Amsterdam Business School
Agency costs at crowdfunded companies
Name: Viacheslav N. Konizhevskiy Student number: 10902139
Thesis supervisor: prof. dr. L.R.T. van der Goot Date: 28 January 2017
Word count: 13,711
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by student Viacheslav N. Konizhevskiy who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is 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
This paper examines the manner in which the agency costs emerge at equity crowdfunded companies. This novel way of financing an enterprise gets more attention of entrepreneurs every year allowing them to access the necessary capital faster, easier and cheaper in control sense than more conventional types of financing can offer. As crowdfunding attracts more entrepreneurs and investors, the research on this development is still scarce. While studying the ways crowdfunding projects are initiated and managed, scholars are yet to examine consequences of this financing method after these projects acquired the necessary funding. This paper contributes to the existing literature by testing the agency cost hypothesis in the new crowdfunding setting using a sample of Dutch equity crowdfunded companies and adding the wisdom-of-the-crowd effect to the model. I find no support for the conventional agency costs hypothesis, stating that agency costs are positively associated with percentage of equity issued to new shareholders. I do find evidence of the wisdom-of-the-crowd effect, meaning that the number of crowdfunders is negatively associated with agency costs. There is also evidence that agency costs are positively associated with crowdfunding target. To the best of my knowledge, this paper is the first that examines the possible effects of crowdfunding for a company after the funding has been raised.
Contents
1 Introduction ... 5
2 Literature review ... 7
2.1 Agency costs ... 7
2.2 Crowdfunding ... 8
2.3 Agency costs at crowdfunded companies ... 10
2.4 Hypothesis development ... 12 3 Methodology ... 14 3.1 Agency Costs ... 15 3.2 Crowdfunders ... 17 3.3 Controls ... 17 4 Data ... 19 4.1 Zero-agency baseline ... 19 4.2 Resulting data ... 20 5 Empirical Results ... 24
5.1 Equity and Agency Costs ... 28
5.2 Wisdom-of-the-crowd ... 30 5.3 Controls ... 31 5.4 Robustness checks ... 32 6 Conclusion ... 33 Acknowledgments... 36 References ... 37
1 Introduction
Organizations’ managers continuously seek the ways to push their companies’ activities to a (even greater) success. As Bloom & Van Reenen (2010) show in their research the decisions made by the management in the progress do matter: they impact the performance level of the organization as a whole. According to these authors, the decisions on the matters of monitoring, setting targets and providing incentives influence the organizational results.
Among such decisions, the management is responsible for finding the ways to finance company activities. The managers-owners are already making this type of decisions at the company’s foundation moment. Until recently, companies could have been financed, among others, by own private capital of the owners, getting a loan by a bank or acquiring the required investment in exchange for a percentage of equity (Cosh, Cumming, & Hughes, 2009). Some of these financing decisions could cause the ownership structure to change, as it is the case with equity financing. The change of the ownership structure may lead to various impacts at different levels of the organization. The rise of the number of owners may create an agency problem in cases when one of the owners continues to manage the company (Jensen & Meckling, 1976). Other new owners experience this problem by the rise of the agency costs. Some studies report on the similar effect as a company assumes more debt (Berger & Bonaccorsi di Patti, 2006; Margaritis & Psillaki, 2010). As a result, the capital providers are getting less return on their investment.
The rise of agency costs has been extensively studied in different settings. This study contributes to the existing literature by examining this matter at equity crowdfunded companies. Crowdfunding has become a popular financing method in the more recent years. Now entrepreneurs are able to get the required investment from a broad and heterogeneous public using services provided by specialized platforms (Belleflamme, Lambert, & Schwienbacher, 2014). This way of acquiring capital is particularly popular among start-up entrepreneurs due to its low barriers: there is mostly no need for a credit rating, surety or a track record which would mostly be required if one would consider getting a loan at a bank.
The popularity of crowdfunding is rising from the moment this financing method has been introduced. Dutch research agency Douw & Koren reports that various initiatives have acquired 128 million euro’s using this financing method in the Netherlands in 2015 (Hupkens, n.d.). Remarkably, this figure is twice as high as it has been one year earlier.
There are multiple methods to get the required capital using crowdfunding. These methods differ from one another in the form of service that would be required from the entrepreneur in exchange
crowdfunding whereby the capital is being provided in exchange for percentage of equity or other forms of future profit participation (Belleflamme et al., 2014). Consequently, the ownership structure of a company changes. Therefore, I expect the emergence of agency costs at companies using this crowdfunding method, which would be positively related to the percentage of equity acquired by investors (crowdfunders). I also test if this relationship is moderated by the wisdom-of-the-crowd effect; that is the shared knowledge accumulated by the crowd that possibly lowers information asymmetry as the crowd is getting bigger, allowing crowdfunders to identify the projects more aligned with their goals.
Despite the extensive research on the agency theory and the agency costs hypothesis, this study is, to the best of my knowledge, the first that examines these issues in a crowdfunding setting as suggested by Huynh (2015). Furthermore, the academic research on crowdfunding as a financing method is still scarce. The researchers studying this method mostly focus on the way the crowdfunding campaigns are being managed and on the aspects which do or do not lead to getting the required investment using crowdfunding (Belleflamme, Lambert, & Schwienbacher, 2013; Cordova, Dolci, & Gianfrate, 2015). However, at the time of this study I was not able to find any academic research, which would examine the possible impacts of this type of financing on the organization after the investment has been secured. I address the main question of this research by combining and analyzing crowdfunding projects data and financial records of the companies, which have acquired capital, using equity crowdfunding.
The remainder of this paper is organized as follows. In Section 2 I provide the literature review on the topics of agency costs and crowdfunding as well as the hypotheses development. In Section 3 I describe the methodology, which I use to test the hypotheses. Section 4 presents the data, which I use in this research and Section 5 the empiric results from testing the hypotheses using this data. Finally, Section 6 concludes the paper.
2 Literature review
Agency costs have been extensively studied in various settings. Scholars used different samples containing information on different industries to test possible relationships between agency costs and various variables. Crowdfunding companies have been studied much less as this method of obtaining external capital is still relatively new. The following provides a literature review on both subjects as far as it is relevant for the current research.
