Digital platform dynamics: exploring the paradoxical role of financial literacy in crowdlending adoption Open Access

Crowdlending, often equated with peer-to-peer (P2P) lending, is an alternative financing model wherein individuals or enterprises raise funds from multiple independent lenders online (Ferrer et al., 2023; Pena and Breidbach, 2021). By leveraging technology, these platforms connect individuals seeking capital (borrowers) and those supplying it (lenders). These platforms broaden and enhance efficiency in accessing capital. By bypassing traditional financial intermediaries, crowdlending platforms enable borrowers, particularly SMEs and individuals, to access financing at competitive rates while allowing lenders to achieve potentially higher returns, albeit with greater exposure to default risk (Ribeiro-Navarrete et al., 2021).

Crowdlending has become an increasingly visible component of the global fintech ecosystem. However, evidence from recent reports suggests that growth in digital lending has been uneven, with 2023 marked by subdued investment and heightened investor caution rather than rapid expansion. Existing studies emphasize crowdlending's institutional role in improving access to finance, particularly for SMEs and sustainability-oriented projects, rather than documenting large-scale market growth or long-term volume projections (Berkowitz and Souchaud, 2023; KPMG, 2024). The broader crowdfunding market, valued at $153bn in 2022, is forecast to grow to $1.7tn by 2032 ^{[1, 2]}. Complementing these figures, Polaris Market Research (2023) valued the broader crowdfunding market at USD 19.86bn in 2023, while Grand View Research (2023) estimated a narrower segment at USD 2.14 bn in 2024. Each figure projects steady growth toward 2030–2032. These variations reflect differing definitions of crowdlending versus broader crowdfunding, highlighting the need for conceptual clarity when examining platform engagement.

This study focuses on borrowers, those requesting funds whose creditworthiness and repayment behavior shape platform trust and long-term viability. Their decisions on whether, when and how much to borrow are critical for understanding platform sustainability and risk management (Pena and Breidbach, 2021; Ribeiro-Navarrete et al., 2021). While most studies emphasized lenders or capital providers by examining their trust, risk perceptions and portfolio diversification strategies, our study addresses how borrowers or individuals seeking funds navigate and respond to the technological, financial and social features embedded in crowdlending platforms.

Globally, crowdlending has expanded rapidly but displays diverse regional patterns (Ferrer et al., 2023; Maier, 2016; Martínez-Climent et al., 2021; Rosavina et al., 2019). Indonesia, for instance, offers a distinctive context whereby a youthful tech-savvy population and unmet financial needs drive adoption (Perdana et al., 2021). It creates growth dynamics, unlike in Western markets. This heterogeneity highlights the need to understand how socio-technical conditions shape engagement and inform more inclusive platform designs.

The crowdlending sector in Indonesia has grown rapidly owing to the rising demand for alternative credit and a large population lacking access to conventional banking (Rosavina et al., 2019). As of July 2023, the fintech lending industry comprises 102 companies with total assets of IDR 7,062 billion, maintaining equity of IDR 3,398 bn against liabilities of IDR 3,664 bn. The Financial Services Authority (OJK) reported a 4.04% dip in monthly loan disbursements to IDR 18.73 tn in January 2023, which still marked a 35.72% year-on-year increase. Around 15.93 million borrowers, mainly from Java, received loans, with IDR 7.08tn directed to productive sectors such as trade and agriculture ^{[3, 4]}. OJK's regulatory oversight aims to sustain the sector's integrity and stability. Indonesia's high mobile penetration, uneven financial inclusion and evolving regulations create a distinctive context for studying borrower behavior in emerging platform economies.

Research on crowdlending has illuminated key aspects of participation, particularly how individuals transition from traditional finance, navigate trust during crises, such as the COVID-19 pandemic and balance emotional and economic motivations when seeking funds (Ferrer et al., 2023; Maier, 2016; Riahi and Garrouch, 2023). Evidence highlights that those providing capital prioritize platform quality and risk-mitigation mechanisms, such as transparent processes and fair credit assessment, over loan terms alone (Perdana et al., 2023). In emerging economies with uneven digital literacy, technological readiness and intuitive platform design are vital for encouraging participation (Okine et al., 2023). Few studies, however, have integrated behavioral, technological and contextual dimensions to explain borrower engagement comprehensively.

Our study uses affordance theory to examine how borrowers interpret platform features, transparency, convenience and accessibility and how these perceptions shape engagement (Du et al., 2019; Faik et al., 2020). The affordance theory emphasizes the user enactment of technological possibilities and explains shifts from conventional finance to platform-based lending in emerging markets. Our central research question is: What determinants shape borrowers' engagement decisions in crowdlending platforms?

Our study advances the theory and practice in three ways. First, it extends affordance theory, commonly applied in organizational settings (Leonardi, 2011; Volkoff and Strong, 2013), to alternative finance. We show how functional and technological affordances (TAs) influence engagement in crowdlending, where risk and trust dynamics differ from those in traditional information systems (IS) contexts. Second, it broadens the geographical scope of crowdlending research by examining Indonesia's fintech landscape, in which distinct socio-technical conditions influence adoption patterns. Third, we identify the affordances that build trust, mitigate risk and sustain participation. We offer guidance to platform designers and regulators in balancing innovation with consumer protection within emerging financial ecosystems.

We analyzed survey data from 254 participants using ordinal regression. Model 1 examined functional affordance (FA), TA, social influence (SI), personal factors (PF) and financial literacy (FL). Model 2 incorporated both demographic and economic controls. This methodological approach enables us to isolate the unique effects of different affordance types while controlling for alternative explanations, thus providing robust evidence for how various determinants shape crowdlending participation decisions. Although crowdlending platforms involve both lenders and borrowers, our study focuses on borrower engagement. Occasional references to lender-oriented mechanisms are included only to illustrate broader platform dynamics. All empirical measures and interpretations are grounded solely in the borrower's perspective. The remainder of this paper reviews the relevant literature (Section 2), details the methodology (Section 3), reports the results (Section 4), discusses theoretical and practical implications (Sections 5 and 6) and concludes the study (Section 7).

The business model of crowdlending, also known as P2P lending, has evolved from advancements in financial technology (fintech) and continues to reshape alternative finance ecosystems (Pena and Breidbach, 2021; Ribeiro-Navarrete et al., 2021). Originating from crowdfunding, crowdlending enables multiple individuals to contribute toward a target loan amount, bypassing traditional financial intermediaries and fostering direct borrower–lender interactions through digital platforms (Maier, 2016; Ribeiro-Navarrete et al., 2021). This model broadens access to capital for individuals and small enterprises, excluding conventional banking, while providing investors opportunities for potentially higher returns. Its efficiency stems from technological innovations that facilitate rapid loan processing, enhanced transparency and competitive interest rates (Maier, 2016; Wang et al., 2015).

Prior research highlights that crowdfunding research spans finance, innovation and social sciences, emphasizing both opportunities and challenges for entrepreneurs and investors (Camilleri and Bresciani, 2024). Their review links adoption of crowdfunding to theories such as the resource-based view, signaling and diffusion of innovations, showing how these frameworks explain the interplay between technological features and investor behavior in alternative financing ecosystems. Similarly, Mora-Cruz and Palos-Sanchez (2023) find that equity, reward and lending models dominate global research, with growing attention to technological aspects that enable platform scalability and investor engagement. Despite these insights, few studies integrate behavioral, technological and contextual dimensions into a unified explanatory model (Barbi and Mattioli, 2019; Ferreira et al., 2022; Ferrer et al., 2023).

