaction.

viewed as securities and are subject to securities regulation (Bradford, 2012). To assess the “risk” on most platforms offerings get a score. There is a great variety between the credit-scores, interest offered, and (financial) information given between the different platforms. Lending-based crowdfunding grows more popular under businesses while banks and other credit facilitators seem to withdraw under stricter regulations after the financial crisis. Also starting businesses, that lack the financial history for a substantial credit-score, get funded through lending-based

crowdfunding.

In obtaining credit, smaller businesses tend to have more problems than larger businesses due to higher failure rates and vulnerability to economic downturns. Information to make grounded decisions on granting a loan is less available in comparison to larger companies. The lack of detailed financial information makes it difficult to determine the creditworthiness of small businesses (Degryse & van Cayseele, 2000). Banks provide small business with other financial products other than loans. For this reason, they are thought to be able to gather useful information to make a proper assessment of the creditworthiness of small businesses (Degryse & van Cayseele, 2000). Banks used to be the main issuer of loans to small businesses (Degryse & van Cayseele, 2000).

Traditional business lending

While lending-based crowdfunding is growing, the most prominent source of credit for businesses are still banks. Banks, like all other investors, face a situation of information asymmetry when evaluating a lending application (Binks & Ennew, 1997). This counts especially for new businesses as they cannot provide a substantial track-record (Berger & Udell, 1998). Information to make a grounded lending decision is either unavailable, too costly to assess or very difficult to interpreted, creating the two problems: adverse selection and moral hazards. The adverse selection could potentially lead to two types of errors. The first error is lending to businesses that default, the second error is not lending to businesses that are able to return the principal and interest. Margins on loans are low and for this reasons banks try to minimize the type I error (Mason & Stark, 2004). The risk of moral hazards arises because banks are not able to monitor businesses once the loan has been granted. Businesses could potentially put less effort in or take on riskier projects. For this reason, a bank is focused on financial ratios in the lending decision (Mason & Stark, 2004). Especially collateral is important as it secures (a part of) the principal amount beforehand. It aligns the interest of the business with that of the lender and shows the confidence of the entrepreneur in his abilities and the success of the project. Therefore, it addresses the potential problem of adverse selection before the lending decision and the problem of moral hazards afterwards (Berger & Udell, 1998).

Essentially the two main considerations in the lending decision will be the presence of collateral to guarantee the loan and the company’s ability to generate sufficient cash flow to be able

to repay the loan. Collateral can recover the funds from the liquidation of the company assets or the personal assets of the entrepreneurs. In the process of evaluating a business , collateral is a

necessary condition for the loan and only after sufficient collateral is available the cash flows will be evaluated (Mason & Stark, 2004).

Information asymmetry

Individuals in households, businesses and governments make decisions based on information (Connelly et al., 2011). This information can be either public and available to everyone, or private and only available for a certain amount of people (Connelly et al., 2011). Because some information is private, information asymmetries occur between people who have this information and people who could possibly make better decisions if they would have this information. Many economic models of decision-making processes would ignore information asymmetries by implying perfect information. Economists assumed that markets with minor information imperfections would have the same characteristics as markets with perfect information (Stiglitz, 2000). Stiglitz (2000) has highlighted the importance of information in the decision-making process in the marketplace and the differences in having perfect and imperfect information. This has revealed limits to existing economic models and showed additional insights to others (Stiglitz, 2000). Two types of information are important in relation to information asymmetry. The first one is concerned with the quality of the information given. The second one is concerned about intent, this is important when one party is also concerned with another party’s behavioral intentions (Stiglitz, 2000).

Information and incentive problems can cause inefficient allocation of resources in capital markets for businesses. To overcome these problems disclosure and institutions that facilitate credible disclosure between companies and investors play an important role (Healy & Palepu, 2001). There are in general many businesses looking for investments and there is a certain amount of capital available by investors. Information problems occur while trying to match supply and demand in an efficient way. Entrepreneurs in need of capital have bette r information about the underlying value of the company than investors. Because they want fast access to capital at the low cost they also have an incentive for overstating the value of the company (Healy & Palepu, 2001). In a crowdfunding context, an entrepreneur can leave certain information behind or give false information (fraud) (Agrawal et al, 2013).

The big issue associated with crowdfunding is the lack of information available to the

investor. The platform only enables a limited amount of information to be obtained by investors and smaller investors lack resources and experience to evaluate investment opportunities properly (Ahlers et al., 2015). This results in a concern about information asymmetries between

may not be able to determine anything concrete about the true quality of the company and even high-performing companies will receive no investments at all (Ahlers et al., 2015). The “lemons” problem (Akerlof, 1970) or the adverse selection problem is caused by information differences between entrepreneurs and investors and could potentially lead to market failure (Agrawal et al., 2013; Healy & Palepu, 2001). After the investment is made, an agency problem arises because investors do not intend or are not allowed to play an active role in the management of the businesses they invested in. Once an investment is made, the entrepreneur has an incentive to engage in risky behavior that would maximize his own interest because he does not (completely) bears risks anymore (Jensen & Meckling, 1976).