2.1 Agency costs
In the second half of the past century, scholars developed the agency theory to explain the relationships present in organizations. This theory states that an agency relationship exists when the agent is engaged by the principal to perform certain tasks while the latter delegates decision-making authority (Hill & Jones, 1992). The interests of the principal and the agent are not completely aligned and the principal does not possess all the information about agent’s actions creating the need for contracts, which limit the actions of the latter (Bosse & Phillips, 2016). The incomplete alignment of principal’s and agent’s interests will in most cases cause the emergence of the agency costs. Jensen & Meckling (1976) argue that as the principal tries to ensure that the agent acts as closely as possible to the principal’s best interest the organization of which they are parts will unavoidably incur at least one of the following types of costs: (a) monitoring expenditures, (b) bonding expenditures, and (c) residual loss. The principal incurs the monitoring expenditures as he tries to monitor and/or control the activities of the agent. The agent might incur the bonding costs, which are associated with certain guarantees for the principal, as compensation for taking inappropriate actions. The residual loss is defined as the reduction in possible welfare of the principal due to the actions taken by the agent, which were not motivated by the best interest of the principal. Jensen & Meckling (1976) define agency costs as the sum of these types of costs and argue that these are inherent to the agency relationships in which both principal and agent are utility maximizers. These leads the authors to the agency costs of outside equity proposition which states that the agency costs are expected to emerge in situations where new shareholders are introduced at a company leaving the manager-founder with a less percentage of equity. The authors argue that the manager’s utility is defined by both intangible as monetary benefits and the emergence of the new shareholders reduces the latter. Being the utility maximizer the manager-founder will compensate this loss by other available means using the company resources for this purpose, leading to agency costs. The authors further expect the market to be aware of such effects and the minority shareholders to consider the possible agency costs while
Ang, Cole, & Lin (2000) test the agency costs hypothesis as defined by Jensen & Meckling (1976) by investigating a sample of small business firms in the United Sates. By measuring the agency costs as the efficiency ratios (operating-expense-to-sales and sales-to-total-assets), the authors study various types of ownership structure and their relationship with such costs. Their findings are consistent with the effects predicted by Jensen & Meckling (1976): the firms with equity distributed among various shareholders experience higher agency cost than those with a sole owner-manager. They also find that as the equity share of non-managing shareholders gets higher, so do the agency costs and this effect is observable among all types of tested industries, company sizes and capital structure differences. The authors find the same association between the agency costs and the number of non-managing shareholders due to the free-rider problem: some of the shareholders may chose not to monitor the owner-manager, decreasing the total monitoring expense and increasing what Jensen & Meckling (1976) call the residual loss.
Building on the agency theory Ross (2013) argues that the entrepreneurial and managerial functions should be separated to avoid conflicts of interest and to lower the level of agency costs. The entrepreneurial function includes strategic decisions while the managerial function focuses on daily operations. If the principal has only limited separated view on the outcomes of the entrepreneurial and managerial functions, the potential of opportunism by the agent increases and so do the associated agency costs.
While all the previously discussed studies consider the association of the factual or legal ownership with agency costs, Sieger, Zellweger, & Aquino (2013) show that the psychological ownership can be comparable to factual ownership in the terms of company performance. When senior managers consider the company as their own, this consideration aligns their interests with those of the principal. As the various interests get more aligned, the need for monitoring declines resulting in lower agency costs. In those cases, the monitoring is found to be contra productive as it reduces the feeling of ownership by the agents.
2.2 Crowdfunding
As of the beginning of twenty-first century, entrepreneurs had various types of financing potentially available to them. Besides from investing their own time and money and that of their close relatives, entrepreneurs could seek an investment by, among other, the banks, venture capitalists or angel investors (Cosh et al., 2009). However, it could take much more time and sources than expected in order to acquire the investment, especially for small firms. According to Cosh et al. (2009), banks prefer investment ideas to be profitable, less risky and backed-up by assets while venture capitalists are more likely to invest in innovative opportunities. This latter type
of investors could also provide entrepreneurs with managerial expertise so that the future profitability is secured (Bengtsson, 2011). In return for both the investment and expertise an entrepreneur would have to give up a certain percentage of his firm’s equity. In those cases, the choice to seek outside finance ultimately leads to the ownership structure change.
The availability of various types of investors did not prevent many entrepreneurs’ ideas from remaining unfunded (Carpenter & Petersen, 2002). While not being able to insure the traditional investors of either the innovative power of their ideas or being backed-up sufficiently, the entrepreneurs started to explore the new opportunities for investment sources: the crowd.
The crowd, being the general public, has gained the attention of various companies at the beginning of the twenty-first century as the Internet penetrated deeper into the households (Huynh, 2015). Instead of remaining customers, consumers engaged in co-creation process of products and started to help companies to solve problems (Kleemann, Voß, & Rieder, 2008). This way outsourcing gained a new form, referred to as crowdsourcing, whereby a company placed an open call on the general large group of people to perform tasks, which were in their essence a part of its own employees’ job. According to Kleemann et al. (2008), crowdsourcing allowed companies to reduce costs of interaction with customers, to use their resources in a more efficient manner and even to increase their turnover. Moreover, firms integrated users of their products as partners in value creation process while employing their knowledge and subsequently increasing the quality of the products. In some cases the engagement of the crowd while developing new products lead to more feasible ideas than when the product is fully designed by hired professionals (Poetz & Schreier, 2012).
The idea of seeking investment at the general heterogeneous public, or crowdfunding, rather than at professional investors is very similar in its nature to that of crowdsourcing. According to Belleflamme et al. (2014), crowdfunding is an open call for provision of financial resources which mostly takes place through the Internet. Entrepreneurs seeking an investment are referred to as crowdfounders and the investors as crowdfunders1. Crowdfunding platforms, specialized websites,
which facilitate the crowdfunding process, allow crowdfounders to initiate a fundraising project, engage with potential investors and provide all the necessary information. The crowdfunding platforms collect the investments made by crowdfunders until the required amount has been reached (Wieck, Bretschneider, & Leimeister, 2013). At that moment, the crowdfounders obtain their investment and obligations towards the crowdfunders.
In some cases, there are no obligations towards the crowdfunders in exchange for the capital they provide. This happens when the crowdfunding project has been initiated using the donation method (Schwienbacher & Larralde, 2010). In those cases, crowdfunders are not being driven by economic reasons to invest into the project (Mollick, 2014). Their motivation is of a more humanitarian nature, similar to the motives of those giving to charity. Artists commonly use this crowdfunding method to finance their art projects (Agrawal, Catalini, & Goldfarb, 2015).
Reward-based crowdfunding is another crowdfunding method, which might not provide financial benefits to the investors (Schwienbacher & Larralde, 2010). Although the latter expect to receive some sort of reward, it might as well be of non-monetary or even intangible nature. As an illustration, producers might employ this method to attract additional financing for movies they are working on and promise to mention the crowdfunders’ names in closing credits (Mollick, 2014). Crowdfounders might also employ this method in order to attract customers for the products, which are still at the development stage. This way the crowdfunders practically pay upfront for the product they will receive although the price is generally lower than it is at the moment the product becomes officially introduced to the general market (Belleflamme et al., 2014).