Our study applies affordance theory, rooted in ecological psychology and developed in IS research, to examine how technological features of crowdlending platforms translate into actionable possibilities for users, specifically for borrowers. This perspective enables a dynamic understanding of engagement, bridging gaps in prior research. Figure 1 presents the research model guiding this investigation.

The concept of affordance finds its roots in Gibson's (1986) ecological psychology, where it marked a shift away from traditional stimulus–response models of perception and action. Instead of viewing behavior as merely reactive to external stimuli, Gibson argued that environments present affordances, opportunities for action that arise through the relationship between an actor's abilities and the surrounding features. Crucially, affordances are neither objective attributes of objects nor purely subjective perceptions. Instead, they are relational, unfolding dynamically at the interface of material structures, embodied capacities and situational goals.

This perspective challenges deterministic accounts that link specific technological features to fixed behavioral outcomes. As Leonardi (2011) and others (e.g., Strong et al., 2014; Volkoff and Strong, 2013) have argued, what a technology affords depends greatly on how it is used, by whom and under what circumstances. Within IS research, these insights have been extended to explore how digital technologies operate within complex organizational and social environments. Leonardi (2011), for instance, reconceptualized affordances as the potential for technology-enabled change. The affordance concept emphasizes that identical tools may support very different actions across user groups.

Markus and Silver's (2008) introduction of FAs advanced this line of inquiry by distinguishing between an artifact's material attributes and the possibilities these enable for specific users. This reconceptualization shifted attention from static technological features toward dynamic enactments of technology in practice. Later work, such as that by Volkoff and Strong (2013), emphasized that affordances are not fixed but emerge through interactions shaped by organizational routines, institutional norms and user expertise. Strong et al. (2014) further clarified these distinctions by contrasting FAs, which are closely linked to a system's technical features, with TAs, which capture broader interpretive and experiential possibilities. Affordance theory, as one of the central lenses in IS literature, offers a compelling framework for analyzing how digital tools both enable and constrain user behavior. At its core, the theory posits that meaningful possibilities for action only arise when system features align with user capabilities and intentions. In their absence, affordances may go unrealized, resulting in disengagement or ineffective use. This principle informs the analytical framework of our study.

Importantly, affordance theory has encouraged a more user-centric approach to technology evaluation and design. By focusing on what users are able to do rather than merely what systems offer, researchers have gained deeper insights into how technologies can be made more intuitive, responsive and context-sensitive (Leonardi, 2011; Strong et al., 2014). Within IS research, affordances function not only as descriptive concepts but also as explanatory tools. They help clarify how users interpret and respond to technological possibilities, how these interpretations evolve and how they influence behavior and outcomes (Dremel et al., 2020; Goel et al., 2013).

This analytical perspective proves especially relevant for digital finance platforms, where user engagement depends heavily on the ability to assess risk, navigate complex interfaces and trust algorithmic decision-making (Du et al., 2019; Faik et al., 2020). In the context of crowdlending, for example, affordance theory can capture dynamics that traditional models, such as the technology acceptance model (TAM), often miss. While TAM assumes a linear relationship between perceived usefulness, ease of use and adoption, affordance theory highlights variations in user goals, knowledge and context (Jiao et al., 2022). Two users may encounter the same platform feature, such as a loan listing or repayment schedule, but interpret and respond to it in fundamentally different ways depending on experience and motivation.

Moreover, crowdlending platforms often serve diverse user bases with contrasting goals, for instance, some lenders pursue high financial returns, while others emphasize social impact or financial inclusion (Du et al., 2019; Jiao et al., 2022). Affordance theory accounts for such divergence by foregrounding the relational nature of technology use. Key platform features, such as borrower transparency, risk disclosure and transaction design, play a critical role in shaping user trust and involvement (Jin et al., 2021; Perdana et al., 2023). Yet these same features may be empowering for some users and overwhelming for others, depending on factors such as FL and risk tolerance. This tension is particularly pronounced in emerging economies, where crowdlending is often facilitated by mobile-first platforms. While these interfaces lower entry barriers, they also introduce new challenges due to regulatory uncertainty and digital literacy gaps (Okine et al., 2023). Affordance theory helps explain these outcomes by examining not only technological attributes but also the user competencies required to engage them meaningfully.

Additionally, the theory incorporates motivational and relational dimensions of platform use. It accommodates both altruistic and profit-driven behavior, while also identifying deterrents such as complexity, security concerns or institutional mistrust (Chen et al., 2019; Wasiuzzaman et al., 2021). Central to this view is the idea that technology's value is not embedded in the platform itself but realized through use (Faik et al., 2020; Leonardi, 2011). Recent literature has sharpened this by distinguishing between perceived affordances (what users believe they can do), actualized affordances (what they manage to do) and outcomes (Meske et al., 2023; Roy et al., 2024). Affordance research has largely focused on organizational and firm-level settings, emphasizing how TAs shape managerial behavior and governance outcomes (Faik et al., 2020; Leonardi, 2011; Sun et al., 2025). Recent studies have begun to extend this perspective to fintech contexts in emerging economies by situating technology use within broader institutional and infrastructural conditions. Sun et al. (2025), for instance, show that fintech adoption affords firms strategic flexibility in managing environmental legitimacy, sometimes enabling symbolic sustainability practices. Complementary evidence from informal-sector micro-entrepreneurs highlights that fintech affordances are relational, shaped by trust in intermediaries, perceptions of state oversight and the reliability of digital infrastructure (Joseph et al., 2025). Together, these studies highlight both a gradual broadening of affordance research and its continued emphasis.

Management literature suggests that digital platform adoption and use are shaped by the coherence of the broader service ecosystem rather than by isolated technological features alone. Evidence from emerging-market firms shows that digital transformation outcomes depend on the alignment of digital technologies with institutional support, business ties and complementary organizational capabilities such as skills development and governance arrangements (Shu and Srimuang, 2025). Digital technologies create value not in isolation, but through their embedding within organizational routines, human–technology interactions and external institutional environments (Cranney et al., 2025).

Research on digital platforms in emerging markets further indicates that user engagement is influenced by more than core transactional functions. Studies of mobile payment services and digital platforms demonstrate that trust, perceived benefits and design-related attributes form part of users' overall perceptions, shaping satisfaction, continued use and behavioral intentions (Nguyen, 2025; Srivastava and Shunmugasundaram, 2025). At the same time, this literature tempers optimistic narratives of digital “leapfrogging.” Empirical evidence shows that disparities in education, age and user capabilities systematically affect how digital affordances are perceived and enacted, producing uneven adoption and usage patterns rather than uniform inclusion (Srivastava and Shunmugasundaram, 2025).

In Southeast Asian contexts, social capital and interpersonal interactions embedded in social networks further shape platform engagement by facilitating knowledge sharing, reducing uncertainty and strengthening trust (Nguyen, 2025). These insights suggest that countries such as Indonesia, characterized by rapid fintech expansion alongside institutional and capability heterogeneity, provide a theoretically appropriate context for extending ecosystem-based analyses of digital platforms.

Building on these perspectives, our study distinguishes between FAs, which refer to features that directly support financial goals (e.g., credit access, loan flexibility) and TAs, which concern usability, system reliability and interface design (Faik et al., 2020; Joseph et al., 2025; Roy et al., 2024). This distinction matters because a platform might offer strong financial incentives, but if the interface is confusing or the infrastructure unreliable, users may struggle to act on those possibilities. Conversely, a well-designed interface cannot compensate for unattractive financial offerings. In this sense, TAs can shape whether and how FAs are realized.