Solutions to information asymmetry

Agrawal et al. (2013) describe three ways to overcome the problems related to information

asymmetry. The first way is to provide quality signals. Agrawal et al. (2013) describe many different online settings in which quality signals are important. The example related to crowdfunding derives from what is called venture quality in the research conducted by Ahlers et al. (2015). The growing equity crowdfunding market and a substantial amount of funds that have been raised through equity crowdfunding show that investors have been able to obtain at least some information regarding the quality of the start-ups on the platform (Ahlers et al., 2015). Especially in a crowdfunding context, the real quality of a company cannot be observed directly and this is the reason why investors look for signals that co-vary with the underlying quality (Ahlers et al., 2015). These signals are considered useful indicators for the probability the start-up will succeed. A set of observable characteristics gives potential investors the opportunity to make an estimation of the value of the start-up (Ahlers et al., 2015). Effective signals differ in the extent to which they are observable and the signal cost (Connely et al., 2011). A signal is observable when it is noticed and understood by investors. The cost of producing a signal must not be greater than the benefits and dishonest signals mu st not be rewarded (Ahlers et al., 2015). Ahlers et al. (2015) make a distinction between fact-based signals and performance-based signals. The entrepreneurs cannot choose the information given by fact-based signals, e.g. external certification and board experience. Performance-based signals, e.g. capital market roadmap and risk level, are set by the entrepreneurs themselves (Ahlers et al., 2015).

Healy & Palepu (2001) state that there are several solutions to overcome the problems related to information asymmetry. They state that optimal contracts between investors and entrepreneurs and regulation on information disclosure can be effective ways to solve the

information problem. To uncover the superior information held by entrepreneurs there is a demand for information intermediaries such as financial analysts and rating agencies (Healy & Palepu, 2001). Healy & Palepu (2001) provide a schematic overview of the role of disclosure, information and

financial intermediaries in the working capital markets (figure 1). Essentially the capital and information flows are between household savings and business firms. Capital flows are either directly or between a financial intermediary. Examples of direct capital flows are private equity and business angel financing. Indirect capital flows go through financial intermediaries, such as banks, venture capital funds and insurance companies (Healy & Palepu, 2001). The information flows show that investors can communicate directly with businesses or through intermediaries. For example, direct information can be communicated with press releases or financial statements. Information flows can also go through financial intermediaries or information intermediaries, such as financial analysts (Healy & Palepu, 2001). Information intermediaries collect private and public information and evaluate that information in order to give advice.

Figure 1. Financial and information flows in a capital market (Healy & Palepu, 2001)

To what extend information intermediaries can limit information asymmetry depends on economic and institutional factors (Healy & Palepu, 2001). Depending on factors that include proprietary costs, regulatory imperfections, and incentive problems intermediaries will either eliminate information asymmetry or leave a residual information problem. Though the schematic overview was created before crowdfunding existed, crowdfunding platforms fit in the model as both financial

intermediaries and information intermediaries. Both capital and information flow through the crowdfunding platform.

Credit-scores

To inform investors about the quality of the offering, platforms usually provide investors with a credit-score on the company that is lending money through the platform. This credit-score is mainly

based on quantitative financial figures or ratios like liquidity, solvability, profitability and relief capacity (www.collincrowdfund.nl, 2016; www.geldvoorelkaar.nl, 2016). Another method of obtaining quantified information on lenders are credit-scores by third parties (Peterson, 2004). Third-party credit scoring agencies gather data on a company’s (credit) payment history, market data and financial information to assess the probability a company will repay their loan obligations (Berger & Frame, 2007).

Third-party credit-scores used to be only available for larger companies but are now widely available for all types of businesses, through agencies like Dun & Bradstreet, Graydon and Creditsafe. Dun & Bradstreet credit scores are the most widely used credit score for small businesses and are based on a company’s credit history, information from public filings (e.g. bankruptcy proceedings and lawsuits) and information on sales, company value and working capital (Kallberg & Udell, 2003). Companies do not necessarily need to submit financial data to credit-scoring agencies. When they do, often only balance sheet information is given while income statements, for competitive reasons, are not. Kallberg & Udell (2003) found that Dun & Bradstreet’s credit-scores are a good predictor for small business failure, mainly based on a company’s payment history. According to Cassar, Ittner & Cavaluzzo (2015), companies with a good third-party credit-score face lesser loan denials of banks.

Third-party credit-scores have been a very cost effective tool for lenders to evaluate

companies with a loan request (Berger & Frame, 2007). Reduced cost of transferring information has led to the broader availability of third-party credit scores. Cowan & Cowan (2006) have found that an increasing number of banks is using these scores in their loan approvals. The main reasons for banks to use third-party credit-scores is to quantify the credit, simplify loan applications and have

inexpensive access to additional information (Cowan & Cowan, 2006). Still, banks rank cash flow information more important in loan approvals (Cowan & Cowan, 2006).

