Crowdfunding (CF) is a digital-financial innovation that, bypassing credit crisis, bank system rigidities and constraints of the capital market, is allowing new ventures and established companies to get the needed funds to support innovations. After one decade of research, mainly focused on relations between variables and outcomes of the CF campaign, the literature shows methodological lacks about the study of its overall behavior. These reflect into a weak theoretical understanding and inconsistent managerial guidance, leading to a 27% success ratio of campaigns. To bridge this gap, this paper embraces a “complex system” perspective of the CF campaign, able to explore the system's behavior of a campaign over time, in light of its causal loop structure.
By adopting and following the document model building (DMB) methodology, a set of 26 variables and mutual causal relations modeled the system “Crowdfunding campaign” and a data set based on them and crafted to model the “Crowdfunding campaign” with a causal loop diagram. Finally, system archetypes have been used to link the causal loop structure with qualitative trends of CF's behavior (i.e. the raised capital over time).
The research brought to 26 variables making the system a “Crowdfunding campaign.” The variables influence each other, thus showing a set of feedback loops, whose structure determines the behavior of the CF campaign. The causal loop structure is traced back to three system archetypes, presiding the behavior in three stages of the campaign.
The value of this paper is both methodological and theoretical. First, the DMB methodology has been expanded and reinforced concerning previous applications; second, we carried out a causation analysis, unlike the common correlation analysis; further, we created a theoretical model of a “Crowdfunding Campaign” unlike the common empirical models built on CF platform's data.
1. Introduction
The last decades have been characterized by growing entrepreneurial opportunities, availability of information, ease of networking and venturing, support of policymakers to entrepreneurship and economy. In contrast they have been characterized by financial crisis and bank fragility (Mamatzakis et al., 2021; Casey and O'Tool, 2014).
The increased demand of capitals for new entrepreneurial initiatives and, on the opposite, the weakening of the capital market triggered a wave of financial innovations, leveraging on digital technologies, able to compensate the gaps of the traditional financial systems.
Based on the Internet and digital platforms a set of new systems tapping into management, finance and digital technologies started supporting a strategic yet unsatisfied demand for capitals and services.
This set of innovations like crowdfunding (CF), mobile payments, cryptocurrencies and blockchain became a challenging research area for management scholars, named Financial Technology (FinTech) (Elia et al., 2022; Milian et al., 2019; Thakor, 2019).
Among the many FinTech systems, CF has gained relevance both theoretically and practically. It is a process for both collecting capitals to finance business ventures and entrepreneurial initiatives and for personal microinvestments. It is free of physical intermediaries and bureaucratic constraints and enables the matching between many individual small funders (crowd) and entrepreneurial capital seekers. The matching and the funding are operated by a virtual intermediary, namely, a digital platform (Brokaw, 2014; Hagiu and Wright, 2015; Linzalone et al., 2020).
CF is playing a relevant role in today's economy: supports ventures in the seed stage of entrepreneurial development (e.g. technology digital start-up) (Wonglimpiyarat, 2018), funds companies seeking to amplify or internationalize their business (Belleflamme et al., 2014; Elia et al., 2018; Giudici et al., 2013; Troise et al., 2022a) and offers a market for digital companies (“cybermediary”) (Runfola et al., 2013).
Although a consistent body of literature focus on actors (Beier and Wagner, 2016; Valančienė and Jegelevičiūtė, 2014), technologies (Koch and Siering, 2015), drivers (De Luca et al., 2019; Mollick, 2014), projects (Jinwook and Lee, 2015; Xiao et al., 2014) and outcomes (Chemla and Tinn, 2019; De Luca et al., 2019; Mollick and Robb, 2016), few studies focus the causes of CF behavior over time (e.g. the raised capital) considering the complex bundle of influences among the components. Such analysis addressable as CF dynamics (Haji Gholam Saryazdi et al., 2020; Mollick, 2014; Rao et al., 2014) is detectable through a study of CF as a complex and dynamic system.
After the foundational stage of the CF literature, aimed to identify and study the relations between components and critical outcomes of a campaign, a further theoretical advancement requires the study of the properties emerging from the system, to investigate the behavior, in particular, the trend of raised capital over time.
Understanding CF dynamics allows driving the raised capital toward the planned target, having cleared the causal mechanisms that preside over its behavior. Such an understanding is crucial to increase campaigns' success ratios, though requiring an underexplored perspective (Harris and Wonglimpiyarat, 2020; Mollick, 2014; Rao et al., 2014; Shneor and Vik, 2020).
The few, previous attempts to explore the system's behavior of a CF campaign are mostly empirical, using data from CF platforms, thus hampering theoretical relevance (Shneor and Vik, 2020). Also, the previous literature considers CF raised capital in static conditions, namely at the campaign's end date, but very limited studies explore its dynamics (Shneor and Vik, 2020).
To fulfill this gap, this paper investigates the behavior of the CF campaign, namely, with regard to the raised capital linking it with the underpinning causal structure. To this aim, a causal loop diagram (CLD) model (Forrester, 1971; Kiani et al., 2009; Lane, 2000; Sterman, 2000) of campaign is built, following a document model building (DMB) methodology (Haji Gholam Saryazdi et al., 2020).
The remainder of the paper is organized as follows. Section 2 presents a literature review of the main conceptual elements and working mechanisms underpinning CF together with a review of existing literature about CF dynamics; Section 3 outlines the research methodology, while Section 4 presents analysis and findings emerging from the CLD model. Section 5 discusses the results and, finally, the concluding Section 6 wraps up insights, implications and limitations of the study.
2. Theoretical background
2.1 Crowdfunding: theoretical elements and concepts
Funding is a vital input for entrepreneurship and business. It secures new ventures and existing companies the capitals required to transform new ideas into marketable products or services. Traditional funding systems like venture capital funds, banks and business angels, have been recently paired by new systems like CF.
CF was born during the global recession of 2008 (Cai, 2018) and increased during the following “worldwide credit crunch” (Krebsz, 2011), given the limited capacity of traditional funding systems to fund entrepreneurship (Elia et al., 2018).