By integrating ecological, IS and management perspectives, affordance theory offers a sophisticated lens for understanding user behavior in digital finance. It allows researchers and practitioners to move beyond simplistic adoption metrics and toward richer, context-sensitive explanations of platform use. Ultimately, this can inform better design and regulatory strategies aimed at building trust, reducing friction and supporting more inclusive financial ecosystems (Ribeiro-Navarrete et al., 2021).

Drawing on the foundational work of Markus and Silver (2008), FAs refer to goal-oriented possibilities for action that emerge when users with particular capabilities engage with technological features. Unlike feature-based perspectives that treat system attributes as fixed determinants of behavior, FAs emphasize the relational nature of user–technology interaction. The same artifact may enable very different actions depending on users' objectives, skills and situational constraints. This relational framing is especially relevant in crowdlending, where heterogeneous actors, from borrowers, novice retail participants to experienced investors, interact with identical platform features in fundamentally different ways (Fang, 2019; Jiao et al., 2022).

Within IS research, FAs are understood as the capacities of technological artifacts to support specific actions while accounting for user goals and competencies (Markus and Silver, 2008). This perspective shifts attention away from interface aesthetics toward the broader purpose of enabling meaningful action within organizational or financial ecosystems (Benbunan-Fich, 2019). Prior literature has applied this logic across domains. Treem and Leonardi (2013), for example, showed how social media affordances reshape organizational communication and socialization, while Benbunan-Fich (2019) examined how wearable technologies support health monitoring practices. In the consumer domain, Fang (2019) demonstrated that functional features of branded mobile applications influence perceptions of competence and loyalty, underscoring how affordances shape both behavior and evaluative outcomes.

Extending this perspective to crowdlending, we conceptualize FAs as emergent possibilities for lending and borrowing that arise when users with financial objectives encounter platform features designed to facilitate these activities. Rather than treating features such as borrower profiles, risk indicators or search filters as inherently influential, this approach emphasizes how they enable specific actions relative to user intentions and financial capabilities. For example, crowdlending platforms commonly allow users to filter loan requests by amount, purpose, interest rate or credit rating (Riahi and Garrouch, 2023; Ribeiro-Navarrete et al., 2021). These mechanisms support efficient decision-making by helping users align loan selection with their risk tolerance and return expectations (Wang et al., 2015).

The underlying theoretical logic is that well-aligned FAs reduce cognitive load and transaction costs, thereby facilitating engagement. Conversely, poorly aligned affordances introduce friction that discourages participation. Importantly, affordances are not deterministic. As Leonardi (2011) argued, affordances require enactment whereby features only become consequential when they are perceived and used in ways that align with user goals and competencies. In crowdlending, intuitive navigation, effective filtering tools and clear information structures are more likely to be actualized when they match users' financial needs and experience levels (Okine et al., 2023; Rosavina et al., 2019; Shi et al., 2019). When FAs are effectively designed and enacted, they enhance satisfaction, support informed borrowing decisions and encourage sustained platform use. Accordingly, we propose:

TAs, as conceptualized by Volkoff and Strong (2013) and further elaborated by Strong et al. (2014), capture a class of affordances that extend beyond discrete system functionalities. Unlike FAs, which are tied to specific technical capabilities, TAs reflect users' perceptions of broader possibilities, for instance, how a technology may enable new forms of interaction, coordination or value creation beyond its originally intended use. This distinction is theoretically important because users frequently appropriate technologies in ways not anticipated by designers, generating emergent practices and adoption patterns.

At their core, TAs emphasize the dynamic relationship between a goal-directed user and an information technology artifact. They focus on potential actions that arise through users' interpretations of what a technology can do and how it fits their objectives and contexts (Leonardi, 2011; Strong et al., 2014). Whereas FAs foreground tangible system properties, TAs highlight the subjective and interpretive dimension of use, in which how users imagine possibilities and reconfigure technology to support their goals (Volkoff and Strong, 2013). This perspective has become central in IS research, where technological effects cannot be understood independently of user sensemaking and enactment (Fang, 2019; Jiao et al., 2022).

Empirical studies have demonstrated the relevance of TAs across diverse domains. In social commerce, perceived affordances such as interactivity, information richness and navigability significantly influence user satisfaction and engagement (Dong and Wang, 2018; Shao et al., 2020; Tuncer, 2021). Shao et al. (2020) further showed that experienced users prioritize interactivity and information quality, whereas novices focus on navigation. This illustrates how affordance perceptions evolve with experience. Beyond consumption contexts, Nelson (2017) demonstrated how digital technologies afford occupational mobility, enabling new career pathways across interconnected labor markets. In financial and organizational settings, Du et al. (2019) identified blockchain's TAs, such as automated transactions and secure credit arrangements, as key drivers of adoption across actors.

The application of TAs is particularly salient in fintech, where digital systems often enable opportunities that extend beyond transactional efficiency. Pal et al. (2021) showed that mobile payment technologies afford not only convenience and security but also reflective tracking of financial behavior, shaping both current use and future adoption intentions. These findings demonstrate that perceived technological possibilities, rather than functional features alone, play a central role in shaping user behavior.

In crowdlending, TAs reflect users' perceptions of a platform's capacity to reshape financial intermediation and relational dynamics. Beyond facilitating loan transactions, crowdlending platforms can bypass traditional intermediaries, foster peer-based trust and broaden access to investment opportunities (San-Jose and Retolaza, 2016). Features such as borrower profiles, visible credit histories and detailed loan purposes enable lenders to conduct due diligence, reduce information asymmetry and manage risk (Chen et al., 2019; Hu et al., 2019; Jin et al., 2021; Shi et al., 2019). Communication tools, real-time performance tracking and data-driven insights further enhance transparency and credibility (Perdana et al., 2023).

Crucially, these affordances are interpreted heterogeneously. Experienced users may value analytics and portfolio diversification tools, while less experienced users prioritize secure transactions and ease of use. When users perceive TAs as empowering and aligned with their broader financial goals, they are more likely to engage with the platform (Jiao et al., 2022; Li et al., 2023; Luo et al., 2023). This perception enhances trust, perceived legitimacy and sustained participation. Building on this logic, we argue that TAs, manifested through real-time information, secure transactions and mobile accessibility, positively influence borrower engagement on crowdlending platforms. Accordingly, we propose:

The influence of social networks on financial decision-making is well established in social network theory and behavioral finance. Granovetter's (1985) concept of social embeddedness challenges the assumption that financial decisions are purely individual or rational, arguing instead that economic actions are shaped by ongoing social relationships. This perspective is especially salient in crowdlending, where trust, credibility and information asymmetry play central roles. In such contexts, peer recommendations, shared experiences and visible endorsements provide critical signals that shape borrowing and lending decisions under uncertainty (Martínez-Climent et al., 2021; Perdana et al., 2023).

Behavioral finance further clarifies how SI operates through distinct but interrelated mechanisms. Ouimet and Tate (2020) identified three primary channels, i.e., information transmission, whereby factual knowledge is shared within networks; social proof, in which individuals adopt behaviors perceived as common or accepted; and normative pressure, which encourages conformity to group expectations. In digital financial environments, these mechanisms frequently reinforce one another, amplifying peer effects on individual behavior.

