Second-Hand Consumption and the Circular Economy: Exploring the Factors Driving Sustainable Consumption Among Young Consumers

2.1 Perceived behavioral control

According to the TPB, PBC is one of the three key factors explaining the formation of an individual's intention and serves as a precursor to their actual behavior. In several studies related to sustainable consumption, like green product consumption, Yadav and Pathak [9] demonstrated that the likelihood of consumers adopting this behavior increases in accordance with their perceived sufficiency of requisite conditions, time, and purchasing opportunities. Kumar [8] and Laheri et al. [10] indicated that consumers are more motivated to buy products when they believe they are capable of doing so. Moreover, research also concluded that PBC strongly influences buyers' intentions in this context [5, 11]. Regarding second-hand purchasing behavior [12], PBC refers to consumers' perception of how easy it is to purchase second-hand goods. This “perceivability of ease” encompasses the availability and diversity of pre-owned inventory, store accessibility, and the consumer's ability to evaluate secondhand items, ... Given the burgeoning prevalence of e-commerce, these convenience-related factors have become pivotal in differentiating consumer purchase intentions [13]. While online channels significantly enhance PBC by offering 24/7 accessibility and sophisticated search filters, offline channels necessitate physical mobility and a greater investment of time [14].

Therefore, the author proposes the hypothesis that consumers with a higher level of PBC are more inclined to purchase secondhand products:

H1: PBC has a significantly positive influence on consumers' INT secondhand products.

2.2 Perceive risk

PR is a state that arises in consumers when they try to avoid unforeseen consequences resulting from their shopping behavior, and it ranks among the most pivotal for elucidating and anticipating consumer actions [15]. The theory of PR has been increasingly applied in studies on consumers' green purchasing behaviors, such as recycled products [16] and green products [17]. In research on secondhand product consumption behavior, studies all indicate that PR significantly influences this consumption trend, with concerns potentially arising from various aspects such as hygiene, aesthetics, and psychology [18-20]. In online shopping environments, PR is frequently heightened due to the inherent inability to physically inspect goods prior to purchase and the potential susceptibility to fraudulent transactions [21]. Conversely, offline channels tend to mitigate these concerns by facilitating direct sensory evaluation [6]. Nevertheless, an elevated level of PR serves as a deterrent that significantly diminishes purchase intentions across both distribution channels [22]. The higher the PR, the less likely consumers are to purchase such products, this conclusion was supported by previous studies [11, 19, 23]. Building upon these findings, the author proposes the following hypothesis:

H2: PR has a negative impact on consumers' INT secondhand products.

2.3 Environmental concern

With the growing consumer interest in environmental issues [7, 24, 25], the concept of EC has garnered increasing attention. EC encapsulates an individual's awareness of environmental problems, coupled with their support for addressing them and their personal readiness to contribute to protection efforts [26]. Many authors have incorporated this factor into the TPB to explain consumers' sustainable purchase intentions [24], demonstrating that ECs drive the INT eco-friendly products such as secondhand products [8] and green products [27]. Building upon these empirical foundations, it is theoretically grounded to posit that higher levels of EC correlate with a stronger intention to consume secondhand goods. This relationship, however, may exhibit variance across distribution channels; for instance, online platforms frequently amplify the “green” dimension of secondhand consumption through targeted marketing, potentially rendering environmental motivations more salient for online shoppers than for those engaging in traditional brick-and-mortar transactions [28]. Notwithstanding potential fluctuations in magnitude and intensity, this study hypothesizes a positive influence of EC on secondhand purchase intentions across both online and offline modalities, as formulated below:

H3: EC has a positive impact on consumers' INT secondhand products.

2.4 Perceived economic benefits

PEB represent one of the primary reasons consumers choose to purchase secondhand goods online [29]. Borusiak et al. [7] further examined this phenomenon and identified three key factors influencing online secondhand shopping: price, product evaluation, and bargaining potential. The motivation behind consumers' preference for secondhand products often stems from cost-efficiency considerations [20], and Turunen and Leipämaa-Leskinen [30] emphasized that the appeal of secondhand goods lies in their perceived affordability, as they are typically priced significantly lower than new alternatives. Supporting this trend, Park and Lin [4] observed a substantial increase in secondhand sales in recent years, with economic factors being the most commonly cited rationale for this shift. Notably, the emergence of online shopping has provided consumers with novel avenues for cost optimization. While the primary advantage of offline establishments lies in the facilitation of immediate price negotiation, online platforms offer superior price transparency and the capacity for seamless comparative analysis across multiple vendors [31]. Consequently, although discrepancies may arise in the formation of purchase intentions between online and offline channels, the influence of PEB remains consistently positive across both modalities [14]. Building upon these empirical foundations, the author proposes the following hypothesis:

H4: PEB exert a positive influence on consumers' INT secondhand products.