Crowdfunders who seek a financial return on their capital mostly invest using lending and equity methods (Ordanini, A., Miceli, L., Pizzetti, M., & Parasuraman, A., 2011). These two methods are in their essence similar to the more conventional investments by debt or equity: an outsider provides a company the required capital and expects to receive return on this provided capital. In case of the equity method, providers of capital acquire equity shares and become company’s shareholders (Belleflamme et al., 2014).
2.3 Agency costs at crowdfunded companies
As the discussed literature on crowdfunding highlights, this financing method might change the company ownership structure making the crowdfunders the new shareholders. This change happens when the company employs the equity method issuing its shares. According to Jensen & Meckling (1976) and empirical evidence provided by later studies, which I have discussed above, such a change has implications on the emergence of agency costs when company utilizes more conventional financing methods. Jensen & Meckling (1976) state that the agency costs are unavoidable when agency relationship is present and exist due to the information asymmetry between the manager and the shareholders and the lowered incentive for the manager to put an effort in achieving the best results for the company and consequently for its shareholders. Previous research on crowdfunding has not addressed issues related to the agency theory in general or the
agency costs hypothesis in particular (Huynh, 2015). As the core of the agency problems lays at the divergence of agent’s interests from those of the principal (Hill & Jones, 1992), this section will highlight the literature on the motives of crowdfounders and those of crowdfunders to determine the possibilities for agency costs to occur.
According to Mollick (2014), one cannot attribute a single motive of crowdfunders to a specific crowdfunding method. Crowdfunders providing capital as a loan might be as much motivated by their believes in the general good which can be accomplished by the project as by the expected financial return on their investment. However, the crowdfunders investing in a particular project expect it to succeed independent of the crowdfunding method used or wherever the project ultimately has a for-profit goal (Mollick, 2014). This expectation is driven by both intrinsic and extrinsic motives as people investing in a non-profit goal get emotional reward from ‘doing the right thing’ (Van Wingerden & Ryan, 2011). At the same time specific crowdfunding platforms attract different types of investors giving rise to the expectation of a monetary goal while choosing for platform facilitating the equity crowdfunding (Ordanini, A. et al., 2011).
Gerber & Hui (2013) study the motivations of crowdfounders to use this financing method. Aside from raising the funds to finance their project in general, crowdfounders seem to take the shorter funding time into account as opposed to conventional financing methods. However, the most remarkable finding is the wish to remain control over the company. Gerber & Hui (2013) explain that crowdfounders perceive this financing method as an opportunity to continue running the business the way they want while the conventional financing methods would force them to take the preferences of the investors into account. As venture capitalists may limit the freedom of the entrepreneur in order to minimize risks and agency costs using various methods including explicit contracts, this perception does seem reasonable (Bengtsson, 2011): equity crowdfunding does not require any additional contracts by default. It might not even allow a possibility of negotiating a price per share if crowdfunders would consider a project to be more risky or expect higher agency costs. It is also consistent with the general independence expectation by entrepreneurs when founding their own company (Estay, Durrieu, & Akhter, 2013). This perception implies that crowdfounders do not expect any significant changes in the way they run their companies after crowdfunding has taken place. While not being bounded by the presence of outsiders in the management of the company they will also be able to make decisions as they see fit. As the crowdfounders retain less equity as the result of the equity crowdfunding and do not perceive any boundaries for their actions, the maximization of their utility by using company’s resources in their sole interest will be most likely to occur (Jensen & Meckling, 1976).
The magnitude of this utilization will depend on the number of factors, including the motivation of the crowdfounder to be entrepreneur in the first place. Scott Morton & Podolny (2002) distinguish utility-maximizers and profit-maximizers while looking at the motivation of entrepreneurs. The latter are expected to value higher profits over other, non-financial perquisites while running a business. They are also more likely to continue maximizing profits even while keeping less equity, creating less agency costs.
The motivation of profit-maximizing crowdfounders is more aligned with that of professional investors as the latter expect return on the invested capital due to higher profits (Ordanini, A. et al., 2011). The projects initiated by this type of crowdfounders are therefore of high quality for capital providers. As professional investors have the necessary knowledge and experience, they are able to identify this high quality projects while choosing among investment opportunities on crowdfunding platforms (Mollick, 2014). Other crowdfunders, who do not have professional knowledge or expertise, build their perception of project quality based on investments made before them. Kim & Viswanathan (2014) report that even though the majority of the crowd does not possess any knowledge that should assist them in identifying high quality investments ex-ante, it still manages to do so accumulating this information through one another’s behavior. The investment decisions of professional investors have thereby a leading part. Schwienbacher & Larralde (2010) add that crowdfunders might not possess investment knowledge but they often know the industry of the project well. This knowledge might help them to distinguish feasible investment opportunities better than a professional investor would do. They also take into consideration the information, which would not be considered relevant by professionals, such as crowdfounder’s social networks’ accounts (Agrawal et al., 2015). While gathering the information from different sources, the crowd creates the wisdom of the crowd effect that lowers the overall information asymmetry.
2.4 Hypothesis development
In sum, the literature, which I have discussed above, produces the following input for this research. Companies with one or more owner-managers are expected to incur agency costs. This effect is considered inherent to the nature of agency contracts and will be present at any company, which incorporates the principle-agent relationship. The magnitude of agency costs will differ depending on the motives of owner-manager and information asymmetry between agency counterparts. Not taking the possible differences into account, the relationship between ownership structure and agency costs is straightforward: as the percentage of equity of new shareholders rises, so do the agency costs. Crowdfunding, which is a relatively new method of financing a company, can change
the ownership structure making the founder-manager to keep less equity and a less share of future profits. The agency costs hypothesis and empirical evidence in various settings suggest that in such a case the percentage of agency costs will rise as the percentage of crowdfunded equity rises, thus, the higher the percentage of equity issued, the higher the agency costs. Therefore, the first hypothesis states:
H1: Agency costs are positively associated with the issued percentage of equity to the new owners at crowdfunded companies.
This relationship might be altered by the wisdom-of-the-crowd effect, which lowers information asymmetry, at least at the crowdfunding project stage. The literature suggests that crowdfunders accumulate the necessary knowledge to distinguish high quality projects. The crowdfunders lacking investment experience follow the lead of professional investors. As the crowd gets bigger, more information gets available which helps identify the best investment opportunities. The latter partially depends on the motives of the crowdfounder and their alignment with the crowdfunders’ motives. This directly affects the height of agency costs to be expected as a company proceeds after acquiring external capital by reducing their percentage as the number of crowdfunders gets higher. Thus, the second hypothesis states:
H2: Agency costs are negatively associated with the number of crowdfunders.
I still expect the percentage of agency costs to be higher when higher percentage of equity is being issued to crowdfunders, even with the wisdom-of-the-crowd-effect present. However, the number of crowdfunders is expected to have a moderating effect on the equity-agency-costs relationship. I expect that when the number of crowdfunders is high, the higher equity percentage will lead to lower percentage of agency costs than when the number of crowdfunders is low. As the number of crowdfunders gets lower, I expect the association between equity percentage and agency costs to become stronger. Therefore, the third hypothesis states:
H3: The association between crowdfunded equity and agency costs will be lower when the number of crowdfunders is high than when it is low.