Recent research indicates that digital platforms can intensify users' reliance on social and heuristic cues when decision environments become information-dense. Although greater information availability is often assumed to improve decision quality, evidence from crowdfunding contexts shows that excessive or poorly structured information increases cognitive load, leading to decision fatigue and reduced willingness to engage (Liao et al., 2025). Under such conditions, individuals are more likely to shift away from effortful, analytic processing toward low-effort strategies that rely on heuristics, emotional responses and salient social signals (Zhu et al., 2023). This dynamic is particularly relevant in digital financial settings where users must evaluate complex trade-offs under uncertainty. While greater financial knowledge can improve individuals' ability to process information, emerging evidence suggests that financially literate users respond to complex digital financial environments with more cautious and selective engagement rather than heightened activity, as they are better able to recognize risks and information overload (Liao et al., 2025). Hence, these findings highlight how platform complexity shapes engagement not only through information provision, but through its interaction with users' cognitive capacity and FL.

On crowdlending platforms, SI manifests in both interpersonal and platform-mediated forms. Lenders often rely on experiences shared by family, friends or colleagues as proxies for platform reliability (Benlian et al., 2012; Shive, 2010). Borrowers, in turn, may draw on endorsements from their networks to enhance legitimacy and attract lenders. Prior studies show that visible social interactions, such as ratings, comments and endorsements, significantly increase trust and accelerate adoption in online settings (Hoegen et al., 2018; Söllner et al., 2016).

These dynamics align with social proof and social learning theories (Bandura, 1977). When facing uncertainty, individuals observe the actions and outcomes of others, particularly those they perceive as credible or similar. In crowdlending, positive experiences shared within networks can encourage participation, whereas negative feedback can deter engagement (Caglayan et al., 2021). Beyond personal networks, platform communities and social media forums further amplify these effects by enabling collective evaluation of risks and opportunities (Gonzalez, 2019). Taken together, these arguments suggest that SI plays a central role in shaping crowdlending engagement. Accordingly, we propose:

PFs in technology adoption are anchored in Bandura's (1982) self-efficacy theory, which posits that individuals' beliefs in their ability to perform specific tasks strongly influence their motivation and behavior. In crowdlending contexts, self-efficacy manifests as users' confidence in navigating digital platforms, interpreting financial information and making informed lending or borrowing decisions amid inherent risks and information asymmetries (Shneor and Munim, 2019). This framework is highly relevant because crowdlending requires autonomous judgment and active participation from both lenders and borrowers.

Self-efficacy theory provides clear theoretical logic for why PF should influence engagement, in which individuals with higher confidence in their abilities are more likely to attempt challenging tasks and persist through difficulties (Shneor and Munim, 2019). In crowdlending, this translates to a greater willingness to navigate complex platforms and make financial decisions. IS research has consistently confirmed the role of self-efficacy in technology adoption. Venkatesh et al. (2003) identified this as a central predictor of technology use through the Unified Theory of Acceptance and Use of Technology (UTAUT). A more recent study by Lin and Hsieh (2023) extended this understanding to financial technologies, showing that self-efficacy underpins willingness to engage with complex platforms that demand both technical and financial competence.

In crowdlending, PF, including self-efficacy, technological familiarity and knowledge of P2P mechanisms, critically influence trust and engagement (Chaouali et al., 2017; Zainab et al., 2017). These factors stem from prior experience, perceived competence and broader confidence in managing financial decisions (Kang et al., 2021; Rieder et al., 2021). Users with higher self-efficacy are better positioned to evaluate platform features, assess risks and leverage opportunities for returns, echoing findings in related domains, such as mobile banking and cryptocurrency adoption (Arli et al., 2020; Changchit et al., 2017).

Empirical studies have highlighted that individuals with stronger technological proficiency and familiarity with online transactions are more inclined to actively engage in crowdlending. Giovanis et al. (2019) found similar patterns in mobile banking, in which prior experience with digital financial services shaped the intention to adopt emerging technologies. This suggests that, in crowdlending, users with prior exposure to e-banking or fintech innovations may transition more readily to P2P lending environments.

Beyond technical skills, general confidence in handling financial technologies fosters positive attitudes toward adoption. A clear understanding of crowdlending processes, including risk assessment, repayment structures and borrower screening, reinforces users' capacity to make prudent investments or borrowing decisions (Zainab et al., 2017). This multidimensional interplay of self-efficacy, knowledge and confidence underscores that successful participation in crowdlending hinges not only on platform usability but also on individual readiness and competence.

Prior research suggests that individuals with stronger PF (higher self-efficacy, greater confidence and more relevant experience) are more likely to engage with crowdlending platforms because they feel capable of managing associated risks and complexities. Consequently, understanding these PF is crucial for scholars and platform designers seeking to enhance user engagement, trust and long-term participation in crowdlending ecosystems. Hence, Hypothesis 4 is proposed:

FL research is grounded in human capital theory, which conceptualizes knowledge and skills as forms of capital that enhance decision-making and improve financial outcomes (Huston, 2010). However, FL may produce paradoxical effects in alternative financial services such as crowdlending. Drawing from Lusardi and Mitchell's (2014) framework, FL comprises declarative knowledge, understanding financial principles, procedural knowledge and the ability to apply these principles in decision-making. This dual perspective is crucial for understanding why financially literate individuals might engage with crowdlending platforms differently than less literate users (Greenberg and Hershfield, 2019; Karki et al., 2023). We argue that the theoretical logic of the effect of FL on crowdlending engagement involves two competing mechanisms that create a paradoxical relationship. Higher FL should theoretically increase engagement through better risk assessment capabilities but simultaneously decrease engagement through heightened risk awareness and the availability of superior alternatives.

This paradox arises from two competing mechanisms. FL should enhance the evaluation of crowdlending opportunities, enabling individuals to assess risks, returns and borrower profiles more effectively. However, greater financial sophistication may heighten awareness of risks and opportunity costs, potentially leading to more cautious engagement. Behavioral finance research supports this dynamic. It suggests that sophisticated investors often exhibit higher risk aversion and adopt more conservative financial strategies (Greenberg and Hershfield, 2019; Hoffmann and Otteby, 2018).

FL equips individuals with the expertise to interpret financial data, evaluate their repayment capacity and compare crowdlending with traditional investment options. This capability fosters informed decision-making in various contexts, from debt management to retirement planning (Zhao and Zhang, 2021). In crowdlending specifically, financially literate lenders may exhibit selective participation, choose only high-quality opportunities and avoid impulsive investment behavior. Similarly, borrowers with higher literacy are more likely to assess their ability to repay loans and to avoid unsustainable borrowing practices (Chan et al., 2022). The theoretical expectation, however, is that the risk awareness effect dominates the capability effect. Financially literate individuals are more likely to recognize that crowdlending typically offers lower risk-adjusted returns compared to diversified portfolios and involves higher information asymmetry risks than traditional financial products.

Conversely, inadequate FL can have negative effects. Borrowers with limited knowledge may underestimate their repayment obligations, making them more susceptible to overborrowing and high-interest loans, thereby perpetuating debt cycles (Skiba and Tobacman, 2008). This highlights the broader societal implications of FL gaps, particularly in emerging fintech ecosystems, where regulatory frameworks are still evolving.