Currently CF is just one of a large set of financial innovations leveraging on digital technologies, known as FinTech (Cai, 2018; Harris and Wonglimpiyarat, 2020; Milian et al., 2019; Thakor, 2019).
FinTech includes, besides CF, systems like cryptocurrencies and blockchain, new digital advisory and trading systems, artificial intelligence and machine learning, peer-to-peer lending and mobile payment systems, robot advisory, big data, and cloud computing (Cai, 2018; Elia et al., 2022; Thakor, 2019).
FinTech systems, often used in combination, provide advanced financial products and services for insurance, regulations and legislative foundations, loans, payments and billing, personal finance and asset management, money transfer and remittance, cryptocurrency, capital markets, and CF (Elia et al., 2022; Milian et al., 2019; Thakor, 2019).
FinTech is not only a prosperous market but also an interesting and challenging research field, where CF has gained a prominent position.
CF rises at the intersection of three literature streams, that of digital platforms (Brokaw, 2014; Hagiu and Wright, 2015), of microfinance (Beaulieu et al., 2015), digital business models (Wirtz, 2019; Zaheer et al., 2019) and strategic outsourcing of organizations (Haji Gholam Saryazdi et al., 2020).
CF consists of “pooling small amounts of capital from a potentially large pool of interested funders” (Short et al., 2017, p. 149) in contrast to the traditional pooling of large amounts from few professional investors and backers (Belleflamme et al., 2014).
From the digital platform's perspective CF is the system of activities performed by a digital platform to match the demand of capitals coming from entrepreneurs, with the demand of investment projects coming from small investors. The CF platform is owned by a digital intermediary company, is accessible via the Internet and supported by a multisided digital platform (Brokaw, 2014; Hagiu and Wright, 2015) (i.e. a web portal designed for matching demand and offer). The platform company keeps a share of the capital raised by successful campaigns (Kimes and Wirtz, 2003; Zott et al., 2011).
The design, the operational settings and the management of the CF platform are critical elements of campaigns' success (Belleflamme et al., 2015; Lacan and Desmet, 2017) and antecedents of performance and sustainability (Ammirato et al., 2021).
CF is gaining theoretical relevance also from a strategic management perspective since the knowledge feedback going from investors to the entrepreneurs are strategic innovation inputs and sources of innovation and change (Troise et al., 2021).
Returns or compensations for funders can range from monetary to rewards, up to equity or to just moral satisfaction, depending on the CF model (Haji Gholam Saryazdi et al., 2020; Mollick, 2014).
About CF projects, they can be of various capitals (targets) (Schwienbacher and Larralde, 2010) and various types (e.g., business and entrepreneurship, public administration, art, journalism, social, etc., all types) (Giudici et al., 2013).
The CF campaign is limited in time and starts with the founders that post on the web platform (e.g. kickstarter.com, eppela.com, etc.) a set of information and digital contents about their entrepreneurial project to promote and showcase it. The platform's audience can browse, seek, compare and evaluate hundreds of projects, very easily and effectively, before they decide the project to fund. Key data are the capital required by founders (the target). However, “signals” are the strategic elements of a project caught by the audience and used to decide if to support it or not (Massa Saluzzo and Alegre, 2021; Shneor and Vik, 2020; Troise et al., 2022b) among plenty of competing projects.
During the campaign, funders can pledge using the payment services made available by the platform (e.g. PayPal, credit card, wire transfer or autonomous agreements of payment between parts). Once the campaign is expired, if the target is achieved the campaign is successful and the funders are obliged to fulfil their promised funding (Etter et al., 2014). If the target is not achieved, the raised capital is returned back to the backers (the “All-or-Nothing” (AoN) model), or alternatively is kept by the founders to carry out the project anyway (the “Keep-it-All” (KiA) model) (Beaulieu et al., 2015; Belleflamme et al., 2016; Mollick and Robb, 2016). This depends on the model adopted by the CF platform.
From an investment point of view there are four models of CF, depending on the type of return: donation-based, lending-based, reward-based and equity-based (Agrawal et al., 2015; Aitamurto, 2015; Mollick, 2014) (see Table 1).
Despite the operational differences, all four CF models have a common mechanism of interests, motivations and transactions, that is, a causal structure; it underpins and determines the behavior of the CF campaign over time and then the raised capital.
The research gap on CF system's dynamics, acknowledged by some scholars (Haji Gholam Saryazdi et al., 2020; Shneor and Vik, 2020), hampers the discovery of behavioral properties. These properties are not deductible by effects' overlay of the single components of CF, and a complex system perspective is suited to the analysis (Belleflamme et al., 2015; Mollick, 2014; Poutanen et al., 2016; Windsor, 2010). Also, dynamic system's properties are not observable in a static situation but with the system in motion, since the complex socio-economic processes of CF (a self-organized mass phenomenon typical of crowds) show in motion and develop in time according to mutual influences and adjustments between the components (i.e. in dynamic conditions).
We support that the study of CF campaign dynamics enables the transition from an initial and definitional stage of research, characterized by the static analysis of parts and effects, to a mature stage focused on systemic and dynamic analysis of CF. This objective provides a valuable base to design CF projects, to control campaign behavior and then to increase the success ratio of campaigns (Belleflamme et al., 2014; Burtch et al., 2013; Xiao et al., 2014). This last is attested at 27% (Shepherd, 2020; Todorov, 2022), which imposes a reflection on the need to go beyond the extant knowledge and research.
2.2 Approaching crowdfunding campaign as a complex system
To study a CF campaign's behavior it is necessary to step forward the single relations between CF's parts (e.g. an actor, an actor's attribute, a platform feature or a setting, etc.) and their outcome (e.g. the raised capital). It is necessary to approach it as a system, whose behavior is produced by parts and variables in action, interacting and influencing each other over time (a complex system). The system's outcome over time, the behavior, is in fact a property emerging from the causal structure of the system (Poutanen et al., 2016).