The loan term and the interest component

Under the consideration of information asymmetry, the loan term can be considered as a quality signal (Flannery, 1986). Flannery (1986) differs between good companies and bad companies and the quality can be assessed by the firm’s private information. Good companies will borrow short-term at a relatively lower interest rate compared to bad companies when the private information is available to lenders. Flannery (1986) predicts a positive relation between loan term and the risk of the loan, i.e. short-term loans predict lower risk and a higher loan term predicts higher risk. In line with the research of Flannery (1986). Berger, Espinosa-Vega, Frame & Miller (2005) find the same risk-loan term relation conducting research on small firms in the U.S. based on data from 1997. In this research, bank ratings are the proxy for risk. Ortiz-Molina & Penas (2008) further explore these findings, controlling for the size of the company and its age and also detect a positive relationship

between risk and the loan term. Ortiz-Molina & Penas (2008) also investigate the relationship between a lenders relationship with their bank and the loan term and cannot find a significant effect. Berger et al. (2005) find that third party credit scoring reduces information asymmetry, this is consistent with the findings of Flannery (1986). Finally, Kirschenmann & Norden (2012) find a

positive relationship between borrower risk and loan term, where this relationship is stronger for loans for individuals compared to businesses. An important note is that all these findings are within a banking context and are not tested within the crowdfunding context.

Hypotheses and conceptual model

Following the literature in the theoretical framework there is a profound basis to form hypotheses and test a conceptual model for predicting crowdfunding success. According to Agrawal et al. (2013) and Ahlers et al. (2015) quality signals are important to engage in the crowdfunding decision. Following Mollick (2014) quality signals can indeed predict crowdfunding success in a reward -based crowdfunding setting. Healy & Palepu (2001) stress the importance of credit-scores as an important quality signal. These findings give relevance to an overall model in which quality signals predict crowdfunding success in a lending-based crowdfunding context.

From the literature is evident that institutions that provide loans to businesses use both their own credit-score and a third party credit-score. Kalberg and Udell (2003) find that third party credit-scores are a good predictor for business failure. They base their findings on data f rom Dun and Bradstreet’s credit-scores. Dun & Bradstreet’s credit scores are commonly used by lending institutions. Their results are in line with research from Cassar, Ittner & Cavaluzzo (2015) who find that good third-party credit scores result in lesser loan denials from banks. Berger & Frame (2007) underline these findings by stating that third-party credit-scoring is an efficient tool for assessing the risk associated with a loan. Following this academic literature, the following hypothesis is stated: H1: Third-party credit-scores are a predictor for the speed of the funding.

Cowan & Cowan (2006) find that banks rely on third-party credit-scores in their lending decision. Banks are more reliant on cash flow information in their lending decision. As this information is important in the platform credit-scores (www.collincrowdfund.nl, 2016; www.geldvoorelkaar, 2016), the following hypothesis is stated:

H2: Crowdfunding platform credit-scores are a predictor for the speed of funding.

Flannery (1986) conducted research on the relationship between loan term and loan risk. Flannery (1986) compares between good companies and bad companies and finds that good companies have shorter loan terms than bad companies. These findings are confirmed in research conducted by Berger et al. (2005), Ortiz-Molina & Penas (2008) and Kirschermann & Norden (2012). The latter two also control for company size and age (Ortiz-Molina & Penas, 2008) and loans to

individuals compared to businesses (Kirschermann & Norden, 2012). The following hypothesis is stated:

H3: Loan terms are a predictor for the speed of funding.

Flannery (1986) also stresses the relationship between the interest rate and loan risk. Where bad companies with higher risk borrow at higher interest rate than good companies with lower risk. For this reason, the following hypothesis is stated:

H4: Interest rates are a predictor for the speed of funding.

This results in the following conceptual model:

Kalberg and Udell, 2003; Cassar, Ittner & Cavaluzzo, 2015; Berger & Frame, 2007

Third-party credit-score

Crowdfunding platform

credit-score

Loan term

Interest rate

Campaign success (Speed

of funding)

Cowan & Cowan, 2006

Flannery, 1986; Berger et al., 2005; Ortiz-Molina & Penas, 2008; Kirschermann & Norden, 2012

Flannery, 1986

3. Data and method

Data collection

To collect the data, a data scraper2 _{has been build using Import.io (www.import.io). Import.io is an} online application that allows for scraping data without the need for coding. Data is obtained from the Dutch lending-based crowdfunding platform Collin Crowdfund (www.collincrowdfund.nl). Collin Crowdfund is a leading crowdfunding platform in the Netherlands and has, to date, no crowdfunding campaigns that were not fully funded. For this reason, the success of a campaign (for this platform) cannot be obtained by a campaign either being funded or not. The success of the crowdfunding campaign is measured by the amount of time (speed) in which the campaign is funded. This makes the success of a campaign a continuous measurement instead of dichotomous when success is measured as either funded or not. Collin Crowdfund has all its previous crowdfunding campaigns listed on the website which allows for scraping the necessary data. Though all crowdfunding campaigns on Collin Crowdfund could be included in the sample, the choice has been made to include all campaigns from the point that campaigns got funded within one hour. The rationale behind this choice is that from this point in time it can be assumed that there is enough capital i n this particular marketplace to get the highest success measure. The effect of the lack of capital to fund crowdfunding campaigns through unfamiliarity with the marketplace is therefore assumed to be eliminated.