Now that the hypotheses have being formulated, the next section describes the methods used to test them.
3 Methodology
In this section follows the description of the methodology used in this research. In order to test the hypothesis, I use a sample of Dutch firms, which were crowdfunded using equity method. The information about the crowdfunding projects has been requested and provided by Symbid, a single crowdfunding platform in the Netherlands, which facilitates crowdfunding projects allowing direct offering of equity shares to crowdfunders. Symbid provided a list of successfully funded projects starting from December 2011 to October 2016. This list contains 133 projects stating the project’s title, targeted and acquired amounts in euro’s, percentage of equity offered and date of reaching the target funding. As this research requires financial information on a company after it has been crowdfunded and there is no such information available for the current year, all projects crowdfunded in 2016 were left out. The same applied to the projects crowdfunded starting July 1, 2015: while acknowledging the arbitrary nature of the chosen date, I expect the agency costs to take at least half-a-year to affect the company’s financial reports. These constraints result in leaving 49 projects out of the scope of this research. Furthermore, for 6 projects the date-field was left blank. The inspection of these projects’ webpages on the website of Symbid makes it more likely to assume these projects were crowdfunded in 2016. One project is left out, as it did not make use of the equity method. In order to identify the companies behind the projects, I screen each webpage on Symbid for projects remaining on the list in order to hand-collect the missing data on the company name and number of crowdfunders. In the next stage, I search each company’s website in order to acquire the official company’s name and registration number at the Dutch Chamber of Commerce2. The missing registration numbers are found using the companies’ official
names on the website of Dutch Chamber of Commerce. The validity of the companies found is checked while screening the website found on the Internet and identifying the projects or products, which were the stated purpose of crowdfunding project on Symbid. Using this method, I am able to retrieve companies’ official information on each project remaining on the list. Several companies initiated more than one crowdfunding project for the same product or service, sometimes requesting funds for different stages of development and/or production. For the reasons of comparability and to avoid possible bias created by a single company having low/high agency costs, I only leave the first projects initiated by the same company on the sample’s list. In order to gather the financial information on the remaining companies I use the ORBIS database. ORBIS-database contains information on over 200 million companies, both listed and unlisted, from 228
countries (Bureau van Dijk, n.d.)3. This information is gathered and structured by Bureau van Dijk
from legal sources and publications in these countries. For all of the companies I use financial information available on all years after crowdfunding has taken place. I retrieve information on companies’ fixed tangible assets, fixed assets, NACE industry code (4 digits)4, status and status
date, date of incorporation, total assets and total debt. If no information on fixed tangible assets is available, I use the value of fixed assets. In all such cases this value is 0 making such substitution appropriate. In some cases, the financial data is missing or states that the necessary data is not available. The companies without the necessary financial data are scraped from the remaining list. Finally, the data from Symbid is merged with the data from ORBIS, using companies’ registration numbers as keys. For each available year a separate record is created, producing a data set with 59 cases on 39 companies.
In order to test my hypothesis I use OLS-analysis. The suggested hypotheses can be expressed by the following equation:
ACi = ƒ(EQUITYi, CROWDFUNDERSi, Zi), (1)
whereby ACi stands for agency costs of the company i; see hereunder for the way agency costs are
measured, EQUITYi equals the percentage of equity the crowdfunders get, CROWDFUNDERSi
represents the number of crowdfunders in the company’s crowdfunding project, and Zi represents
other factors, which might also influence the level of agency costs. These factors are described further in this section.
3.1 Agency Costs
Recently researches use the measures of agency costs as proposed by Ang et al. (2000). They measure the agency costs as the ratios of sales-to-assets as an indication of insufficient asset utilization and of operating-expense-to-annual-sales as an indication of excessive perk consumption. Although these measures seem to be the most appropriate to capture the agency costs, it is not possible to use them for Dutch crowdfunded companies. These companies are mostly small enterprises and are not obliged to publish profit-and-loss statements by Dutch law. Indeed, for all of the companies in the current sample no information on expenses or sales has been found. Therefore, I have to use another measure as a proxy for agency costs.
3 As of January 25, 2017 ORBIS-database covers total number of 209,138,566 companies.
In his paper on venture capital Gompers (1995) suggests that the tangibility of assets measured by tangible assets to total assets affects the expected agency costs: the less tangible the assets become, the more agency costs are expected. In case of liquidation the capitalists might be able to recover (a part) of their investment by selling the remaining tangible assets, thus reducing the need and costs to monitor the manger’s actions. Furthermore, investments into research and other type of intangible assets might be more valuable to the manager but not necessarily result into more value for the shareholder. A similar explanation is considered by Rajan & Zingales (1995), Aivazian, Ge, & Qiu (2005) and Cassar & Holmes (2003). Following this approach, I measure agency costs by the distance of the intangibility5 ratio to the baseline intangibility, whereby the higher this distance
gets, the higher the expected agency costs are. As stated by Jensen & Meckling (1976), the agency costs equivalent the total amount of money which negatively affects the principals’ (potential) welfare as a result of agency problems. To measure agency costs in a particular company one must therefore define a baseline for comparison at which total agency costs equal zero. This should be the case at companies with one single owner: the single-owner companies do not incur agency costs related to manager-shareholder relationship by their definition. To define this zero-agency baseline for this research I collect financial information on companies with a single owner using the ORBIS-database. As the practical assets tangibility might be industry related, I only gather information on companies from the same industries as the crowdfunded companies in the sample, using four digits NACE industry code.
Then, the mean intangibility ratio is calculated for each industry, giving the zero-agency baseline. Finally, I calculate agency costs, measured by the distance to this baseline, for each company from the sample matching its industry, using equation:
ACi = Ii - Ῑnace, (2)
whereby ACi stands for agency costs of company i, Ii equals the intangibility ratio of company i
and Ῑnace equals the mean intangibility of company i’s industry. To exclude possible industry wide
effects in a particular year, while performing this calculation I use ratio’s matched by financial years.
5 The intangibility ratio used in this paper is calculated as total assets less tangible assets divided by total assets. This somewhat excessive calculation method is specifically chosen to avoid any possible confusion with coefficient direction during empirical tests.