In general the behavior of the CF campaign is then the trend of a campaign's focal effect over time. Assuminga system's focal effect as the raised capital, the dynamic analysis relates CF behavior with the structure of causal influences.
A CF campaign can be modeled as several subsystems, made up of components that have complex influences (Jackson, 2009; Poutanen et al., 2016); the isolated analysis of its components is neither suited to explain nor to predict the system's behavior. Its study requires a holistic and non-reductionist investigation approach, framed within the complexity theory (Jackson, 2009; Maani and Cavana, 2007; Poutanen et al., 2016). It is a suitable and effective theoretical perspective to study complex systems, nonlinear relationships, systemic interaction and the emergence of system behavior (Cilliers, 2011). This theoretical lens enables research to look at the evolution of human-technology systems (Fleming, 2001), that would be not possible otherwise.
Approaching CF as a system enables the discovery of forms of regular interaction or interdependence (Mele et al., 2010), thus providing generalizable and insightful knowledge about the system's behavior. A system can be defined as an entity, which is a coherent whole (Ng et al., 2009) such that a boundary is perceived around it to distinguish internal and external elements and to identify input and output relating to and emerging from (Mele et al., 2010).
CF, being a physical, dynamic and human system, is made up of (1) components (e.g. people, processes, products, capitals, etc.), (2) attributes (e.g. input, processes and output characteristics of each component) and (3) relationships between components and characteristics (Tien and Berg, 2003, pp. 23-24). Among this system's elements, some mutual cause–effect relationships are recognized (Haji Gholam Saryazdi et al., 2020; Harris and Wonglimpiyarat, 2020; Wonglimpiyarat, 2018). The behavior of a complex system arises from its structure. Complex systems' structure consists of the feedback loops and nonlinearities created by the interaction of the physical and institutional structure of the system with the decision-making processes of the elements acting within it (Sterman, 2000).
2.3 Models to study crowdfunding campaign's dynamic behavior
The existing literature provides few studies of CF as a complex system. Elia et al. (2018) outline three main subsystems: project initiators (individuals or start-ups searching for funds), crowdfunders (people keen on providing money to finance business projects) and a CF platform company (an organization that allows interaction among project initiators and crowdfunders). However, CF has three other subsystems: the project, the platform and the platform's audience. Each of them is made up of components.
What gives birth to the (complex) system's behavior is the feedback loop. It is a circular chain of cause–effect relations among n elements (or components or variables) of the system. So, if a change in the state of one attribute of Component A determines a change in the state of one attribute of the element B and the change in B, in turn, influences newly a change in the attributes of A, a causal feedback loop governs the system's behavior.
Humans' systems are governed by many feedback loops, and the overall systems' behavior is the determined by the system's feedback loop structure.
Feedback loops determine “hard to predict” and counterintuitive behavioral effects of the system that struggle with the human mental models and capacity to detect them and analyze their consequences (Ammirato et al., 2022; Groesser and Schaffernicht, 2012; Takahashi, 2005); thus specific and structured approaches of analysis are required.
The complex causal structure of CF is just nuanced in the study by Belleflamme et al. (2016) who highlight the key feedback loops among the three principal subsystems: fundraisers, funders and platforms (see Figure 1).
Exploring and theorizing the key mechanisms underpinning CF's behavior (at the global, platform and campaign levels) is a promising and viable way to increase the CF success ratio and indeed (Chen et al., 2018; Lee et al., 2016; Yang et al., 2015) show diseconomies and failures’ rates of CF platforms due to the lack of system studies. Strategic management of CF, thus, requires the analysis of the CF system's dynamics (Lee et al., 2016). Analyzing CF as a complex system allows increasing learning and understanding (Chemla and Tinn, 2019).
Despite the aforementioned studies, CF dynamics are far from being detected and understood; besides a complex and dynamic phenomenon like CF is elusive without a suited theoretical “lens” (Ammirato et al., 2022).
2.4 Detecting and analyzing crowdfunding campaign's causal structure: the causal loop diagrams
One of the most effective approaches to studying complex systems is system dynamics (Forrester, 1961; Moellers et al., 2019; Sterman, 2000). It provides a set of conceptual elements, guidelines and tools to support the analysis of the feedback loops of complex systems.
“Real life” complex systems can be traced back to combinations of archetypes, whose behavior is known. “Common modes” (Sterman, 2000) or “system archetypes” (Senge, 1990) are causal structure archetypes, who give rise to emergent (standard) trends over time of a given effect of the system (behavior).
The main tool able to identify and model a complex system's causal structure is the CLD (Forrester, 1961; Giordano et al., 2007; Kiani et al., 2009; Lane, 2008; Sterman, 2000). A further tool is the stock and flow diagram (Duggan, 2016; Sterman, 2000). The former is suited for wider scope qualitative analysis; the latter is needed in addition to the former for quantitative analysis.
A CLD is a standardized conceptual diagram (Lane, 2000, 2008) that represents a system's elements (displayed by texts), connected by arrows representing causal links (Lane, 2000, 2008; Sterman, 2000). Graphically, arrows connect the texts, with incoming or outgoing direction depending on the cause–effect relation flow (incoming for effects and outgoing for causes).
Each causal link (arrow) is marked with a polarity, positive (+) or negative (−), to indicate how the effect variable changes with respect to the cause. A positive link is assigned when a rise of the cause corresponds to a rise of the effect, or a decrease of the cause corresponds to a decrease of the effect. A negative polarity is assigned to the causal link if a rise of the cause generates a fall of the effect, or a fall of the cause makes the effect rise (Sterman, 2000) (see Figure 2).
Once drawn the CLD for the entire system model, the feedback loops show. They are the closed chains of arrows that start and end at the same element. The behavior of the system is determined by the structure of the feedback loops (i.e. the number, the type and interconnection). It is responsible for the behavior of the system and explains how and why it changes over time (Cronin and Gonzalez, 2007; Maani and Cavana, 2007; Sterman, 1989). The feedback loop type can be reinforcing (R) or balancing (B) and are marked by a loop identifier that shows whether the loop is reinforcing or balancing (Sterman, 2000).