Sample

The first time a crowdfunding campaign was funded within one hour on the Collin Crowdfund platform was on the 17th _{of April 2015. The latest crowdfunding campaign in the dataset was fully} funded on the 26th _{of July 2016. The dataset includes campaigns that were funded within a}

timeframe of one year, three months and 9 days. This results in a sample size of N=184. One case is excluded from the sample because it is the only case that has a crowdfunding platform credit-score of “Excellent”. This is the highest crowdfunding platform credit-score on the Collin Crowdfund platform. All other cases have a score of either “Goed” (good), “Ruim voldoende” (ample) or “Voldoende” (sufficient). There are no cases that had the crowdfunding platform credit-score of “Matig” (moderate), the lowest crowdfunding platform credit-score. This results in an ordinal scale with three groups for the crowdfunding platform credit-scores in the sample. The final sample size that is used to test the hypotheses and the conceptual model consists of 183 cases, N=183.

2 _{A data scraper or web scraper is a computer software technique that helps transforming unstructured data}

from the world wide web into structured data on a local database. Once stored on a local database the data can be used for analysis and research purposes.

Variables

The model consists of four predictor variables and one outcome variable. The first predictor variable is the party credit-score. Collin Crowdfund uses the Dun & Bradstreet credit-score as a third-party credit-score. The Dun & Bradstreet credit-score is a commonly used by financial institutions in a lending decision and has four categories ranging from a score of 1 as the highest (best) credit -score and a score of 4 as the lowest score. Collin Crowdfund requires a Dun & Bradstreet creditscore of at least 3 for a company to apply for funding on the platform. A Dun & Bradstreet credit -score of 3 is reported as “Verhoogd risico” (increased risk), a credit--score of 2 as “Laag risico” (low risk) and a credit-score of 1 as “Minimaal risico” (minimal risk). The third-party credit-score is a categorical (ordinal) variable ranging from 1 to 3, where a score of 1 has a value label “Minimaal risico”, a score of 2 has a value label “Laag risico” and a score of 3 has a value label “Verhoogd risico”. The second predictor is the crowdfunding platform credit-score. Collin Crowdfund uses five different categories for its platform credit-score. As explained earlier both the highest category and lowest category are not occurring in the sample. The highest platform credit-score in the sample is “Goed” (good), the middle score “Ruim voldoende” (ample) and the lowest platform credit-score “Voldoende” (sufficient). The crowdfunding platform credit-credit-score is a categorical (ordinal) variable ranging from 1 to 3, where a score of 1 has a value label “Goed”, a score of 2 has a value label “Ruim voldoende” and a score of 3 has a value label “Voldoende”. Table 2 provides an overview of the descriptive statistics for these two ordinal variables.

Table 2. Descriptive statistics for the sample of crowdfunding cases, only categorical variables (N=183)

N % Minimum Maximum Mean SD

Third-party credit-score 1. “Minimaal risico”

(minimum risk) 2. “Laag risico” (low

risk)

3. “Hoog risico” (high risk) Crowdfunding platform credit-score 2. “Goed” (good) 3. “Ruim voldoende” (ample) 4. “Voldoende” (sufficient) 20 78 85 51 84 48 10.9 42.6 46.4 27.9 45.9 26.2 1.0 2.0 3.0 4.0 2.4 3.0 .7 .7

The third predictor is loan term. This is the time that companies borrow money for. All companies borrow with a linear payment schedule and start (re)paying principal and interest after one month and on a monthly basis. The loan term is consequently stated in months and for this reason, this variable is also measured in months. This variable is continuous with scores ranging from 8 months as the lowest score to 120 months as the highest score. The fourth predictor is interest rate. The interest rate is a percentage of the total amount that companies borrow through the platform. As companies start repaying on the principal after a month and on a monthly b asis the amount of interest paid per month declines during the repayment of the loan. The interest rate is a continuous variable that ranges from 6,5% to 9% on the Collin Crowdfund platform, which is in line with interest rates on loans on other crowdfunding platforms.

The outcome variable is the time in which a crowdfunding campaign on the Collin

Crowdfund platform is fully funded (speed of funding) and indicates the success of the crowdfunding campaign. Crowdfunding campaigns that are funded faster are considered as more successful compared to crowdfunding campaigns that took longer to be funded. Collin Crowdfund reports within a timeframe of hours: campaigns are either funded within an hour, within two hours (meaning between one hour and two hours), within three ours ( meaning between two hours and three hours), etc. The speed of funding is a continuous variable. In addition, the total amount of the loan (capital requirement) is added to the model as a control variable. In general , it will take more investors to complete the total amount. As it takes more time to have a larger group of investors informed and in the position to invest this is considered a necessary assumption. Table 2 provides an overview of the descriptive statistics. For the analyses, a z-score of the speed of funding is used. The next section will elaborate on the rationale behind using z-scores instead of the direct measure of speed of funding.