3.2 Crowdfunders
The main objective of this research is the relationship of agency costs and the ownership structure, which I operationalize by the percentage of equity acquired by the crowdfunders. The expected relationship between the percentage of equity and the agency costs is a straightforward operationalization of the agency costs hypothesis (Jensen & Meckling, 1976). However, this relationship should be moderated by the wisdom-of-the-crowd effect as discussed in the previous section of this paper. I operationalize this main effect by (natural log of) the number of crowdfunders who participated in the crowdfunding project. As the crowd gets bigger, so does its wisdom and less are the agency costs. Previous research has already discovered the existing relationship between the number of crowdfunders and quality of the crowdfunding project (Cordova et al., 2015) and I expect this effect to have impact on the later stage, after the project has been successfully crowdfunded, as well. The moderation effect is measured by multiplying the percentage of equity and (natural log of) the number of crowdfunders. Hereby the mean-centered values of variables are used as using the nominal values will produce undesired results. As explained by Field (2013), while using nominal values in models containing moderation, one should interpret the resulting coefficients of moderation term as being true if one of the variables producing this term equals zero. As both percentage of equity issued and the number of crowdfunders never equal zero in case of crowdfunded companies, such interpretation is not suitable for current research.
3.3 Controls
The intangibility of the company’s assets as well as the overall efficiency might be related to the company’s age. The start-ups might be founded with a limited number of tangible assets and build them up as the company begins to book revenues. I therefore include the company’s age at the reporting date, measured as the number of months, as a control variable.
The data includes financial information from several years on some of the companies. Previous research indicates that agency costs occur after ownership gets divided (that is, the equity shares have been issued) but does not specify the period for this to take place. It is possible that the agency costs as expected in the current research do not emerge immediately after equity issue or that these rise for a period of time as the original owner finds his or her way to compensate the monetary reward loss by other means until reaching an equilibrium. I therefore control for the number of months after funding has taken place. For this, I use four dummy variables, which are being coded as following. First variable equals 1 if the number of months between the funding
same way for the number of months between 8 and 13, between 14 and 23 and higher than 23. These numbers are chosen according to percentile values for the nominal number of months. The company’s size, measured as the natural log of total assets, is also included as control variable. Larger companies have more financial opportunities to invest in tangible assets and are expected to employ better technology (Margaritis & Psillaki, 2010).
Previous research suggests two control variables, which have been associated with the crowdfunding project’s success. Successful crowdfunding projects are those that are able to acquire their funding target or exceeded that target. As described in earlier section, crowdfunders are more likely to invest in a project when they perceive it as being of high quality; that is being feasible for accumulating return on crowdfunders’ investment, based on the information available to them. This links the antecedents of the project’s success to information asymmetry. Mollick (2014) finds that the increasing crowdfunding target is negatively associated with the projects’ success while Cordova et al. (2015) show positive association with the mean contribution by the crowdfunders. I only include crowdfunding target, measured as log of target amount, as control variable: mean contribution to the project is a product of crowdfunding target (which equals the obtained amount) and the number of crowdfunders. Therefore, I expect it to be highly correlated to these variables creating a multicollinearity issue. Indeed, while testing for correlation I find a Pearson’s r of 0.537 for log of target and natural log of mean contribution which is significant at p<0.001.
Finally, it is possible that a company financed by crowdfunding does not succeed in reaching its corporate goals and eventually faces bankruptcy. In those cases, crowdfunders among other possible capital providers incur costs that also fall under the definition of agency costs: they lose their investment, possibly entirely. Thus, I should include bankruptcy into my analysis as additional control variable. However, as I show while describing the data in the following section, there is a very limited number of Dutch equity crowdfunded companies, which faced bankruptcy at the time of this research. Therefore, including this variable does not provide meaningful information for my analysis. Thus, bankruptcy variable is not included.
In this section I provide the description of the methods used to test the hypotheses. In the next section I describe the data, which is being used in this research.
4 Data
This section provides description of the data used for this research. As mentioned in the previous section, I use two different sets of data: one to define the zero-agency baseline and the other containing information on crowdfunded companies. In order to estimate the agency costs for each crowdfunded company I ultimately combine these two sets of data by matching the NACE industry code from the zero-agency baseline set with that in crowdfunded companies set. I also match financial year while combining the data. The resulting set of data represents a copy of crowdfunded companies set extended by two variables: mean industry intangibility and agency costs, measured as distance between each company’s intangibility ratio to the zero-agency baseline. In the following, I therefore provide the description of the data from zero-agency baseline set and that of the data from the final data set, which I use to test the hypotheses.
4.1 Zero-agency baseline
To define the zero-agency baseline I retrieve financial information on medium and small Dutch companies with sole owner, using ORBIS-database. I only retrieve information for industries in which the crowdfunded companies operate using four digits NACE industry codes. Obtaining data using these criteria results in 61,985 observations. In the following stages records are being removed if these contain no information on total assets or if these equal 0 and if the information needed to calculate intangibility ratio is not available. Also only those records are being kept which contain necessary financial data on reporting years from 2012 to 2015.
The data retrieved from the ORBIS-database contains information on direct shareholder as well as global ultimate shareholder. In case this are not the same person or entity, these records are also being excluded to avoid companies with complicated ownership structures as these might contain more shareholders which are not detectable without further closer investigation.
Limiting the data of companies with a sole owner is further complicated by constructions whereby a company is not owned by a person directly but by a holding company. In most of the cases a company is stated to have an ultimate shareholder indeed. To limit the data as close as possible to that on sole owner companies I exclude the records whereby a shareholder’s NACE industry code is not 6420 (Activities of holding companies), 6430 (Trusts, funds and similar financial entities) or 7010 (Activities of head offices). The holding companies, set up to be intermediate between ultimate owner and operational company and their only purpose being to manage the latter, may be classified under any of these three industry codes. Finally, companies with non-Dutch ultimate
shareholders are being excluded to achieve maximum comparability to the crowdfunded companies as all of the latter have Dutch ultimate shareholders.
Because of filtering the obtained data as described above, the number of observations for calculating zero-agency baseline is reduced to 27,700. Using this data I calculate mean intangibility ratio for each industry from the crowdfunding data set. This calculation is performed for each year from 2012 to 2015. Calculated mean values are being added to the final data set through comparing appropriate NACE industry codes and years. As a final step, I calculate the distance from intangibility ratio of each company at each available reporting period to the mean intangibility of appropriate industry in appropriate year.
4.2 Resulting data
The data set provided by the crowdfunding platform contains 133 projects. I immediately reduce this to the projects that reached their crowdfunding target from January 2012 to July 2015. This reduction is based on financial data availability (there is no financial data available later than 2015 at the time of this research) and on the assumption that the agency costs will occur and become visible in financial records no earlier than six months after the project has reached its target and the equity share of the company’s founder is reduced as result. After merging the data with financial information as described in the previous section, I leave out the companies on which there is no financial information available.