The R loops tend to amplify any external stimuli since a rise (or a fall) in any of the loop's variables will increase (or absorb) the initial stimuli. Alternatively, in the case a system is in equilibrium, the stimuli that disturb this situation will be amplified and the dynamic behavior of the system will deviate exponentially from the initial equilibrium state (see Figure 3).
The reinforcing loop, therefore, gives rise to a self-reinforcement of a behavioral trend of the system (reinforcing loops are marked by the letter “R”). Negative feedback loops, on the other hand, tend to dampen any external stimulus. The balancing feedback loop, therefore, gives rise to a self-regulation or self-balancing process and is indicated by the letter “B” (balancing) (Sterman, 2000) (see Figure 4).
Reinforcing and balancing are the two basic feedback loop types that preside over the two basic behaviors of a system: amplification and dampening. Real systems embody much more complex causal structures than the two basics; however, Senge's nine system archetypes (Senge, 1990) allow modeling any real system to a combination of archetypes, whose behaviors are known.
A CF campaign can be conveniently analyzed with a CLD to explore the causal structure and discover properties and mechanisms underlying the behavior. In this context, causal links connect tangible (e.g. rewards, capitals, web pages of the platform, etc.) and intangible (e.g. regulations, reputation, trust, satisfaction, etc.) system components (Meadows, 2008).
3. Research methods
DMB (Haji Gholam Saryazdi et al., 2020) is a methodology that employs a literature review to create a data set for building a model. DMB is a data-driven research strategy (Bleijenbergh et al., 2009), where sources of data and numerical databases are written (Forrester, 1980). Sources can include articles, reports, books and the like.
In line with the research protocol of Haji Gholam Saryazdi et al. (2020) the following phases were carried out (see Figure 5):
clarification and motivation (C and M) of the question to investigate,
sampling, collection and screening (S, C and S) of the literature sources,
critical literature review (CLR) of the sources and extraction of the relevant model data and
interpretation, composition and presentation of the results in a suitable form.
3.1 Clarification and motivation of the question to investigate
Approaching CF as a complex system, we can detect the feedback loop structure that determines its dynamics. We assume that CF has characteristics, behaviors and properties to discover with this approach. Approaching CF as a complex system has a high chance of coming to new and valuable findings. The research question we stated is the following: What is the causal feedback structure underpinning a CF campaign's behavior?
We found a few similar pieces of research addressing this question. Literature explored many single cause-effect relations occurring between the system's elements and a CF campaign's outcome.
Discovering and analyzing CF's causal structure give ground to study many aspects of its behavior and support business management and decision-making (e.g. entrepreneurs, small individual investors, platform managers, policymakers and regulators) about the system's settings, platform strategies and system's policies.
3.2 Sampling, collection and screening of the literature sources
The authors searched for relevant sources and inherent documents, reviewed the contents and analyzed them. Namely, a critical literature review was carried out (Snyder, 2019; Torraco, 2005). The sources were assessed for their relevance, and the key content was summarized. Hence, a model of evidence was created for the system under investigation (Khan et al., 2003).
This approach reduces the bias of researchers, allows detailed information about the phenomenon, promotes the grounding with the literature and supports a coherent combination of various theoretical elements with a central theoretical subject.
3.2.1 Critical literature review
The critical, or integrative, review aims to assess, critique and synthesize the literature on a research topic (Torraco, 2005). In this case, the review was driven by the aim of detecting the comment and relations of the CF system model.
The review involved a “creative” collection of data as the purpose is to combine perspectives and insights from different fields of research and different types of sources (Snyder, 2019), to identify the main theoretical parts and then interconnect them.
Setting the model boundary was crucial at this point, to circumscribe the scope of the collection of the sources. For our aim, the unit of analysis was the CF campaign. It comprehends the whole system of activities, relations and actors, involved in a single, successful CF. By project, instead, we mean the entrepreneurial initiative behind the campaign.
Our boundary is defined by the input and the output of the system to analyze; input and output are, respectively, CF platform audience and CF campaign's raised capital. The main components involved in the transformation of a platform's audience (i.e. web page visitors) in the campaign's raised capital are included in the system's boundary. Other relevant elements outside of the model boundary are parts of the outside environment. The focal variable of the analysis is raised capital (see Figure 6). This is a function of the time, whose initial value is zero at the kickoff of the campaign, and the final value is achieved at the expiration date. The campaign is successful if the capital raised at the expiration date is equal to or over the target.
Transparency and documentation of the process of review are required: how the integrative review was done and how articles were selected (Bryman and Bell, 2011; Snyder, 2019) are reported in the following sections.
3.2.2 Collection, screening and selection: creating the list of sample sources
In this phase, the scholarly relevant Internet search engine Google Scholar (https://scholar.google.com/) was used to search refereed papers published on recognized international journals or selected conference proceedings.
To collect and select the relevant and coherent papers to be included in the list L of theoretically significant sources, the search was conducted on December 7th, 2021. The aim was to identify and collect a significant number of sources, which brought any relevant element to the CF campaign model system, with respect to the model boundary. To this aim, no restrictions and filters were applied to the search whose entries were “crowdfunding” AND “dynamics” OR “system” OR “campaign”. To secure a scientific reliability of the data, only journal and conference proceeding articles were accepted (see Table 1).
The sources that resulted from the search were reviewed by abstracts and then sorted in order of relevance and coherence. A full review of the sources was then performed, processing the items per batch. The sample list of sources was stopped when the authors considered theoretically significant the set of data against the CF campaign model (theoretical saturation). A list of 23 articles (see Table 2) was then agreed as exhaustive to cover the CF campaign's system elements (see Table 3).
3.2.3 Critical review and extraction of the model data
The key elements emerging from the sources were reported on a review form, and all the reviews were summarized and reported in a model boundary diagram (Haji Gholam Saryazdi et al., 2020; Sterman, 2000). A model boundary diagram shows the elements to be included in the model (Sterman, 2000).