Empirical research underscores the role of FL in digital financial service adoption and responsible usage (Chan et al., 2022; Hoffmann and Otteby, 2018; Lusardi and Mitchell, 2014). Within crowdlending, literacy functions as both an enabler and a moderating factor. It facilitates responsible engagement while tempering participation among those aware of risks. This nuanced role suggests that financial education initiatives can promote the more informed and sustainable use of crowdlending platforms. The theoretical prediction is that greater FL leads to more conservative behavior and reduced borrowing engagement with alternative financial services such as crowdlending. Based on these arguments, we propose Hypothesis 5:

To address our research questions by examining the relationships between affordances, social factors and borrower engagement in crowdlending, we collaborated with a prominent Indonesian crowdlending platform to conduct an online survey using Qualtrics. This partnership was strategically chosen to ensure direct access to active borrowers, thereby enhancing the ecological validity of the findings. The survey guaranteed participants' anonymity and facilitated access to a diverse respondent base, following the guidelines established by Moore and Benbasat (1991).

Our instrument development followed a systematic three-phase approach, designed to ensure construct validity and reliability. First, we identified relevant constructs from the existing literature, compiling measurement items for affordances based on Du et al. (2019), Faik et al. (2020), Lin and Hsieh (2023), Luo et al. (2023) and Seidel et al. (2013). The selection of affordance theory as our theoretical foundation was driven by its proven ability to explain user-technology interactions in financial contexts, directly addressing our first hypothesis. For SI, PF and crowdfunding engagement constructs, we drew from Eckhardt et al. (2009), Lin and Hsieh (2023), Okine et al. (2023), Perdana et al. (2023) and Venkatesh et al. (2012). FL items were adapted from Lusardi's (2019) Big Three framework, chosen for its widespread validation and relevance to financial decision-making contexts, central to our study.

Second, rigorous refinement procedures were implemented to eliminate redundancy and ensure item clarity. Each construct was measured using multiple items on an 11-point Likert scale ranging from “strongly disagree” to “strongly agree”. This extended scale was selected to capture the nuanced attitudinal variations that are essential for ordinal regression analysis. To mitigate common method bias, we followed the recommendations of MacKenzie et al. (2011) and Podsakoff et al. (2003, 2012), including item randomization, reverse coding and temporal separation of predictor and outcome measures.

Third, we conducted expert validation with four specialists, i.e., two Indonesian fintech practitioners and two academics from Australia and Singapore, ensuring both practical relevance and theoretical rigor. A pilot study with 30 participants confirmed the survey's comprehensibility and user friendliness. Demographic variables (age, sex, education, risk preference, occupation and income) were included as control variables to account for alternative explanations and enhance the robustness of our findings.

Ordinal regression was selected as our primary analytical approach because our dependent variable is ordinal and cannot be meaningfully analyzed using linear regression, which assumes equal intervals between the categories. Unlike linear methods, ordinal regression respects the ranked structure of responses and estimates how predictors shift the probability of higher versus lower engagement levels (Harrell, 2015; Kamariotou and Kitsios, 2022).

Crowdlending engagement was measured using an 11-point Likert scale (such as “Not at all”,“Extremely Rarely”, “Rarely”, “Occasionally”, “Sometimes”, “Moderately Often”, “Frequently”, “Often”, “Very Often”, “Almost Always”, “Always”) (see  Appendix). In ordinal regression, this scale produces ten thresholds, each representing a cut-point between adjacent categories (e.g., “Not at all | Extremely Rarely”). These thresholds are standard in cumulative logit models and preserve ordinal ordering, while enabling meaningful interpretation of transitions between response levels (Harrell, 2015; Kamariotou and Kitsios, 2022).

This model relies on the proportional odds assumption, which posits that predictor effects remain constant across all thresholds. In practice, this means a single odds ratio describes the shift in engagement probability regardless of where the threshold lies. Where this assumption is violated, results highlight that predictor effects vary across engagement levels rather than indicating a flaw in the variable itself. Ordinal regression aligns with our goal of analyzing hierarchical crowdlending engagement levels, using coefficients and odds ratios to test how affordances and social factors differentially influence participation across ordered categories (see Table 1).

The control variables were incorporated to isolate the unique effects of FA, TA, SI, PF and FL on CE. Demographic attributes, such as AGE, SEX and EDU, have been shown to shape both trust and participation in crowdlending (Jünger and Mietzner, 2020), while INCOME and OCCUP influence perceived access to capital and risk preferences (Rosavina et al., 2019). PRIS is a critical factor moderating platform adoption decisions (Chen et al., 2019; Emekter et al., 2015) and LENGTH affects how users enact affordances over time (Leonardi, 2011; Okine et al., 2023). Including these controls ensures that the hypothesized relationships are not confounded by contextual or experiential heterogeneity among respondents, strengthening the internal validity of our ordinal regression results.

This study had several limitations. First, the cross-sectional survey design constrains causal inference; longitudinal studies should better establish temporal relationships between variables. Second, reliance on self-reported measures may introduce bias, particularly when assessing FL and engagement behaviors. We mitigated this using validated multi-item scales, although social desirability bias may persist. Third, sample representativeness is limited, as data were drawn from users of a single Indonesian platform; findings may not generalize to other cultural or platform contexts and self-selection may bias results toward more engaged or technologically adept users. Fourth, common method variance remains possible owing to the single-source design, despite the mitigation efforts. Employing mixed methods or objective behavioral metrics could strengthen future research. These limitations do not undermine the study's contributions, delineate the scope of interpretation or highlight opportunities for further empirical validation.

Of 394 respondents, 254 useable responses were retained after excluding incomplete surveys, straight-line patterns and failed attention checks (see Table 2 for the detailed demographic statistics). While we distributed 500 questionnaires, achieving a 78.8% initial response rate, the final useable response rate was 52.8%, reflecting rigorous data quality standards rather than methodological weakness. Nearly half of the respondents (47.2%) were aged 26–35, indicating the prevalence of crowdlending among young adults, followed by those aged 36–45 (30.3%) and 46–65 (8.7%). Men comprised the majority (70.5%) of the users, with women accounting for less than one-third. Educational attainment was evenly split between high school and higher education (diploma, bachelor's degree or above). Most respondents (65.7%) were employed in the private sector, with smaller proportions of entrepreneurs (17.7%), civil servants (4%) and unemployed individuals (1.2%). Risk preferences were balanced across risk-averse, risk-neutral and risk-tolerant categories, suggesting diverse attitudes toward financial risk.

Regarding income and spending, 61% earned below IDR 9.99 million monthly and 86.6% spent below that threshold, while 31.9% earned between IDR 10–20.99 million, indicating predominance of lower-income groups. Crowdlending engagement duration revealed that 42.5% had used platforms for one to two years and 34.3% for less than a year, highlighting the recent growth. Smaller segments had three to five years (16.9%) or more than five years (6.3%) of experience, suggesting emerging user retention trends within the market.

The reliability and validity analyses presented in Table 3 demonstrate the strong psychometric properties of the measurement model. Cronbach's alpha values for all constructs ranged from 0.816 to 0.950, exceeding the widely accepted 0.70 threshold, indicating high internal consistency across items (Nunnally and Bernstein, 1994). The Kaiser–Meyer–Olkin (KMO) Overall Measure of Sampling Adequacy (MSA) of 0.95 reflects excellent sampling adequacy, while Bartlett's test of sphericity (χ² = 5196.378, p < 0.001) confirmed that the correlation matrix was factorable and suitable for structural modeling. Confirmatory factor analysis (CFA) further supported the construct validity of the model, i.e., the Comparative Fit Index (CFI = 0.922) and Tucker–Lewis Index (TLI = 0.909) met the recommended cutoffs for good model fit (≥0.90). The Standardized Root Mean Square Residual (SRMR = 0.045) falls well within the 0.08 threshold and although the Root Mean Square Error of Approximation (RMSEA = 0.088) is slightly above the ideal 0.08, it remains acceptable, given the model's complexity and sample size (Pituch and Stevens, 2016).