Table 3. Descriptive statistics for the sample of crowdfunding cases, only continuous variables (N=183)

Minimum Maximum Mean SD

Speed of funding (hours)

Z-score speed of funding (hours) Loan term (months)

1.0 -0.3 8.0 624.0 3.0 120.0 33.9 -0.2 53.3 100.2 0.7 11.0 Interest rate % 6.0 9.0 7.4 .6 Capital requirement (€x1000) 50.0 1500.0 187.8 159.2 Data changes

For the outcome variable, there is a large amount of cases that are funded within an hour. On the other side, a substantial group of cases deviates largely from the mean on the right side of the distribution, implying a skewness of the data caused by outliers. To make the data more applicable

for analysis a data transformation is, in this case, justified (Field, 2013). When transformations do not sufficiently reduce the skewness of the data, changing scores can be taken into consideration (Field, 2013). Because the data is heavily skewed the choice is made to convert back from a z-score. To do this, first, all scores are converted to z-scores for the outcome variable. As z-scores of 3.29 constitute an outlier, all z-scores higher than 3 are reduced to a z-score of the mean plus three standard deviations. Ideally, the data would not consist of too many outliers because this tends to violate the assumption of normally distributed errors (Field, 2013). When assumptions of parametric tests are violated it is harder to generalize conclusion to the total population of the sample .

Statistical approach

The nature of this research is quantitative. Quantitative research often includes a deductive approach. This research focuses on testing hypotheses. The research design is cross-sectional, meaning all different cases reflect a single point in time. This way a framework with quantifiable data is created that allows for an analysis of two or more variables (Field, 2013). The starting point will be a correlational analysis between all variables measured to investigate relationships. Because there will be one outcome variable and multiple predictors (both categorical and continuous) the statistical model will be tested with a multiple regression analysis. The categorical predictors are dummy coded so the categorical variables can be entered into the regression model. As a baseline group, for both variables the lowest credit-scores were assigned a 0. This implies that negative beta coefficients that are statistically significant for the dummy variables take fewer hours to get funded compared to the baseline group. Stated differently, negative beta coefficients that are statistically different for the dummy variables implies that the better credit-scores get funded faster. Finally, as the total amount of the loan is a control variable, it should be entered into the model first. In the second step, all predictors are entered into the model at the same time.

4. Results

Correlations

Because both credit-scores are ordinal variables Spearman’s correlation coefficient is used to test the relationships between variables. Another advantage is that it is a non-parametric statistic and can be used when the data has violated assumptions for parametric tests (e.g. non-normally distributed data) (Field, 2013). Table 4 shows the correlation matrix for all variables that will be put in the regression model.

Table 4. Spearman’s rho correlation matrix Speed of funding (z-score) Crowdfunding platform credit-score Third-party credit-score

Interest rate Loan term Capital requirement Speed of funding (z-score) .049 .122 -.122 .163* .399** Crowdfunding platform credit-score .282** .512** -.064 -.277** Third-party credit-score .344** -.139 -.194** Interest rate -.102 -.303** Loan term .063 Capital requirement **p < 0.01, *p < 0.05

There are several significant results reported in the correlation matrix. For the predictor variables there is only one significant relationship with the outcome variable. There is a significant positive relationship between the speed of funding and the loan term meaning longer loan terms take longer to fund, rs = .163, p < .05. This is in line with the third hypothesis of this research H3: Loan terms are a predictor for the speed of funding. There is also a stronger, more significant and positive

relationship between the speed of funding and the capital requirement, rs = .399, p < .01. This invigorates the decision to take the capital requirement into account as a control variable.

Furthermore, there is a positive, significant relationship between the credit-score, rs = .282, p < .01, and a positive, significant relationship between crowdfunding platform credit-score, rs = .512, p < .01. As a higher credit-score implicates higher risk this is in line with the theory. There is a negative, significant relationship between crowdfunding platform credit-score and the capital requirement, rs = -.277, p < .01, suggesting that companies with a higher capital requirement have less risk.

Also the third-party credit-score has a positive, significant relationship with the interest rate, rs = .344, p < .01. In line with the crowdfunding platform credit-score, a high score implicates a riskier project and interest rates seem to be higher. Again, there is a negative, significant relationship between the credit-score and capital requirement, rs = -.194, p < .01. Although this relationship is smaller compared to the crowdfunding platform credit-score this again implies that companies with a higher capital requirement are considered to have less risk. In conclusion, interest rate has

negative, significant relationship with capital requirement, rs = -.303, p < .01, indicating that companies with a higher capital requirement borrow at cheaper interest rates.

Multiple regression

After reporting the results of the correlation matrix, the conceptual model can be tested using multiple regression as a statistical method. As mentioned before, capital requirement is a control variable and is entered into the model first. After entering all other variables into the model the model significantly predicts 12.1% of the variability in the outcome variable. The second model predicts 10.5% more of the variability compared to the first model. The adjusted R2 _{accounts for a} shrinkage of 3.5% if the model were taken from the population. The Durbin-Watson statistic lies very close to 2, indicating independent errors. All assumptions for regression were met although the assumption of normality, after changing the data to z-scores and the outliers to 3 standard deviations, could still be questioned. The histogram of the standardized residuals shows some positive skewness and the P-P plot deviates slightly from the linear line. Table 5 shows the regression coefficients for all variables. There are only two significant beta values suggesting that all but one of the hypotheses should be rejected. Interesting enough, capital requirement does not significantly predict speed of funding in the first step but does significantly predict speed of funding in the next step of the model, assuming some interaction effects.