The final data set contains 59 observations of financial data on companies financed by equity using the crowdfunding method in the period from January 2012 to July 2015. One company is removed from this data set: this company issued 53 percent of its equity shares in exchange for the investment while the equity share percentage for all remaining companies ranges from 0.50 to 20 percent. Remarkably, this company executed two crowdfunding projects and filed for bankruptcy a year after it acquired capital by second round of funding, which makes this case even more exceptional: of all the remaining companies only one has faced bankruptcy after the crowdfunding project has been completed.
Ten of the remaining cases in the data set show negative agency costs: the intangibility in these cases is lower than that of their industry mean. These cases are also left out as it is assumed that the industry mean represents the zero-agency baseline. One case is left out as its intangibility distance equals 24.50 percent points while that of the remaining cases does not exceed 17.98 and two cases with a number of crowdfunders of 1,765 which is at least four times as high as for any other case in the data set. Finally, I leave out five more cases after natural log transformation of
the dependent variable as they reported extremely low values compared to the remaining sample, creating distribution issues.
Table 1 – Descriptive statistics of Dutch crowdfunded companies during the years 2012-2015
Mean Median Minimum Maximum Standard deviation
INTANG 9.580 8.473 3.059 17.979 4.340 EQUITY 8.393 8.000 0.500 20.000 5.957 CRWDFS 145.250 102.500 30.000 398.000 110.595 INT 1.286 0.020 -7.920 8.720 4.251 SIZE 4.409 5.201 -2.690 7.400 2.314 AGE 36.980 29.000 7.000 156.000 31.249 TARGET 79.285 52.500 20.000 300.000 67.118 REP1 0.175 0.000 0.000 1.000 0.385 REP2 0.325 0.000 0.000 1.000 0.474 REP3 0.225 0.000 0.000 1.000 0.423 REP4 0.275 0.000 0.000 1.000 0.452 Number of observations 40
INTANG = distance (percent points) from intangibility ratio of a company to zero-agency baseline
(corresponding industry mean intangibility), EQUITY = percentage of equity shares issued to crowdfunders (percent points), CRWDFS = nominal number of crowdfunders, INT = interaction of mean-centered equity issued and mean-centered natural log of the number of crowdfunders, SIZE = natural log of total assets, AGE = company’s age at the reporting date (months), TARGET = total amount invested by the crowdfunders (euros x 1,000), REP1 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is lower than 8, REP2 = dummy
variable, number of months from the moment crowdfunding has taken place to the reporting date from 8 to 13, REP3 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date from 14 to 23, REP4 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is higher than 23.
The descriptive statistics of the remaining 40 cases are presented in Table 1 and include the mean value, median, minimum and maximum values and finally the standard deviation. It shows that the mean intangibility distance between the company and its appropriate industry value equals 9.58 percent points (Mdn=8.47, SD=4.34) and ranges from 3.01 to 17.98. The mean equity percentage issued to crowdfunders equals 8.39 percent (Mdn=8, SD=5.96) with the minimum of 0.50 percent and maximum of 20. This equity percentages have being issued to at least 30 and no more than 398 crowdfunders, mean value of 145.25 (Mdn=102.50, SD=110.60). The interaction term, being the product of the mean-centered values of equity percentage issued to crowdfunders and natural log of the number of crowdfunders, reports the mean value of 1.29 (Mdn=0.02, SD=4.25) and
equals 4.41 (Mdn=5.20, SD=2.31) and ranges from -2.69 to 7.4, meaning that a number of companies has a total assets value lower than one thousand euros. The mean age of a company at the reporting date is 36.98 months (Mdn=29, SD=31.25) with the minimum of 7 and maximum of 156 months. The mean target achieved during crowdfunding campaigns equals 79.29 thousand euros (Mdn=52.50, SD=67.12) with the minimum amount of 20 and the maximum of 300 thousand euros. The mean values of the dummy variables for the number of months between the funding and reporting dates are as follows: for the number of months lower than 8 – 0.18 (SD=0.39), from 8 to 13 months – 0.36 (SD=0.47), from 14 to 23 months – 0.23 (SD=0.42), and for the number of months higher than 23 – 0.28 (SD=0.45). All of the dummy variables report median value of 0 and range from 0 to 1 as coded.
Table 2 – Correlation matrix for Dutch crowdfunded companies during the years 2012-2015
V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V1: lnINTANG 1.000 V2: EQUITY 0.104 1.000 0.525 V3: lnCRWDFS 0.009 0.298* 1.000 0.958 0.062 V4: SIZE -0.239 -0.393** 0.003 1.000 0.138 0.012 0.985 V5: lnAGE -0.138 -0.137 0.099 0.117 1.000 0.395 0.399 0.544 0.471 V6: logTARGET 0.394** -0.081 0.444*** 0.224 0.402** 1.000 0.012 0.618 0.004 0.165 0.010 V7: REP1 -0.029 0.102 0.323** 0.200 0.023 0.376** 1.000 0.858 0.532 0.042 0.217 0.890 0.017 V8: REP2 -0.011 -0.157 -0.169 0.126 -0.230 0.017 -0.320** 1.000 0.949 0.333 0.296 0.439 0.154 0.918 0.044 V9: REP3 -0.160 -0.168 -0.106 -0.076 0.060 -0.078 -0.248 -0.374** 1.000 0.325 0.301 0.516 0.640 0.715 0.633 0.123 0.017 V10: REP4 0.185 0.235 0.002 -0.231 0.166 -0.265* -0.284* -0.427*** -0.332** 1.000 0.253 0.144 0.990 0.152 0.306 0.098 0.076 0.006 0.036
Note: P-values listed below correlations coefficients.
*,**,*** denote significance at the 10% level, 5% and 1% level, respectively.
lnINTANG = natural log of distance in percent points from intangibility ratio of a company to
zero-agency baseline (corresponding industry mean intangibility), EQUITY = percentage of equity shares issued to crowdfunders (percent points), lnCRWDFS = number of crowdfunders, SIZE = natural log of total assets, AGE = company’s age at the reporting date (natural log of number of months), logTARGET =
log of the total amount invested by the crowdfunders, REP1 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is lower than 8, REP2 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date from 8 to 13,
REP3 = dummy variable, number of months from the moment crowdfunding has taken place to the
reporting date from 14 to 23, REP4 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is higher than 23.
Table 2 provides Pearson correlations of variables used in this research. There is a number of significant correlations between independent variables, however all are lower than 0.5. I also test all of the independent variables for multicollinearity based on tolerance, which is much higher than 0.1 for all of the variables, except for dummy variable REP2 for which the tolerance equals 0. Therefore, the latter is excluded from further analysis. Both statistics on remaining variables suggest that there are no multicollinearity issues in the data set used in this research.
5 Empirical Results
In this section I present the results of testing the hypotheses. For all independent variables as described in Section 3 there are certain predictions of association between them and intangibility distance as a proxy for agency costs. First, I split the sample into low-intangibility-distance and high-intangibility-distance groups using the entire sample median value of intangibility distance. Table 3 presents the results of t-test of differences in mean values of independent variables between the resulting groups. As all predictions imply specific directions of these differences, a one-sided t-test is used.