Ordinal logistic regression with a proportional odds model relies on several assumptions to produce valid estimates (Harrell, 2015; Pituch and Stevens, 2016). The first requirement is that the dependent variable is ordinal and categories follow a meaningful order without implying equal spacing. Our crowdlending engagement variable meets this criterion. The proportional-odds assumption implies that predictor effects are constant across engagement thresholds. Tests indicate that FL and length of crowdlending use (LENGTH) violate this assumption, indicating that their effects vary across levels of engagement. These violations reflect genuine heterogeneity rather than measurement issues; accordingly, both variables were retained and their coefficients interpreted with caution. In substantive terms, this suggests that FL and usage duration influence engagement differently at lower versus higher levels, rather than exerting a uniform effect. Robustness checks using logit, probit and complementary log–log (cloglog) models yield consistent effect directions and significance levels, confirming the stability of the findings despite non-proportional effects.

The model also assumes no perfect multicollinearity among predictors; none should be exact linear combinations of the others (Harrell, 2015). Variance Inflation Factor (VIF) diagnostics (see Table 4) show all predictors below the commonly accepted threshold of five, with FA highest at 3.02, indicating no serious multicollinearity concerns. Additionally, the absence of significant outliers or high-leverage points was verified to prevent an undue influence on the coefficient estimates. These diagnostics, combined with proportional odds testing, enhanced the transparency and robustness of our statistical analysis. They ensure that interpretations drawn from the regression results are methodologically sound and provide confidence in the subsequent theoretical and practical implications derived from the findings.

The data reveal notable correlations among the crowdlending platform variables (Table 5). A strong positive correlation exists between FA and PF (r = 0.738), indicating that higher FA aligns with more favorable personal perceptions. This highlights how users' financial evaluations influence their broader platform experience. Similarly, FA correlated positively with TA (r = 0.632), suggesting that users who find the platform functionally supportive also recognize its technological adequacy. These findings highlight the intertwined roles of usability, cost and technological suitability in shaping user engagement. The results of the correlation matrix are presented in Table 5.

The relationship between FA and SI (r = 0.553) implies that borrowers who perceive high FA are also more likely to be influenced by social factors. This highlights the potential role of social and peer influences on individual perceptions and behavior regarding online credit. In contrast, the data show several statistically significant negative correlations between FL and FA, TA and PF, with correlation coefficients ranging from −0.346 to −0.411, implying that positive perceptions of functional facilitation, technological facilitation and PF associated with crowdlending platforms decrease with increasing levels of FL. This provides a nuanced understanding of the interaction between financial knowledge and technological use.

Moreover, the positive correlation between the duration of online credit use (LENGTH) and age (AGE) (r = 0.194) suggests that older individuals have been using online credit platforms for longer periods. By contrast, a negative correlation between monthly income (INCOME) and AGE (r = −0.221) suggests that older users tend to have lower monthly incomes, indicating possible age-related financial trends within the user base.

We also found that those who have higher monthly expenditure (EXP) tend to have been engaged in the platform for a longer time (LENGTH) (r = 0.240), suggesting an association between spending habit and tendency to borrow funds in crowdlending platforms. Finally, a strong negative correlation between monthly spending (EXP) and INCOME (r = −0.344) indicates that as income increases, monthly spending decreases and vice versa. This inverse relationship warrants further investigation of borrowers' financial behavior and spending patterns.

While these correlations provide an insightful understanding of variables and their relationships, it is important to emphasize that correlation is not synonymous with causation. These relationships do not imply that changes in one variable necessarily cause changes in another variable. Furthermore, the insights gained from the correlation matrix should be complemented by additional statistical measures and placed in a broader research context to gain a comprehensive understanding.

This model includes five variables: FA, TA, SI, PF and FL. All the predictors had significant effects on the dependent variable (p < 0.05) (Table 6). The positive estimates suggest that as these variables increase, the likelihood of moving to a higher category of the outcome variable increases. Notably, FL has a negative relationship, implying that, as FL increases, the odds of falling into a higher category decrease. Their coefficients can be interpreted as follows: for every one unit increase in the given variable (while keeping others constant), the log-odds of being in a higher frequency category of crowdlending engagement (e.g. moving from “Rarely” to “Occasionally”) increases by the amount of that variable’s estimate. For example, an increase in FA by one unit (with an estimated value of 0.391) increases the log probability of being in the higher category of crowdlending exposure by 0.391. However, it is important to note that FL has a negative impact on the outcome variable; that is, an increase in FL decreases the probability of being in a higher category.

This model included the same five variables as in Model 1, in addition to AGE, SEX, EDU, OCCUP, PRIS, INCOME, EXP and LENGTH. Compared with Model 1, the significance and direction of the effects for the variables included in both models largely remained the same, with some slight variations in the estimates and significance levels (FA, TA and SI were slightly less significant but still meaningful at p < 0.10). FL is conceptually vital, in which Greenberg and Hershfield (2019), Hoffmann and Otteby (2018) and Lusardi and Mitchell (2014) demonstrate its paradoxical role, improving evaluation and heightening risk aversion. This explains the negative coefficient observed, as financially literate individuals may engage cautiously, despite understanding the mechanisms and potential benefits of crowdlending.

Of the additional variables, only EDU and INCOME were significant (p < 0.05). EDU has a negative relationship, suggesting that higher education levels are associated with a lower category of the outcome variable. Similarly, higher income levels (INCOME) were associated with lower engagement levels AGE, SEX, OCCUP, PRIS, EXP and LENGTH were not statistically significant in Model 2, suggesting that these factors did not meaningfully contribute to the prediction of the outcome variable. Although statistically non-significant, LENGTH was retained for its theoretical importance. Prior studies show that engagement duration shapes perceptions of trust and platform affordances, as experienced users enact possibilities differently than newcomers (Leonardi, 2011; Okine et al., 2023). Controlling for LENGTH captures user heterogeneity rather than omitting the core affordance factor.

Interval estimates add an interpretive layer to our analysis by clarifying the baseline log-odds of transitioning to higher levels of crowdlending engagement when all predictors are set to zero. For instance, the interval “Not at all|Extremely Rarely” in Model 2 yields a log-odds value of −6.851. This means that, without any influence from functional or demographic variables, the likelihood of borrowers moving from no engagement to minimal engagement is very low. As borrowers have been using the platform more frequently, they are more likely to engage more. The statistical significance of these thresholds helps to determine which shifts between categories are meaningful, highlighting where predictors exert the greatest influence.

When predictor variables are incorporated, we can assess how they alter the thresholds. Positive coefficients (e.g. FA, TA, SI and PF) suggest that enhancing these aspects can encourage higher engagement. Conversely, negative coefficients (e.g. FL, education and income) indicate potential dampening effects. The model fit improved when control variables were added, as shown in Table 7. McFadden's R² increased from 0.271 to 0.301, and both Cox and Snell and Nagelkerke increased by 0.038. Such gains, moderate to strong for ordinal logistic models, justify including controls, capture additional variance without overfitting and strengthen the explanatory power of the research model.