Table 5. Regression coefficients

Variable B SE B β

Step 1

Capital requirement 5.204E-7 .000 .126

Step 2

Capital requirement 6.900E-7 .000 .166*

Interest rate -.109 .107 -.091 Loan term .007 .004 .123 TPCS_low_vs_high -.402 .107 -.302** TPCS_minimum_vs_high -.253 .162 -.120 CPCS_good_vs_sufficient -.181 .151 -.123 CPCS_ample_vs_sufficient -.175 .116 -.132

Note. R2 _{= .016 for Step 1; ΔR}2 _{= .105 for Step 2 (p < .002).}

The main finding is that there is a highly significant, negative effect of TPCS_low_vs_high on the speed of funding. This means there is a significant difference between these two credi t-scores. TPCS_low_vs_high is the dummy code for third-party score “low risk” vs. third-party credit-score “high risk”. The negative effect means “low risk” companies get funded faster compared to “high risk” companies. This partially supports H1: Third-party credit-scores are a predictor for the speed of the funding. When comparing “minimum risk” companies to “high risk” companies there is no significant relationship between the dummy variable and the speed of funding. It should also be mentioned that both the dummy variables for crowdfunding platform credit-score were close to significant.

5. Discussion

To elaborate further on the last statement in the result section, the interesting question arises why companies with third-party credit-scores that indicates “minimum risk” do not get funded faster compared to “high risk” companies while “low risk” companies do get funded faster. When looking at the descriptive statistics for the categorical variables (credit-scores) there are only 20 cases (N=20) for the third party credit-score “minimum risk”. This fairly low number of scores could potentially bias the data. The sample of both other groups of the third-party credit-score is around four times as high (“low risk” N=78, “high risk” N=85). Once there is a more substantial sample it would be

interesting to test again and see if the effects are still the same.

Almost all of the hypotheses in this research are not supported, actually, just one is semi-supported. A potential explanation for this problem is the choice of crowdfunding platform and the way the outcome variable was measured. Collin Crowdfund reports all their successful campaigns as being funded within a time-frame. This is either within hours and, once it is more than a day, it becomes within days. Because a standard was needed and most cases get funded within one day the choice was made to convert days to hours. Projects listed on website as funded within 2 days were converted to 48 hours but actually fall within a timeframe of 25-48 hours, explaining a lot of hidden variation. Another problem with the measurement goes one step further. A substantial amount of projects gets funded within one hour, causing the data to cluster around one. This causes problems when data transformation techniques are necessary to imply for the assumption of normality (Field, 2013). If future research would be conducted on data on the Collin Crowdfund platform it is strongly recommended to measure the time in which a loan is fully subscribed in minutes rather than hours to prevent biased data. From the experience of the researcher companies often get funded within minutes through this Crowdfunding Platform.

Though the conceptual model should be rejected based on this research, the rationale described in the previous paragraph could be a potential explanation why there were no significant effects. In addition, the effects of the dummy variables for the crowdfunding platform credit -score were close to significance. Still, it possible that the predictors in the model are not appropriate. For the credit-scores it is harder to explain why it has no significant effect on the success of a

crowdfunding campaign. As a successful campaign boils down to investors engaging in making a lending decision it is quite clear from the literature that a risk-assessment on the entity borrowing money to plays an essential role. While a credit-score essentially is a risk-assessment it should make a contribution to the investment decision (Kalberg and Udell, 2003; Cowan & Cowan,2006; Cassar, Ittner & Cavaluzzo, 2015). The positive correlation between the crowdfunding platform credit-score and the third-party credit-score is an indicator of the efforts made by different institutions to provide people with quality information and using (some of) the same techniques.

From the correlation matrix, quite a few interesting relationships came forward. First of all, the loan term correlates with the speed of funding but there is no direct effect. There might be a potential interaction effect within the model. It would be interesting to further investigate the effect of the loan term and the crowdfunding investment decision. From the literature is evident there is a relationship between the loan term and the investment decision for other financial institutions (Flannery, 1986; Berger et al., 2005; Ortiz-Molina & Penas, 2008; Kirschermann & Norden, 2012). There is a positive relation between both credit-scores and the interest rate. Higher risk companies borrow at higher rates in line with Flannery’s (1986) research and the economic principle that there should be a trade-off between risk and return. The relationship between the credit-score and interest, under the assumption there is also a relationship between the interest rate and the lending decision, could be explained by a mediation effect. If businesses borrow at an interest rate in line with their credit-score, the effect of the interest rate should be mediated through the effect of the credit-score. While this would be an interesting research topic, there is a statistical restriction in using categorical variables as mediators. The credit-scores should have a continuous level of measurement. Graydon’s PD-range as a percentage of the probability of default could be an option.