3.3 Interpretation, composition and presentation
Once agreed on the model boundary diagram, the data were combined according to CLD modeling notations and construction guidelines (Bala et al., 2017; Lane, 2000; Schaffernicht, 2007; Sterman, 2000). Model building and analysis were supported with software for system dynamic modeling. Thus, CLD was used to both analyze the data and present the results. Visualizations were preferred to mitigate the difficulties of analyzing complex systems (Ammirato et al., 2022).
4. Findings
4.1 “Crowdfunding campaign” system's structure
The findings show a set of 26 system components that, together with their mutual influence, determine the behavior (i.e. raised capital) of a CF campaign.
Literature addresses the effects of single components on CF campaign behavior, thus without consideration for “side-effects” among components. These are, instead, key determinants for the system's behavior according to the system dynamics perspective.
The 26 components, according to a system perspective, are part of six subsystems. The existent literature does not focus and address explicitly this intermediate level, between the components and the system, theoretically linking operational variables with overall CF outcomes. The six intermediate subsystems that we observe are founders, funders, projects, platforms, campaign audience and platform companies. The following attributes are found to be linked with the campaign's success: founders' worth, funders' worth, project worth, platform quality, campaign audience and platform company worth.
4.2 Components and causal relations
The CF campaign components and inherent cause–effects relations are presented in this section per subsystems.
Founders' worth: Founders' worth is a first subsystem that influences with a positive link polarity the CF campaign: if the founders' worth rises, the success ratio of campaigns rises. Founders' worth is made up of the following:
the founders' direct network
the founders' social network
founders' reputation and trustworthiness
founders' experience in CF campaigns
founders' history
third-party endorsement
The founders' direct network of people (i.e. family and friends) affects the founders' worth; in practice, it is the group of backers who generally support the campaign in the early stage (Mollick, 2014). This variable is crucial for bringing the campaign to a wider audience on the web CF platform. The rise of this variable in the number will correspond, according to CLD building guidelines, to a rise in the founders' worth (positive link polarity).
The founders' social network (i.e. followers of Facebook, LinkedIn and personal web pages) is the amount of the social network contacts of the founders (i.e. the number of contacts). This variable is fundamental to scaling up the audience, just after the opening of the call. This variable is a cause (positive link polarity) of founders' worth (Belleflamme et al., 2014; Mollick, 2014).
Founders' reputation and trustworthiness are acknowledged key elements of founders' worth (Belleflamme et al., 2015; Koch and Siering, 2015; Mollick, 2014) linked to this latter with a positive link polarity.
Founders' experience in CF campaigns is recognized as a cause of founders' worth establishing – according to CLD notation – a positive link polarity (Jinwook and Lee, 2015; Koch and Siering, 2015). The campaigns in which founders have backed other projects before show a positive relation with CF campaign success (Koch and Siering, 2015). In terms of the causal link, if the founders' experience in CF campaigns rises, either as a backer or founder, the founders' worth rises, and the success of the campaign rises too (positive polarity).
A branch of studies has focused on the influence of the founders' history on the success of the CF campaign (Koch and Siering, 2015; Zvilichovsky et al., 2015). Koch and Siering (2015) show that the worth of founders' history is a cause influencing the success of CF campaigns thus linked with positive polarity. The rise of the worth of the funder's history, both personal and professional, implies the rise of the founders' worth.
Third-party endorsement is acknowledged by literature as another concurring cause of CF campaign success since the rise of it is positively linked with crowdfunded campaigns (Massa Saluzzo and Alegre, 2021).
CF platform quality: The CF platform emerges as another critical subsystem of a CF campaign, whose key attribute is quality. The literature highlights its importance to engage the users. Platform quality comprises effective and efficient operations, hosting and processing of all project information, data and transactions, from and to both sides of users. The platform quality is linked with a positive polarity to the raised capital (i.e. the success ratio of the CF campaign) since an increase in the platform's quality results in an increase in success of the CF campaign.
Quality of the CF platform (platform quality) is caused by
effort and specialized knowledge required to users
metadata quality
quality of graphics, styles and web page structure
duration balance
revenue share of the platform company.
Effort and specialized knowledge required by users have a negative polarity with platform quality (Koch and Siering, 2015).
The quality of platform design is a polar cause of platform quality (Lacan and Desmet, 2017). It is made up of the quality of metadata, quality of graphics, styles and web page. They influence the experience of the users, both founders and backers.
The quality of metadata is a cause of the quality of platform design. Well-designed metadata reduces users' effort and increases matching ratios between funders and fundraisers. Metadata allows the audience to browse the posted campaigns, by filtering them per category (e.g. film, dance, arts, design and technology) and get effective and clear information to ponder the decision to invest. This increases the quality of the funders' experience. Metadata quality and quality of platform design are linked with a positive link polarity.
The quality of platform settings emerges from literature as another cause of platform quality. In turn, it is caused by a set of key settings (i.e. parametrizations of the CF platform): duration and share.
The duration of the campaign is the maximum duration a campaign can have, over which the project expires. The platform company sets this constraint. If the duration rises then the fundraised increases, but a longer duration implies both obsolescence of the idea and a longer return period for funders (Beier and Wagner, 2016). The balance is a key cause of the quality of the platform setting.
Share of the Platform is the price for the intermediation, the share the company keeps on the capital raised by the crowdfunded campaigns. If the share falls, the audience rises, while the Platform's quality falls.
- (3)
The project worth: The third subsystem of a CF campaign is the project, whose attribute linked to the system's behavior is worth. Project worth is a fundamental cause of a campaign's success ratio, as pointed out by Koch and Siering (2015). The following components concur with the “Project worth”:
originality of the idea
coherence of the CF model
congruity of the target
value of return to the funder
quality of communication
orientation of the project
additional funding
quality and frequency of project updates
funders' public comments’ positivity.
These components' attributes are acknowledged for effects on the project's worth.
The originality and creativity of the project idea are searched by nonprofessional investors (i.e. people) in the projects they screen when deciding to support a CF campaign. These kinds of investors seek and choose often for engagement and fulfillment of originality (Kang et al., 2016).