The robustness checks presented in Table 8, using three link functions (logit, probit and complementary log–log), show that the direction and relative magnitude of the estimated effects remain stable, although the statistical significance of some predictors (e.g., FA) varies slightly across specifications. Across all models, PF and FL consistently emerged as the strongest predictors, with PF positively and FL negatively associated with the outcome, with high statistical significance (***). TA and EDU also retain significance across specifications, while INCOME remains negatively significant in all the models. Minor predictors such as FA and SI showed weaker but generally consistent effects and variables such as AGE, SEX and PRIS remained non-significant, suggesting their limited explanatory contribution. Importantly, no major sign reversals occur, and the effect magnitudes are comparable, indicating that the results are not sensitive to the choice of link function. This consistency strengthens the credibility of the findings and reassures reviewers that the identified relationships, particularly the roles of PF, FL and EDU, are robust to alternative ordinal modeling assumptions and not to artifacts of a single specification.

Our findings offer empirical support for the relevance of affordance theory in understanding consumer behavior on fintech platforms, while also encouraging a re-evaluation of some key assumptions within conventional technology adoption frameworks. Both FA (β = 0.313, p < 0.01) and TA (β = 0.292, p < 0.01) show statistically significant positive relationships with crowdlending engagement. This aligns with affordance actualization theory, which posits that meaningful engagement arises when users perceive a match between their goals and the technological capabilities available to them (Anderson and Robey, 2017; Strong et al., 2014).

Building on earlier IS research that focused primarily on organizational settings (Strong et al., 2014; Volkoff and Strong, 2013), our study suggests that individual borrowers evaluate fintech platforms not only for the financial solutions they offer but also for the quality of the user experience, namely, whether the platform is accessible and easy to navigate.

The fact that both types of affordances, i.e., functional and technological, have comparable effect sizes indicates that each plays a distinct and necessary role. Strong functionality, if poorly implemented, does not seem to drive engagement, nor does a user-friendly interface make up for limited financial utility. This supports the broader argument in digital transformation theory that value is co-constructed through an interplay of capabilities, rather than emerging from any single feature or component (Cranney et al., 2025; Shu and Srimuang, 2025; Srivastava and Shunmugasundaram, 2025). In this way, user engagement appears to stem from the coherence of the broader fintech ecosystem rather than from isolated attributes.

We also observe a more nuanced picture when applying affordance theory to financial decision-making. While FA and user-specific factors remain strong predictors of engagement, consistent with prior claims that technology must align with users' goals and competencies to be effective (Leonardi, 2011; Strong et al., 2014), the influence of TA diminishes once control variables are included in the model. This may reflect the idea that, in high-stakes financial environments, the perceived ease or appeal of a platform matters less than the concrete financial advantages it provides, building on insights from Du et al. (2019).

One of the more surprising findings relates to FL, which shows a significant negative association with crowdlending participation. This result contradicts human capital theory, which generally predicts that increased financial knowledge leads to higher adoption rates (Huston, 2010). Instead, our data echo findings in behavioral finance, where greater financial sophistication is linked to increased risk sensitivity (Hoffmann and Otteby, 2018). In our case, borrowers with higher FL may be more attuned to the potential risks and trade-offs involved in crowdlending compared to more traditional financial services. This challenges the common assumption in the fintech literature that greater literacy always promotes usage (Chan et al., 2022). A summary of hypothesis testing results is provided in Table 9.

Equally surprising is that higher education and income levels negatively predict engagement, contradicting typical technology adoption patterns (Venkatesh et al., 2012). This aligns with observations by Camilleri and Bresciani (2024) and Mora-Cruz and Palos-Sanchez (2023), which show that demographic influences and adoption drivers vary across regions and platform types, underscoring the need for contextually sensitive theoretical frameworks when analyzing crowdfunding behavior. These unexpected patterns reveal that the traditional technology acceptance models may inadequately capture behavior in alternative financial contexts where risk awareness moderates adoption decisions, suggesting important theoretical refinements for understanding financial technology engagement among sophisticated user populations.

This study contributes to affordance theory and the crowdlending literature by both reinforcing its explanatory power in financial contexts and revealing patterns that call for theoretical refinement. Our findings confirm that affordance theory offers a robust lens for understanding technology-mediated financial behavior, while also exposing dynamics that are not fully captured by existing adoption frameworks.

We extend affordance theory from its traditional organizational focus to alternative finance by showing how FA and TA shape borrower engagement in crowdlending platforms (Leonardi, 2011; Volkoff and Strong, 2013). The positive effects of FA (H1 supported) and TA (H2 supported) align with prior IS research demonstrating that well-designed affordances enable meaningful user action (Faik et al., 2020; Strong et al., 2014). At the same time, the weaker effect of TA in the controlled model suggests that affordance actualization in financial decision-making contexts is more contingent than previously theorized. Our finding extends Du et al.'s (2019) work by showing that perceptions of technological possibilities are mediated by individual capabilities and situational risk considerations rather than platform design alone.

The most theoretically provocative result concerns FL, which exhibits a significant negative relationship with crowdlending engagement (H5 supported). This contradicts human capital theory's assumption that knowledge uniformly promotes participation and aligns instead with behavioral finance research showing that financial sophistication heightens risk awareness and aversion (Greenberg and Hershfield, 2019; Hoffmann and Otteby, 2018; Huston, 2010). Consistent with Lusardi and Mitchell's (2014) paradox, FL appears to both enable informed evaluation and discourage participation in alternative financial services. This challenges prevailing fintech assumptions that literacy straightforwardly drives adoption and suggests that knowledge may generate more conservative behavioral patterns that existing models fail to anticipate (Choung et al., 2023).

The finding that FL negatively correlates with crowdlending participation (β = −0.264, p < 0.05) presents a noteworthy challenge to common assumptions in both financial inclusion discourse and technology adoption theories. Rather than implying avoidance of digital financial tools, the result points to a more discerning approach enabled by financial knowledge. We suggest three possible mechanisms that may help explain this counterintuitive relationship.

First, the informed skepticism explanation highlights how financially literate individuals are more adept at identifying less obvious risks, such as complex terms, high effective interest rates or hidden fees. This heightened awareness can lead to more cautious borrowing behaviors, consistent with research in behavioral finance that shows financial knowledge tends to increase risk aversion rather than simply promote engagement (Feller et al., 2017). Second, the alternative access mechanism considers that individuals with stronger FL often have higher levels of education or income and therefore greater access to conventional financial services. Their lower use of crowdlending platforms may simply reflect a preference for more familiar or advantageous lending options rather than a rejection of fintech itself. Third, the stigma sensitivity explanation suggests that financially knowledgeable users might be more attuned to the social perceptions surrounding alternative lending. In some contexts, using crowdlending platforms may be interpreted as a sign of financial distress, which could discourage participation due to reputational concerns. These mechanisms imply that FL influences not just whether individuals engage with crowdlending, but how they interpret its risks, benefits and social meanings.

The confirmed positive effect of SI (H3 supported) reinforces the relevance of social network theory in financial decision-making (Granovetter, 1985; Ouimet and Tate, 2020). Extending prior crowdfunding research, our findings show that social proof remains influential even after controlling for demographic and economic factors, underscoring that financial behavior is shaped by embedded social relations rather than individual rational calculation alone (Caglayan et al., 2021; Hong et al., 2018).

Unexpectedly, higher education and income negatively predict crowdlending engagement, contradicting established technology adoption models (Venkatesh et al., 2003, 2012). This suggests that conventional acceptance frameworks may inadequately capture behavior in alternative financial settings where risk salience moderates adoption. Consistent with prior research, these results highlight that demographic effects vary across institutional and regional contexts, underscoring the need for context-sensitive theorization in crowdfunding research (Camilleri and Bresciani, 2024; Mora-Cruz and Palos-Sanchez, 2023).