2.5 Subjective norm

The concept of SN describes a psychological state wherein an individual's decision-making is swayed by the anticipated judgments of key individuals within their social milieu [32]. Both the Theory of Reasoned Action (TRA) and the TPB posit that SN significantly shape an individual's intentions and behaviors [33]. Rausch and Kopplin [34] argued that SN positively influence consumption intention, including pre-purchase decision-making processes. This perspective is further supported by Garas et al. [35], who conclude that opinions from individuals whom consumers perceive as personally important are more likely to influence their behavior. If consumers perceive that purchasing a particular product would be socially unacceptable, they are less inclined to buy it. Moreover, previous studies have consistently shown that SN are closely associated with socially responsible behaviors, particularly in collectivist cultures [2, 24]. This relationship was demonstrated in a study conducted in China, which found that individuals' INT second-hand clothing increased when they observed others engaging in the same behavior [36]. Furthermore, the recent proliferation of “thrift haul” culture across digital platforms has catalyzed a significant socio-cultural paradigm shift, transitioning secondhand consumption from an activity driven by economic necessity to a fashionable and socially sanctioned endeavor [37]. As social legitimacy increases, individual intentions to participate in the circular economy undergo a corresponding escalation [38]. This phenomenon provides the theoretical basis for the following hypothesis:

H5: SN have a positive impact on consumers' INT secondhand products.

2.6 Intention to purchase

The TPB posits that human intentions are formed through the careful consideration of available information. There is a significant positive correlation between intention and actual behavior [39, 40]. This perspective was previously demonstrated in the previous studies [41, 42], which emphasized that the willingness to interact with a product is a crucial precursor leading to an actual purchase behavior. Simultaneously, these studies also regard purchase intention as playing a pivotal role in guiding purchasing behavior. The strongest correlation between intention and behavior occurs when both are measured under identical conditions for each of these elements [43]. Consequently, general intentions predict general behaviors, while specific intentions determine specific behaviors [43]. In other words, intention is the most proximal and direct determinant of actual behavior.

Within the context of secondhand shopping behavior research, the conversion from purchase intention to actual behavior is anticipated to be more seamless within online environments, facilitated by the “one-click” transactional convenience [44]. Conversely, offline procurement may be impeded by spatio-temporal constraints, such as geographical distance or restricted retail operating hours [45]. However, while the magnitude of the intention-behavior conversion may fluctuate due to channel-specific determinants, the positive foundational relationship between intention and actual purchasing behavior remains invariant [42]. Consequently, this study proposes the following hypotheses:

H6: Purchase intention influences online secondhand buying behavior (ONL).

H7: Purchase intention influences offline secondhand buying behavior (OFF).

Based on the assumptions outlined above, Figure 1 presents the finalized proposed theoretical model framework.

Figure 1. Proposed research model

4.1 Validity and reliability assessment

The validity and reliability of the model were carefully assessed through various criteria. The evaluation results of Outer Loadings showed that the coefficients of all observed variables exceeded 0.7, consistent with the recommendation of the study [50]. This demonstrates a robust relationship between the observed variables and their corresponding latent variables. This finding confirms that all observed variables are suitable for inclusion in the model, ensuring content validity.

Scale reliability was evaluated based on Cronbach's Alpha and Composite Reliability [51]. Nonetheless, Hair et al. [51] pointed out that Cronbach's Alpha suffers from several drawbacks as a traditional method, potentially distorting reliability estimates. Hence, greater emphasis is placed on the Composite Reliability coefficient (rho_a), which is considered more preferable and appropriate for this study [52].

The evaluation results of Composite Reliability (rho_a) indicated that the eight measurement scales constructed in the study had Composite Reliability (rho_a) values ranging from 0.700 to 0.881, all below 0.95, meaning there was no redundancy among the observed variables [51]. Furthermore, these values were all greater than 0.7, meeting the reliability requirement, and it can be concluded that the scales exhibit good internal consistency.

Next, the author assessed the convergent validity of the eight measurement scales mentioned above. This evaluation was based on the analysis of the Average Variance Extracted (AVE) index in Table 1. It can be observed that the EC scale had the strongest convergence with an AVE value of 0.704, while the SN scale had the weakest convergence (0.527). However, all scales had AVE values greater than 0.5, satisfying the research condition proposed by authors [53].