Also evident from the correlation matrix is the negative si gnificant relationships between the credit-scores and the capital requirement and the interest rate and the capital requirement. This suggests that lower risk companies have higher capital requirements and borrow at lower interest rates. A possible explanation could be that more mature and larger companies have larger capital requirements and bare less risk compared to smaller starting companies. Ortiz-Molina & Penas (2008) control for age and size within the loan term-risk relationship and find a positive relationship. This could be an indicator that there is also a relationship between age and size within capital requirement-risk relationship.

One predictor of crowdfunding success that was not included in the model but potentially has a significant effect on successful campaigns is collateral available to investors ( Berger & Udell, 1998; Mason & Stark, 2004). The reason collateral was not taken into account has a more practical nature. The collateral available to investors is usually described in the text and could not be scraped from the Collin Crowdfund website. The availability of collateral and the value it represents is an important topic for research. Because a company lending through crowdfunding essentially has a lot of contracts with different investors it is hard to resolve issues around collateral when companies default.

From the academic literature on crowdfunding it is clear that there are potential problems related to information asymmetries Agrawal et al., 2013; Ahlers et al., 2012; Belleflamme et al., 2013. The main concern of this research is how investors deal with information asymmetries. Quality

information is important to overcome information asymmetry Stiglitz (2000). This is supported by Ahlers et al. (2015) within the context of crowdfunding and Degryse & van Cayseele (2000) within the context of lending. Healy & Palepu (2001) described the different ways information flows in an investment environment stressing the importance of third-party information intermediaries. Although no significant effect on the speed of funding for both credit-scores was found, it is still strongly recommended to further investigate this relation. Crowdfunding is still in its infancy and it takes time to become clear what risks are associated with crowdfunding. While investing is simple and has no hard limitations it is important that people should be aware of the potential risks. The relationship between the third-party credit-scores is also important when looking at the business model of the crowdfunding platforms. Revenues comes from success fees on funded campaigns giving the platforms short-term incentives to push as many projects as possible through a platform, even if these projects bare (too much) risk. Platforms focused on short-term profits could form a potential danger for crowdfunding, especially because most platforms advice funders in their risk-assessment with their own credit-score.

6. Conclusion

This research was concerned with the effect of information signals on crowdfunding success in lending-based crowdfunding for businesses. Data was collected from Dutch crowdfunding platform Collin Crowdfund. The nature of this research is quantitative. To test the hypotheses Spearman’s correlation was used in combination with multiple regression. Third-party credit-scores,

crowdfunding platform credit-scores, interest rate and loan term were used as predictors in the regression model, while the capital requirement was used as a control variable. The speed of funding was used as a proxy for crowdfunding success. Because both credit-scores were categorical

measures, the hypotheses for these predictors can only be fully supported when all dummy variables show a significant effect.

Based on the regression analysis, there was a significant effect when comparing “low risk” to “high risk” but no significant effect comparing “minimum risk” to “high risk”. H1: Third-party scores are a predictor for the speed of the funding is partially supported. Where “low risk” credit-scores are funded faster than “high risk” credit-credit-scores. There was no significant effect for the dummy variables created for the crowdfunding platform credit-score on the speed of funding. H2: Crowdfunding platform credit-scores are a predictor for the speed of funding was not supported. There was no significant effect of the loan term on the speed of funding. H3: Loan terms are a predictor for the speed of funding was not supported. There was no significant effect of the interest rate on the speed of funding. H4: Interest rates are a predictor for the speed of funding was not supported.

A potential explanation for not fully supporting the H1 could be the small sample size of the “minimal risk” group. A potential explanation for not supporting all the hypotheses could be the level of measurement of the outcome variable speed of funding. Instead of hours, this should be in minutes to get a more accurate estimate and allow for useful data transformation techniques. There were several significant relationships between variables. Notifiable relationships exist between both credit-scores and the interest rate suggesting that riskier loans have higher interest rates.

Considering the credit-score should be a reflection of the risk-return ratio they can be mediators for the effect of interest on crowdfunding success. Another notifiable re lationship are the ones between both credit-scores and the capital requirement, implying larger capital requirements are concerned with less risk.

As crowdfunding is still a young phenomenon it is important to engage in future research. Suggestions for future research include measuring the effect of credit-scores using an improved data sample, data from another platform or data from several platforms. Including collateral as a

predictor could potentially improve the model for crowdfunding success. Credit-scores are potential mediators for the effect of the interest rate.

7. References

Andy Field. (2013). Discovering statistics using SPSS. Sage publications.

Agrawal, A. K., Catalini, C., & Goldfarb, A. (2011). The geography of crowdfunding (No. w16820). National bureau of economic research.

Agrawal, A. K., Catalini, C., & Goldfarb, A. (2013). Some simple economics of crowdfunding (No. w19133). National Bureau of Economic Research.

Ahlers, G. K., Cumming, D., Günther, C., & Schweizer, D. (2015). Signaling in equity crowdfunding. Available at SSRN, 2161587.