Table 3 – One-sided t-test of variables used to analyze agency costs at Dutch crowdfunded companies during the years 2012-2015
Below Median
(n=20) Above Median (n=20) Difference t-value
EQUITY Mean 7.386 9.401 2.015 1.072 Standard deviation 6.603 5.205 lnCRWDFS Mean 4.657 4.777 0.120 0.506 Standard deviation 0.838 0.650 SIZE Mean 4.947 3.871 -1.076 -1.492* Standard deviation 1.687 2.745 lnAGE Mean 3.435 3.289 -0.146 -0.663 Standard deviation 0.874 0.452 logTARGET Mean 1.645 1.900 0.255 2.597*** Standard deviation 0.316 0.304
*,**,*** denote significance at the 10% level, 5% and 1% level, respectively.
EQUITY = percentage of equity shares issued to crowdfunders (percent points), lnCRWDFS = natural
log of the number of crowdfunders, SIZE = natural log of total assets, lnAGE = natural log of company’s age at the reporting date (months), logTARGET = log of the total amount invested by the crowdfunders
As predicted, the group of highly intangible companies contains crowdfunding projects, which issued higher equity percentage (M=9.40, SD=5.21), than the group with low intangibility distance (M=7.39, SD=6.60). However, this difference is not significant and represents a small-sized effect (t=1.072, p=0.146, d=0.31). The companies with high intangibility distance to the zero-agency baseline are smaller (M=3.87, SD=2.75), than low-intangibility-distance companies (M=4.95, SD=1.69). These difference is significant (t=-1.492, p=0.073) and represents a medium-sized effect (d=-0.64). There is also a significant difference, representing a large effect (t=2.597, p=0.007, d=0.81), between the mean values of the (log of) crowdfunding target between high-intangibility-distance (M=1.90, SD=0.30) and low-intangibility-distance groups (M=1.65, SD=0.32). Furthermore, high-intangibility-distance group appears to be younger
(M=3.29, SD=0,45), than low-intangibility difference group (M=3.44, SD=0.87). However, this difference is not significant (t=-0.663, p=0.256) and this effect is small (d=-0.17). Although not all significant, these differences are in the predicted directions. On the other hand, the (natural log of the) number of crowdfunders appears to be higher for the group of high-intangibility-distance (M=4.78, SD=0.65), than for the group of low-intangibility-high-intangibility-distance companies (M=4.66, SD=0.84). These difference is non-significant and this effect is also small (t=0.506, p=0.308, d=0.14). The latter finding challenges the second hypothesis as it predicts for intangibility distance to be higher when the number of crowdfunders is high. However, as the number of crowdfunders is not the only predictor of the agency costs, this finding is not sufficient to reject the second hypothesis.
Splitting the sample using 25th and 27th percentiles for intangibility distance produces similar
results. However, while using the value of 25th percentile to distinguish two groups, the t-test of
the difference in natural log of total assets is non-significant (t=-0.677, p=0.252, d=-0.69). At the same time, the (natural log of) number of crowdfunders is lower for the group with high
intangibility distance (M=1.54, SD=0.17), than for the group with low intangibility distance to the zero-agency baseline (M=1.55, SD=0.16). However, this difference is non-significant (t=-0.198, p=0.422, d=-0.08).
Table 4 presents the OLS estimation results of the three models tested in this research. Model I aims to test the first hypothesis, that is the positive association between the percentage of equity issued to crowdfunders and the agency costs. Model II tests the second hypothesis, which expects a negative association between the number of crowdfunders and agency costs. Model III tests the third hypothesis, which predicts the association between the percentage of equity and agency costs to be lower when the number of crowdfunders is high than when it is low. All three models use the natural log of the distance between intangibility of a given company to that of the appropriate industry as dependent variable and include control variables which are likely to be associated with agency costs in a certain way.
These models produce mixed results in terms of support for the hypotheses. While the
coefficient for (natural log of) the number of crowdfunders is negative and significant in Models II and III as predicted, that of the equity issued to the crowdfunders is non-significant in Model I. It is also negative which is contrary to the prediction. Testing Model III produces
non-significant coefficient for equity as well, although it is positive in this model. The interaction term between (the mean-centered) equity percentage and the (natural log of) number of crowdfunders is also negative and non-significant, contrary to the third hypothesis.
Table 4 – Initial OLS regression estimates with natural log of intangibility distance as dependent variable for Dutch crowdfunded companies during from January 2012 to July 2015
OLS Regressions with dependent variable lnINTANG
Model I Model II Model III
Prediction Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic
Constant EQUITY + -0.007 -0.619 0.004 0.366 lnCRWDFS - -0.226 -2.614** -0.263 -1.844* INT -0.015 -0.644 Controls SIZE - -0.058 -1.960* -0.057 -2.267** -0.051 -1.708* lnAGE - -0.361 -3.449*** -0.385 -4.034*** -0.386 -2.922*** logTARGET + 1.203 5.277*** 1.448 6.326*** 1.368 5.486*** Dummy variables REP1 -0.190 -1.017 -0.111 -0.644 -0.088 -0.306 REP3 -0.008 -0.047 0.025 0.163 0.013 0.073 REP4 0.432 2.475** 0.496 3.102*** 0.502 3.356*** Adjusted R-squared 0.422 0.518 0.497 F-statistic 5.060*** 6.978*** 5.276*** Number of observations 40 40 40
*,**,*** denote significance at the 10% level, 5% and 1% level, respectively.
lnINTANG = natural log of the distance from intangibility ratio of a company to zero-agency baseline
(corresponding industry mean intangibility), EQUITY = percentage of equity shares issued to crowdfunders (percent points), lnCRWDFS = natural log of the number of crowdfunders, INT = interaction of mean-centered equity issued and mean-centered natural log of the number of
crowdfunders, SIZE = natural log of total assets, lnAGE = natural log of company’s age at the reporting date (months), logTARGET = log of the total amount invested by the crowdfunders, REP1 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is lower than 8, REP3 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date from 14 to 23, REP4 = dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is higher than 23. Model III uses mean-centered values for variables EQUITY and lnCRWDFS as explained in Section 3. Model III is tested using PROCESS tool by Andrew F. Hayes (http://www.afhayes.com) for IBM SPSS6.
6 The PROCESS tool, developed by Andrew F. Hayes (http://www.afhayes.com), adds a custom dialogue box to IBM SPSS to test models containing moderation and mediation variables (Field, 2013). It mean-centers appropriate variables and adjusts standard errors for heteroscedasticity. Due to the latter, the standard errors and t-statistics produced using PROCESS tool differ from those produced using standard SPSS dialogue box for regression. These differences did not affect the significance of coefficients found. Furthermore, PROCESS tool structures input data for plotting for further examination of the interaction effects. This is not performed in current research, as the coefficient of the interaction term is non-significant: any graphical differences would therefore be meaningless.