Table 1. Reliability and validity testing

Source: Compiled by the author from survey data

Subsequently, in evaluating the discriminant validity of the scales, the author employed the Heterotrait-Monotrait (HTMT) ratio method developed by Henseler et al. [54]. The analysis results in Table 2 showed that all pairs of measurement scales in the model had HTMT values below 0.85, indicating that the latent variables shared minimal variance with each other, ensuring discriminant validity [54]. Among them, the PR-PBC scale pair had the lowest HTMT value (0.071), followed by PR-ONL (0.135), ONL-OFF (0.200), SN-PR (0.266), and PR-EC (0.267), demonstrating excellent discriminant validity.

Table 2. Discriminant validity (Heterotrait-Monotrait (HTMT))

Source: Compiled by the author from survey data

4.2 Evaluation of the structural model

The author assessed multicollinearity across the nine measurement scales by analyzing the Variance Inflation Factor (VIF). A VIF value of 5 or higher indicates significant multicollinearity that may distort structural model results, while values between 3 and 5 signal a moderate level of concern [52]. Conversely, VIF values below the threshold of 3 suggest minimal or no multicollinearity issues.

The analysis results in Table 3 revealed that all observed variables across the eight measurement scales had VIF values below 3. Among them, variable PEB2 exhibited the lowest VIF value (1.328), while PR2 had the highest VIF value (2.917), approaching but not exceeding the critical threshold of 3.

Table 3. Collinearity statistics results

Source: Compiled by the author from survey data

4.3 Hypothesis testing

4.3.1 Hypothesis testing for offline secondhand buying behavior model

As shown in Table 4, the coefficient of determination (R²) for INT is 0.518, indicating that 51.8% of the variance in personal values is explained by the five independent variables in the model: PBC, PR, EC, PEB, and SN. The offline buying behavior (OFF) variable yielded an R² of 0.250, suggesting that INT explains 25% of the variance in OFF within the model.

Table 4. f², R², and Q² coefficients results of OFF

Source: Compiled by the author from survey data

Based on the R² analysis and following the criteria established by Henseler et al. [55], the R² value indicates that the predictive power of INT for the dependent variable is relatively modest. However, in practice, R² values vary across different research contexts and disciplines; furthermore, in Partial Least Squares Structural Equation Modeling (PLS-SEM), a low R² does not necessarily imply that the model is ineffective or lacks significance [56]. The R² values in this study, hovering around 25%, may stem from the inherent complexity of factors influencing real-world shopping behavior. Consumers' decisions to purchase secondhand products offline can be influenced by numerous external factors, such as the diversity of physical stock, store location, time constraints, and perceived tangible quality. Consequently, the predictive accuracy and practical significance of the model are not diminished by these moderate R² values.

The effect size (f²) of the independent variables on the dependent variables mostly ranges from 0.012 to 0.309, indicating small to medium effect sizes. Notably, the relationship from INT (INT → OFF) exhibits the largest effect size (f² = 0.309), confirming the decisive role of intention in driving actual behavior. Regarding the antecedents of intention, PEB (PEB → INT) shows a medium effect (f² = 0.239), while PBC (PBC → INT) and SN (SN → INT) demonstrate small effect sizes of 0.088 and 0.053, respectively. This highlights the dominant influence of economic factors on consumer intention; while social relationships and personal resource control do exert influence, their contribution to consumer decision-making remains modest. At a lower level, EC (EC → INT) has a minimal effect size (f² = 0.012), suggesting its limited contribution to explaining the variance in intention. Conversely, the relationships for PR (PR → INT) and EXP (EXP → OFF) both yielded f² = 0.000$, implying that these factors do not play a significant role in the formation of intention or offline secondhand purchasing behavior within this specific sample.

Regarding the predictive relevance of the model for INT and OFF, the Q² coefficients are 0.328 and 0.161, respectively. Falling within the 0.15 – 0.35 range, these values indicate medium predictive relevance, confirming that the model possesses adequate predictive power and statistical fit.

For the structural model assessing the dependent variable OFF showed on the Figure 2, the model includes the dependent variable OFF, with independent variables comprising PBC, EC, SN, PR, PEB, and the mediating variable INT. Following statistical conventions and the guidelines [52], a p-value less than 0.05 indicates a statistically significant relationship at a 95% confidence level.