Akerlof, G. (1970). The market for lemons: Quality uncertainty and the market mechanism. Quarterly Journal of Economics, 84(3), 488-500.

Belleflamme, P., Lambert, T., & Schwienbacher, A. (2013). Crowdfunding: Tapping the right crowd. Journal of Business Venturing.

Berger, A. N., Espinosa-Vega, M. A., Frame, W. S., & Miller, N. H. (2005). Debt maturity, risk, and asymmetric information. The Journal of Finance, 60(6), 2895-2923.

Berger, A. N., & Frame, W. S. (2007). Small business credit scoring and credit availability. Journal of small business management, 45(1), 5-22.

Berger, A. N., & Udell, G. F. (1998). The economics of small business finance: The roles of private equity and debt markets in the financial growth cycle. Journal of banking & finance, 22(6), 613-673.

Binks, M. R., & Ennew, C. T. (1997). The relationship between UK banks and their small business customers. Small Business Economics, 9(2), 167-178.

Bradford, C. S. (2012). Crowdfunding and the federal securities laws. Columbia Business Law Review, 2012(1).

Burtch, G., Ghose, A., & Wattal, S. (2013). An empirical examination of the antecedents and consequences of contribution patterns in crowd-funded markets. Information Systems Research, 24(3), 499-519.

Cassar, G., Ittner, C. D., & Cavalluzzo, K. S. (2015). Alternative information sources and information asymmetry reduction: Evidence from small business debt. Journal of Accounting and

Economics, 59(2), 242-263.

Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling theory: A review and assessment. Journal of Management, 37(1), 39-67.

Cowan, C. D., & Cowan, A. M. (2006). A survey based assessment of financial institution use of credit scoring for small business lending. SBA Office of Advocacy.

Degryse, H., & Van Cayseele, P. (2000). Relationship lending within a bank-based system: Evidence from European small business data. Journal of financial Intermediation, 9(1), 90-109.

Collincrowdfund.nl (2016). Risico’s, Collin Credit Score. Retrieved July 15th, 2016, from https://www.collincrowdfund.nl/risicos/

Cumming, D. J., & Johan, S. (2016). Crowdfunding and Entrepreneurial Internationalization. In N. Dai and D. Siegel, eds. Entrepreneurial Finance: Managerial and Policy Implications. The World Scientific Publishers. Chapter 5. Forthcoming.

Crowdsourcing.org (2016). Global Crowdfunding Market to reach 34.4B in 2015, predicts Massolution’s 2015CF Industry Report. Retrieved May 24th, 2016, from

http://www.crowdsourcing.org/editorial/globalcrowdfunding-market-to-reach-344b-in-2015- predicts-massolutions-2015cf-industry-report/45376

Ferrary, M., & Granovetter, M. (2009). The role of venture capital firms in Silicon Valley's complex innovation network. Economy and Society, 38(2), 326-359.

Flannery, M. J. (1986). Asymmetric information and risky debt maturity choice. The Journal of Finance, 41(1), 19-37.

Geldvoorelkaar.nl (2016). Risico, “Met afloscapaciteit wordt het bedrag bedoeld dat een geldnemer maandelijks kan inzetten voor aflossing én rentebetaling”. Retrieved July 15th, 2016, from

https://geldvoorelkaar.nl/geld-investeren/risico.aspx

Healy, P. M., & Palepu, K. G. (2001). Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature. Journal of accounting and economics, 31(1), 405-440.

Kallberg, J. G., & Udell, G. F. (2003). The value of private sector business credit information sharing: The US case. Journal of Banking & Finance, 27(3), 449-469.

Kirschenmann, K., & Norden, L. (2012). The relationship between borrower risk and loan maturity in small business lending. Journal of Business Finance & Accounting, 39(5‐6), 730-757.

Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of financial economics,3(4), 305-360.

Mason, C., & Stark, M. (2004). What do investors look for in a business plan? A comparison of the investment criteria of bankers, venture capitalists and business angels. International Small Business Journal, 22(3), 227-248.

Mitra, D. (2012). The role of crowdfunding in entrepreneurial finance. Delhi Business Review, 13(2), 67.

Mollick, E. (2014). The dynamics of crowdfunding: An exploratory study. Journal of Business Venturing, 29(1), 1-16.

Ortiz-Molina, H., & Penas, M. F. (2008). Lending to small businesses: The role of loan maturity in addressing information problems. Small Business Economics, 30(4), 361-383.

Petersen, M.A. 2004. Information: Hard and soft. Working paper, Northwestern University.

Poetz, M. K., & Schreier, M. (2012). The value of crowdsourcing: can users really compete with professionals in generating new product ideas? Journal of Product Innovation Management, 29(2), 245-256.

Schwienbacher, A., & Larralde, B. (2010). Crowdfunding of small entrepreneurial ventures. Handbook of entrepreneurial finance, Oxford University Press, Forthcoming.

Stiglitz, J. E. (2000). The contributions of the economics of information to twentieth century economics. Quarterly Journal of economics, 1441-1478.