Furthermore, all three models suggest that the intangibility distance to the zero-agency baseline is only significantly higher at least two years after crowdfunding has taken place as the dummy variables REP1 and REP3 are non-significant in these models. I therefore rerun the regressions while substituting all of the dummy variables by one: REP23 which equals 1 if the number of months between the funding and reporting dates is higher than 23 and 0 in all other cases. The results of testing these altered models are presented in Table 5 and discussed in more detail further in this Section.
Table 5 – OLS regression estimates with natural log of intangibility distance as dependent variable for Dutch crowdfunded companies during from January 2012 to July 2015
OLS Regressions with dependent variable lnINTANG
Model IV Model V Model VI
Prediction Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic
Constant EQUITY + -0.010 -0.903 0.003 0.321 lnCRWDFS - -0.239 -2.905*** -0.275 -2.179** INT -0.017 -0.860 Controls SIZE - -0.064 -2.234** -0.060 -2.471** -0.053 -1.912* lnAGE - -0.355 -3.540*** -0.376 -4.163*** -0.383 -3.324*** logTARGET + 1.133 5.349*** 1.409 6.508*** 1.329 6.519*** Dummy variables REP23 0.468 3.140*** 0.501 3.753*** 0.514 4.027*** Adjusted R-squared 0.437 0.538 0.524 F-statistic 7.047*** 10.079*** 7.635*** Number of observations 40 40 40
*,**,*** denote significance at the 10% level, 5% and 1% level, respectively.
lnINTANG = natural log of the distance from intangibility ratio of a company to zero-agency baseline
(corresponding industry mean intangibility), EQUITY = percentage of equity shares issued to crowdfunders (percent points), lnCRWDFS = natural log of the number of crowdfunders, INT = interaction of mean-centered equity issued and mean-centered natural log of the number of
crowdfunders, SIZE = natural log of total assets, lnAGE = natural log of company’s age at the reporting date (months), logTARGET = log of the total amount invested by the crowdfunders, REP23 = dummy variable, dummy variable, number of months from the moment crowdfunding has taken place to the reporting date is higher than 23. Model VI uses mean-centered values for variables EQUITY and lnCRWDFS as explained in Section 3. Model VI is tested using PROCESS tool by Andrew F. Hayes (http://www.afhayes.com) for IBM SPSS.
5.1 Equity and Agency Costs
The first hypothesis (H1) predicts a positive coefficient for the percentage of equity issued to the crowdfunders (EQUITY) and the distance between a company’s intangibility ratio to its industry zero-agency baseline, which I use to measure the agency costs. The results do not support this hypothesis: the found coefficient is negative and non-significant (β=-0.010, t=-0.903). This finding is contrary to the earlier presented results of the t-test, which suggest that there is a non-significant but positive difference in the mean value of equity issued to crowdfunders between the companies with high intangibility distance to the zero-agency baseline and those with low intangibility distance. This contrast is explained by the presence of other variables in Model IV, which are used as controls, while the t-test does not include these variables and therefore does not take possible differences between companies into account other than in the intangibility distance. Thus, other things equal, the empirical results show that there is no evidence that the higher percentage of equity issued to the crowdfunders leads to higher agency costs in
companies using equity crowdfunding to acquire external capital.
Strictly speaking, this result only implies that such an effect is not present in the sample used for this research. However, this particular sample represents 26 of 62 companies that used equity crowdfunding method in the Netherlands in the period from January 2012 to July 2015. That implies that the results may be generalized for at least the Dutch companies seeking
crowdfunding.
This finding is contrary to previous research of agency costs in various settings and it might indicate that crowdfunding companies differ from others in a way the principal-agency relationships function. However, this contradiction might be due to a different measure of agency costs. Starting with Ang et al. (2000), agency costs are commonly measured by the ratios of operating-expenses-to-annual-sales and annual-sales-to-total-assets. However, these measures are not available for current research due to the absence of annual sales data or any other data from profit-and-loss statement. This research uses distance from intangibility ratio of a company to zero-agency baseline as a proxy for agency costs and Model IV does not show significant association between this measure and percentage of equity shares issued to crowdfunders. Another measurement difference might explain the contradictory. I use percentage of equity specifically as predictor while other researchers frequently use the number of non-managing shareholders. The percentage of equity issued to crowdfunders appears to be a more accurate measure as the number of owners might not reflect all of the shareholders. As the maximum number of shareholders for not listed enterprises might be limited by law, the owners might use
holding constructions to circumvent legal restrictions. In these cases, the number of owners might contain measurement errors. Moreover, the information on shareholders is not always published and available in databases such as ORBIS. Indeed, further inspection of the data available in ORBIS-database reveals that the number of owners for the crowdfunded companies is recorded as either 0 or 1. In latter cases, almost all of the shareholding companies also report no shareholders, which cannot be accurate due to the legal forms of these companies. All of the companies except one do report the number of directors, which ranges from one to six. The directors’ lists contain both persons as companies. Assuming that at least some of these directors are also owners of the companies, I use this variable as proxy and substitute percentage of equity issued to crowdfunders by it in both models. I find no significant coefficients for this variable. Thus, the other possible explanation might be that the distance between intangibility ratio’s and the industry mean intangibility does not reflect the agency costs. However, this would contradict with other papers where the negative association between the tangibility of assets and the agency costs has been found.
While using the distance between intangibility ratio’s and the industry mean intangibility as a proxy for agency costs, this research does not find evidence for an association between the percentage of equity issued to crowdfunders and agency costs in the sample of crowdfunded companies in the Netherlands that acquired external capital from January 2012 to July 2015 using this novel financing method. However, this should not be interpreted as absence of agency costs at these companies. The presence of positive intangibility distances suggests they are. Therefore, percentage of equity issued to crowdfunders simply does not seem to drive agency costs. In fact, this should even be expected in situations when equity shares do not affect the owner’s income in significant way. Equity percentage issued to new owners affects original owner’s income when a company reports profit, which can be shared. In cases when the profit is low or absent and the owner’s monetary compensation is formed by a salary, the issuance of equity will not practically affect his or her monetary reward, creating no necessity to compensate by other means. Such cases are not uncommon among small Dutch firms where owners’ main compensation is often formed by a salary or a management fee that are reported as immediate expense. The companies in the sample used in this research have a mean absolute value of total assets of 332 thousand euro’s meaning these companies are indeed small. Furthermore, additional analysis reveals that there is a number of high leveraged companies in the sample, some having even more debt than the total value of assets. This suggests these companies report limited profit or even loss leaving nothing for the shareholders to consume. Therefore, issuing equity to external shareholders, in