Our study advances management theory in three interconnected ways. First, it extends affordance theory into consumer finance by demonstrating that affordance actualization in high-stakes environments is systematically moderated by individual characteristics such as FL and socioeconomic status, responding to calls for consumer-oriented affordance research (Yan et al., 2023). Second, it challenges linear literacy–adoption assumptions by showing that FL functions as a double-edged mechanism (e.g. facilitating platform navigation while amplifying risk evaluation), calling for contingency-based models rather than universal adoption logics (Chen and Bellavitis, 2020). Third, it contributes to emerging market digital finance theory by revealing that crowdlending adoption reflects necessity-driven participation rather than inclusive leapfrogging, tempering optimistic narratives of fintech-enabled financial inclusion (Chen et al., 2025).

While the cross-sectional design limits causal inference and the Indonesian context constrains generalizability (Rosavina et al., 2019), these limitations also point to fertile directions for future research. Longitudinal studies and cross-country comparisons are needed to trace affordance actualization over time and to examine how risk awareness reshapes technology engagement across institutional settings. Overall, the study deepens understanding of affordances in financial contexts and highlights the complex, sometimes counterintuitive ways user characteristics shape engagement in emerging fintech ecosystems.

Our findings offer practical insights for the design of crowdlending platforms and for financial service providers and regulators working to foster sustainable digital lending environments. The recommendations are organized below based on their priority, focusing first on the areas with the most immediate impact on platform sustainability.

A primary observation in this study is the strong link between FA and borrower engagement, which establishes user-centric platform design as the highest priority for managers. This highlights the importance of user-centered design. Platforms that streamline loan applications, make navigation straightforward and present borrower information in a clear manner tend to maintain higher engagement levels (Okine et al., 2023; Rosavina et al., 2019). Design strategies should aim to minimize cognitive burden through intuitive interfaces, while still incorporating reliable tools that support sound decision-making. Features like mobile-first access and responsive customer service further enhance usability by fitting into users' everyday routines.

Addressing the paradoxical role of FL represents the next level of priority, as our results indicate that different user segments engage with platforms in distinct ways. More financially literate or affluent users may approach digital lending with greater caution, indicating a more deliberate, evaluative form of engagement. On the other hand, individuals with lower FL, who engage more frequently, might be more vulnerable to risks. This places an ethical responsibility on platforms to include clear disclosures and educational tools, aligning with broader goals of responsible financial inclusion (Lusardi and Mitchell, 2014). These findings point to the value of tailoring digital experiences to different user profiles, adjusting the complexity of interfaces and decision aids accordingly.

Building trust through SI is another significant factor that managers should incorporate into their ecosystem strategies. Mechanisms such as peer recommendations, verified user testimonials and group-based lending features can help build trust and expand user networks (Caglayan et al., 2021; Hong et al., 2018). Collaborations with community-based groups or microfinance organizations may be especially helpful in markets where digital trust is still developing. However, these social features should enhance, not substitute for, strong platform usability and transparency.

From a regulatory perspective, the concentration of engagement among users who are less educated and possess less FL raises concerns regarding consumer protection. Regulatory bodies might need to go beyond simple disclosure rules, advocating for clearer communication of risks and mechanisms such as cooling-off periods before large borrowing decisions are finalized (Ferrer et al., 2023). These protections become especially important in contexts where users may lack the tools to assess lending risks on their own.

Finally, FL should be seen not only as a precondition but also as a design opportunity. Interactive tools like loan calculators, explanations of fees and personalized financial guidance can be embedded within platforms. While more financially savvy users currently show less engagement, improving the financial knowledge of all users may help encourage responsible borrowing and reduce defaults. In this way, financial education becomes a foundation for long-term platform success and trust.

Crowdlending has emerged as a prominent fintech model shaping global financial ecosystems, particularly in Indonesia, where limited access to traditional banking and a young, digitally adept population have accelerated platform adoption. This study examines decision-making in online lending by applying affordance theory from IS research to explain how users engage with crowdlending technologies.

Drawing on survey data from 254 respondents, we find that FA, TA, SI and PF are positively associated with platform use. In contrast, FL exhibits a negative relationship with engagement, indicating that more financially knowledgeable individuals' approach crowdlending with greater caution. Supplementary analysis further shows that higher education and income levels are associated with lower reliance on crowdlending platforms, suggesting more prudent borrowing behavior and access to alternative financial options. The study has several limitations. Self-reported measures may not fully capture actual behavior and the cross-sectional design limits causal inference. The Indonesian context also constrains generalizability, while the PF construct focuses primarily on confidence-related traits rather than broader characteristics such as risk tolerance or personality.

Theoretically, the study contributes in three ways. First, it extends affordance theory by showing that functional and TA operate as complementary, not substitutable, drivers of engagement. Second, it challenges human capital assumptions by demonstrating that FL has a contingent, rather than uniformly positive, relationship with fintech adoption. Third, it complicates digital leapfrogging narratives by revealing how socioeconomic stratification concentrates platform use among users with limited alternatives. Overall, the findings underscore crowdlending engagement as a multi-level phenomenon shaped by technological, social and economic forces, offering implications for both platform design and policy in emerging economies.

1. Crowdlending platforms facilitate my access to credit by reducing constraints of traditional financial services.

2. Crowdlending platforms make obtaining consumer credit easier compared to traditional financial services.

3. Crowdlending platforms offer a range of products that meet my financial needs.

4. Submitting loan applications on crowdlending platforms is convenient.

5. I can seek assistance from crowdlending customer services to understand the loan products and the application process.

6. Fintech lending saves me time in financial transactions.

7. Crowdlending platforms provide information that helps borrowers find suitable loan products.

1. Navigating and using crowdlending platforms on a web browser is easy.

2. Using crowdlending platforms on a phone is straightforward and user-friendly.

3. The fintech lending application is reliable.

4. Providers promptly fix any issues with crowdlending platforms.

5. I have not experienced technical issues while using crowdlending platforms.

1. I have friends and relatives who borrow money through crowdlending platforms.

2. The success stories of friends and relatives who borrow money through crowdlending platforms have influenced my decision to do so.

3. I ask my family, friends or colleagues for advice on crowdlending procedures to borrow money.

1. I have a sufficient understanding of financial products and crowdlending services.

2. I feel confident in conducting online transactions via mobile and desktop applications.

3. I feel confident applying for a loan through a crowdlending platforms.

Imagine that you have $1,000,000 in a savings account, with an annual interest rate of 5%. How much money do you think you will have in your account after five years if you leave it untouched?

1. Less than $1,050,000

2. Exactly $1,050,000

3. I Don't Know

Suppose the interest rate on your savings account is 1% per year and the inflation rate is 2% per year. After one year, how much money will you be able to buy in your account compared to what you can buy today?

1. More than what I can buy today.

2. The same as what I can buy today.

3. Less than what I can buy today.

4. I Don't Know

Is this statement true or false? “Buying stocks of a single company usually offers a safer return than investing in a stock mutual fund.”

1. False

2. I Don't Know

Please select the frequency of use for each of the following:

1. Applying for credit from a crowdlending app.

2. Repaying a loan from a crowdlending loan.

3. Exploring the crowdlending loan app to find information and products relevant to my financial needs.