Quality and frequency of project updates are acknowledged as crucial for capturing funders during the campaign timeframe. Lee et al. (2016) show the influence of project updates on CF campaign success. An increase in quality and/or frequency causes an increase in CF campaign success.
Congruity of the target is a cause that has effects on the project's worth. As (Mollick, 2014) argues the requested amount of money influences the project's funding success. The project's goal, or target, is set by founders depending on the planned required capital and the (eventual) availability of other funding sources from the third party. The target then can be very different, ranging from small amounts (e.g. < 1$) for a single initiative like an exhibit, to a proper seed capital provision for the start-up of a new venture. Mollick (2014) supports that the rise of the target over the perceived “congruous” target causes the fall of trust and confidence in the success of the campaign. In turn, the rise of the congruity of the target (that corresponds to the cut of the gap between the “perceived congruous” target and the actual target) is linked with a positive polarity to the project worth.
The quality of communication of the project is another cause linked to CF campaign success that links to project worth. Namely, if the first rises the second rises too (positive polarity). The quality of communication, in turn, is given by the combination of the quality of descriptions (Koch and Siering, 2015), depth of descriptions (Koch and Siering, 2015; Xiao et al., 2014), language used (Mitra and Gilbert, 2014) and characteristics specified (Etter et al., 2014), all regarding the way the founders present and communicate the project, through the platform. Effective use of media, meant as the effective combination of various media (video, images and texts), is another cause of the quality of communication. Each medium has its influence on the funding success of projects (Koch and Siering, 2015).
The orientation of the project is acknowledged by the literature as a positive causality with the project's worth. In other words, the orientation (level) of the project toward a collective interest or good (environment, social discrimination, etc.) is causally linked with the worth of the project. In the eyes of the crowdfunders, if the orientation of the CF project rises, the project worth rises too, thus revealing a positive link polarity from the first to the second. The literature shows, in fact, the link between the environmental and/or social responsibility of the project to the success of the CF campaign (Hörisch, 2015; Hörisch and Tenner, 2020; von Selasinsky and Lutz, 2021). Thus, a positive link polarity exists from the orientation of the project to the project's worth.
Additional funding, other than crowdfunders, is another variable linked to project worth. The optional, possible coverage of a share of capital with other financial sources (e.g. of the founders themselves or of third parties) to fill a certain gap, is linked with positive polarity to project worth.
Funders' comments positivity is another cause influencing the level of audience since it orients the decisions of the platform audience to invest or not in a given CF campaign. The supportiveness of the comments posted on the project page by those who backed the campaign influences the level of the audience and orients the decisions toward investing (or not) in the given project. It is assumed that an increase in positive comments (in the amount and content) causes a rise in the CF campaign's audience.
- (4)
Funder worth: Funder's preparation and experience is a variable with a positive causal link with success of CF campaign as reported by Jinwook and Lee (2015).
- (5)
Audience: Audience of CF campaign is the group of Internet users looking for CF campaigns to back. Of course, the audience has a positive link polarity to the number of funders, and this in turn has a positive link polarity with the success of a CF Campaign. Causes of audience are detected in
platform popularity,
audience's trust,
social media fake manipulation and
financial gap.
Platform popularity is the level of acknowledgment of a CF platform, in the CF environment (potential crowd funders). Platform popularity has a positive link polarity with the audience since the rise of popularity determines the rise of the audience. For this reason, the platform companies invest strategically in platform branding (Branzov and Maneva, 2014). Social media shares create the so-called tam-tam, thus raising the audience on a CF platform. This raises the number of funders.
The audience's trust is the confidence about the likelihood of success of a given posted project. It influences the choice of the project to fund (Kang et al., 2016). Trust can be cognitive and affective (Kang et al., 2016). If the audience's trust rises then the number of funders and capital raise.
Social media fake manipulation is a cause of mistrust toward a CF platform. As the manipulation of fake comments rises, the audience of the CF platform falls.
The financial gap is the difference between the target and the actual capital raised. (Wonglimpiyarat (2018) points out that this gap is a cause of an increase or decrease in the audience of the campaign and finally in the funders. As much as the gap falls, the audience's confidence in the success of the campaign rises, and the risk of failure reduces. These signals boost backers in funding the campaign (both in pledges and in speed). As much as the gap falls, the funding (both in pledges and in speed) rises.
CF company worth: CF platforms are owned and managed by companies that, according to multisided platforms’ business model (Brokaw, 2014; Rossi and Vismara, 2018; Wirtz, 2019), keep a share of the capital raised by successful campaigns (Aitamurto, 2015). If the revenue share falls, then the founders rise, and since the share is kept away from the raised capital the project's success rises too.
The specialization and the service offered to users by platforms' companies, being able to reach a large audience, causes engagement and reuse of the platform (Aitamurto, 2015; Rossi and Vismara, 2018).
Combining and summarizing the most relevant findings coming from the literature review, the data set for the CLD model comes out, by pointing to the correspondent (short) name for CLD and the effect (i.e. the link polarity of each link) (see Tables 4 and 5).
4.3 Modeling crowdfunding with causal loop diagrams
The data confirm a set of 26 components (variables) belonging to six subsystems: founder worth, funder worth, platform quality, project worth, platform audience and platform company.
Assuming as the focal effect of the study the CF campaign's capital (i.e. behavior of the campaign) and connecting them according to the data set a CLD model of a CF campaign is carried out (see Figure 7).
The model shows the existence of 19 feedback loops (see Appendix), five of which are “engines” (Haji Gholam Saryazdi et al., 2020) of the CF campaign (see Table 5).
5. Discussion
CF campaign behavior (i.e. the raised capital over time) is critical to address the low CF success ratio (Mollick, 2014). To this aim, we advance the following discussion based on the CLD model's evidence.
5.1 Causal structure of a crowdfunding campaign
Despite the spread and growth of CF platforms and projects, the success ratio of CF campaigns is steady and moderate. A high rate of unsuccessful campaigns produces disengagement of the audience from CF platforms, thus hampering its growth.
Our research returned a set of 26 components of the CF campaign, grouped in six subsystems, unlike Shneor and Vik (2020) who identified 106 labels of variables, grouped into five broad categories (Shneor and Vik, 2020).
The key feedback loops (“engines of growth”) of the behavior work in specific stages of the campaign that we identify in: (1) kickoff, (2) scale-up and (3) maturity.
The interpretation of the results is supported and informed by the concept of system archetypes (Senge, 1990) or common modes (Sterman, 2010). They are paradigms, mental models and frames of reference to interpret the dynamic phenomena that emerge from the observation of reality. They provide the rationale for a given system's behavior, linking the trend of a focal variable (behavior) to the system's causal structure. This way it is possible to analyze or predict a system's behavior over time.
5.2 Stage 1 – start
The start of the campaign is characterized by two reinforcing and interconnected feedback loops: R1: direct network support and R2: indirect network support.
5.2.1 R1: direct network support
The “Start” is the early stage of the CF campaign, characterized by the support of the founders' direct network (i.e. family and friends) (Agrawal et al., 2015), which is also the main audience of the campaign's web page. The driver for supporting the campaign is the direct and personal relationship between founders and funders, rather than the return of the investment, or the platform's audience.
Family and friends, by the mean of word of mouth and social media networks, extend the campaign's audience to the founders' indirect network (i.e. the friends of friends). They start backing the campaign and increasing the capital, driven by the worth of the founders (Troise and Tani, 2021), namely reputation and trustworthiness.
5.2.2 R2: indirect network support
As soon as the direct network mobilizes to back the campaign, the indirect network starts being involved. Founders' indirect contacts (social network contacts, friends of friends and indirect relatives) are informed of the campaign by the direct network and avail social and geographic proximity (Agrawal et al., 2013, 2015) to assess project and founders' worth and finally decide to back the campaign (see Figure 7).
R1 and R2, according to the CLD (Figure 8), are two interlocked reinforcing loops. These two interlocked, reinforcing loops, amplify or dampen in time. Two interconnected reinforcing feedback loops lead back to the system archetype “Success to the successful” (Senge, 1990). This archetype concerns the situation in which two actors compete for common and finite resources. The one who first prevails in acquiring more resource than the other acquires, reinforces his/her success over the other, who is weakened over time until it disappears (see Figure 9).
In light of the aforementioned archetype, direct and indirect networks compete to acquire a share of the capital, which is (theoretically) limited by the target. Since both networks are functional and need to support the capital raising it should be avoided an overcapitalization from one of them. Since a share of the target capital is always pledged by founders' direct and indirect backers, the acquisition from “direct” and “indirect” backers must be appropriately balanced to avoid the success of one over the other and to preserve the investment capacity for both.
This finding fits with an established behavior in the nascent CF literature, in that CF projects are subject to critical mass phenomena (Agrawal et al., 2013; Giudici et al., 2013). Projects that succeed in raising a sizable amount of capital in the early days of a CF campaign (i.e. the start stage) are fated to succeed, while those that fail to attract contributions in the early days are fated to fail.
5.3 Stage 2 – scale-up
In this stage, R1, direct network support and R2, indirect network support have run out of supportive power. This stage of the CF campaign's behavior, shifts the audience (and the backers) from the founders' network, to the worldwide Internet audience of the CF platform. The driving force of the campaign's behavior is the quality of the investment project. This is given by the return's (i.e. compensation for the funding) value and by the worth of the project. The worth of the project comprehends a set of dimensions (orientation, quality, originality and congruity), assumed by the campaign's audience.
This aforementioned shift is much more evident with the project size of the CF campaign: small projects' capital is little affected by quality, while the capital of large projects is influenced by project quality (Burtch et al., 2011).
In this stage, then, the behavior of CF is driven by the worth of the investment and of the project. The ability to scale up from the founders' network to a crowd of funders comes from the quality of the project. This mechanism is critical for the following behavior of the campaign, in order to make it a “crowd” phenomenon. Quality of the project, indeed, captures a key group of backers, which provide a valuable contribution to the campaign capitals and to the campaign's credibility for the following wave of backers.
Even in CF settings driven by altruism and moral return (i.e. social CF projects or donation-based models) success of CF rises for higher-quality projects (Burtch et al., 2013). Thus, the relevance of project quality on campaign's behavior is stably positive (i.e. if project quality rises, then raised capital rises) across types of projects.
According to the CLD model (see Figure 10) two reinforcing feedback loops operate and influence the campaign's behavior: R3: skilled crowdfunders and R4: newcomers and bystanders.
5.3.1 R3: skilled crowdfunders support
Crowdfunding campaign's behavior is driven by a share of the platform's audience, with investor attitude, that seeks, compares and ponders the possible alternate campaigns, before to decide. This group of backers is driven by rational analysis of the campaign's likelihood of success (the financial gap and coherence of the target), the project quality (value of the return on funding, idea originality, congruity of the target and quality of project communication), and the founders' worth (namely founders' history and third-party endorsement). The likelihood of success is the main driver of decision for these potential backers, who ponder the platform's value too (i.e. share of the platform's company).
This R3 reinforcing loop is in line with the well-known Matthew effect. High-quality projects attract more (skilled) investors/funders, whose investments attract other more funders (Macht and Weatherston, 2014; Xu et al., 2016; Vasileiadou et al., 2016). This reinforcing feedback loop is known as Matthew effect (Mollick, 2014).
5.3.2 R4 newcomers and bystanders’ support
R4 is a reinforcing feedback loop interconnected with the R3, caused by newcomers and opportunistic lenders (bystanders); they cover a part of the capital, when the investment risk is low, while evaluate the return and the financial gap (they take advantage of the financial coverage of the precedents and the reasonable certainty that there will be a last group of lenders who will not miss the opportunity to successfully reach the target).
Allocation of the project's capital must be balanced between the two interconnected R loops, which trace back to the system archetype success to the successful. The overallocation of capital to one of these types would produce pathological negative effects. It is necessary to break the link between the two feedback loops, detaching the allocation of capital from the lender's experience.
5.4 Stage 3 – maturity
This stage is characterized by the funding of followers, who put in place a herding effect (Beier and Wagner, 2016). A crowd of funders, who rely on the pioneering investments of the skilled funders, crowd the campaign according to the “herding effect” (Beier and Wagner, 2016).
5.4.1 R5 – “herding” effect
This stage's behavior is characterized by a reinforcing feedback loop (R loop) (see Figure 11). It exponentially amplifies the initial stimulus supportiveness of funders' comments and raised capital.
The capital raised and the supportive comments posted by previous funders on the campaign's web page become new drivers of CF behavior. The campaign's audience, at this stage, is driven and supported by previous backers' comments. Key signals are then the amount and positivity of comments posted on the campaign's web page and the capital gap. Depending on these two elements, new backers fund the project according to the herding effect (Beier and Wagner, 2016), that is they are motivated to participate in a community, to avoid staying aside from a campaign or rather to support an original idea or a certain orientation of the project (e.g. environmental, social, humanitarian and scientific).
This study provided valuable, although preliminary and early, evidence of the theoretical importance of the dynamics of a CF campaign. This study comes to identify the causal foundation of CF campaigns, which makes the underpinning mechanisms of CF behavior. They have been supported and argued by Kuppuswamy and Bayus (2013) who show the existence of dynamics among groups of backers, where herding effects take place between early funders and bystanders.
5.5 Managerial implications and limitations of the study
CF campaign causal structure is a theoretical antecedent of CF dynamics, in particular of the trend of raised capital over time.
The audience, the funders and ultimately the capital of the campaign, naturally increase or dampen in time, depending on the causal structure. Ignoring the rationale and mechanisms underpinning CF behavior makes it hard to manage it and increase the success ratio of campaigns.
Implications from this study suggest that the allocation of capital among the various stages and types of backers must be balanced; this allows switching smoothly across the stages and the groups of backers, from those starting the capital funding to those closing the campaign. It is also suggested to control the allocations of the capital available for funding, between the groups concurring in the same stage (start and scale-up) since they interact according to the archetype “success to the successful.” These system archetypes suggest that the excessive relative allocation of capital to one of the two groups would produce pathological negative effects (i.e. the disappearance of the unsuccessful group). Both groups in both stages, however, are needed to pass all the developmental stages of the campaign, so the existence of a theoretical balance of the funding between the group of backers and a theoretical ratio among the capital raised emerges across the stages.
Alternatively, it is suggested to break the interconnection between the feedback loops, to detach the allocation of capital between groups of funders.
The aforementioned implications are useful for the managers of CF platforms, seeking to increase the performance of the platform; the rise of the overall platform success ratio goes through a higher campaign success ratio. Hence, they should translate the highlighted implications into proper settings and adjustments of platform structures and operational parameters.
Policymakers as well, seeking to foster effective policies in support of CF as an economic flywheel (Harris and Wonglimpiyarat, 2020), can take useful knowledge to bypass the system's resistances, to consider counterintuitive behaviors (Forrester, 1971; Sterman, 2000) and exploit systems' leverage points (Chen et al., 2018).
Despite the perspective and the scope of this study on CF, not circumscribed to just the project and enclosed into a specific model of CF, it is not free of limitations. All the literature sources grounding the model are useful for a preliminary exploration, but a wider and systematic review may provide further elements on the CF structure. Future development of this study line would involve quantitative modeling.
6. Conclusions
This paper, by keeping a holistic, system approach to the CF campaign, contributed to discovering and analyzing the causal structure underpinning the behavior of CF campaigns.
Linking the causal structure with the focal variable raised capital over time, we identified and reported the mechanisms by which groups of backers interact and influence each other during the development of the campaign. Motivated by a common goal (fulfilling the target capital) different groups of backers (the direct network, the indirect network, skilled funders, newcomers and bystanders), according to a system archetypes' interpretation, compete in the acquisition of shares of the target capital. Both in the start and the scale-up stages of the campaign overallocation to one of the groups would weaken the other, thus compromising the stage's capital share. This view also suggests the existence of pondered shares of capital across the three stages, whose balanced achievement brings to the concluding herding effect reinforcing loop.
The Extant literature mainly focuses on CF projects, with empirical data of single projects and in the boundary of one specific CF model (donation, landing, reward and equity) following that theories and implications for management remain within the descriptive and operational level.
This paper approached a system-level exploration, to provide an understanding of the relationship between causal structure and behavior of CF campaigns; this investigation is useful to prevent unbalances or dampening trends in the capital rising, rather than make funding from backers a self-reinforced cycle. These objectives are keys to addressing agreed diseconomies of CF campaigns (e.g. many CF projects fail, while many others overfund) (Shneor and Vik, 2020; Yang et al., 2015).
On this premise, by the means of a DMB methodology, we come to the causal structure of the CF campaign. It revealed 26 campaign system components and attributes, belonging to six subcomponents with attributes: founder worth, funder worth, project worth, platform quality, platform audience and platform company.
Building a CLD we provided evidence of the complex structure of a CF system. Interactions of feedback loops show the mechanisms underpinning the dynamics of CF. The extant literature is, in contrast, mainly focused on a suboptimal-level analysis.
As stated by Mollick (2014) “there is substantial value in further studying the dynamics of crowdfunding, since it sheds light on a variety of subjects of interest to academics and policymakers, with implications for entrepreneurial financing, the role of individual quality and networks in venture success, and the importance of geography in new ventures” (p. 2).
Despite a relatively consistent corpus of literature, the CF theory is in its infancy (Shneor and Vik, 2020): a quantitative orientation of the existent studies and a recursive empirical analysis of data from a few dominant CF platforms reveal that current knowledge is a far from being mature for markets, players, concepts and behavioral patterns (Shneor and Vik, 2020). This paper synthesized and connected results from existing studies, employing qualitative-oriented studies aiming toward theory development and conceptual fine-tuning thus forwarding research according to research directions addressed by Shneor and Vik (2020).