Figure 2. Structural model for OFF

Table 5. Path coefficients results of OFF

Source: Compiled by the author from survey data

As presented in Table 5, the results show that the factors positively and significantly influencing INT among youth, ranked in descending order of impact, are: PEB (PEB → INT, β = 0.426, p = 0.000); PBC (PBC → INT, β = 0.244, p = 0.000); SN (SN → INT, β = 0.193, p = 0.000); and EC (EC → INT, β = 0.088, p = 0.033). This suggests that economic benefits are the most critical factor in driving users' INT secondhand goods offline, followed by PBC, social influence, and EC, though these latter factors have a considerably weaker impact compared to economic benefits. Conversely, PR (PR → INT, β = -0.004, p = 0.897) does not exhibit a significant effect on young consumers' INT secondhand goods, indicating that awareness of potential risks does not influence their intention formation.

Furthermore, the analysis reveals that once an individual develops an INT secondhand goods, it positively affects their actual offline buying behavior (INT → OFF, β = 0.487, p = 0.000). Thus, it can be concluded that Hypothesis H7 is supported.

4.3.2 Hypothesis testing for online secondhand buying behavior model

The analysis results in Table 6 showed that the R² for INT is 0.515, indicating that 51.5% of the variance in INT is explained by the five independent variables: PBC, PR, EC, PEB, and SN. This ratio is highly consistent with the R² of the INT variable in the offline model, suggesting that the influence of these independent variables on the formation of purchase intention is relatively uniform across both online and offline contexts.

Table 6. f², R², and Q² coefficients results of online secondhand buying behavior (ONL)

Source: Compiled by the author from survey data

Conversely, the Online buying behavior (ONL) variable yielded an R² of 0.319, implying that INT explains 31.9% of the variance in ONL. The fact that the R² for ONL in the online model is higher than the R² for OFF in the offline model suggests that, in an online context, intention serves as a more robust predictor of actual consumer behavior. This disparity may stem from the unique characteristics of the e-commerce environment, where the intention-behavior gap is minimized by high levels of convenience and accessibility. In this digital landscape, consumers can convert intention into action almost instantaneously via mobile devices, facing fewer external barriers such as geographical distance, weather, or travel time compared to physical shopping. Furthermore, online secondhand platforms provide superior search and price-comparison capabilities, allowing psychological drivers, such as PEB and PBC, to exert a more direct and potent impact on behavioral outcomes. This further reinforces the assertion by Ozili [56] that R² values fluctuate depending on the research context and that a lower R² does not necessarily denote an ineffective or insignificant model.

Regarding the f², the results are largely consistent with the offline model. Notably, the relationships for PR (PR → INT) and EXP (EXP → ONL) remain non-significant in shaping intention and secondhand purchasing behavior. While other variables influence INT or ONL with a hierarchical strength similar to the offline model, distinct differences in magnitude are observed.

Specifically, although the path from INT (INT → ONL) remains the most dominant, its impact is considerably stronger than in the offline model. With an f² = 0.437, which exceeds the 0.35 threshold for a 'large effect' according to Cohen [57], this confirms that intention possesses stronger predictive power for online purchasing behavior. Regarding the antecedents of intention, PEB (PEB → INT) shows a medium effect (f² = 0.233), while PBC (PBC → INT) and SN (SN → INT) demonstrate small effect sizes (0.089 and 0.054, respectively), showing negligible deviation from the offline results. Additionally, EC (EC → INT) maintains a marginal effect size (f² = 0.012), indicating its limited contribution to explaining the variance in purchase intention.

Finally, concerning the model’s predictive relevance for INT and ONL, the Q² coefficients are 0.327 and 0.191, respectively. Although slightly higher than those of the offline model, these values remain within the 0.15 – 0.35 range, representing medium predictive accuracy. This confirms that the model maintains adequate predictive power and statistical fit.

For the structural model assessing the dependent variable ONL showed on the Figure 3, the model includes the same independent variables, PBC, EC, SN, PR, PEB, and the mediating variable INT.

As presented in Table 7, the results indicate that the factors positively and significantly influencing INT among youth, ranked in descending order of impact, are: PEB (PEB → INT, β = 0.421, p = 0.000); PBC (PBC → INT, β = 0.247, p = 0.000); SN (SN → INT, β=0.194, p = 0.000); and EC (EC → INT, β = 0.089, p = 0.038). These findings demonstrate a pattern consistent with the analysis of offline purchasing behavior. Economic benefits remain the central factor driving young consumers’ INT secondhand goods, followed by PBC, social influence, and EC—all exerting comparable effects to those observed in the offline context. Additionally, PR (PR → INT, β=-0.006, p = 0.850) again shows no significant impact on purchase intention, reinforcing the earlier conclusion. Thus, Hypotheses H1, H3, H4, and H5 are supported, while Hypothesis H2 is rejected.

However, the effect of secondhand purchase intention on actual behavior is notably stronger for online purchasing (INT → ONL, β = 0.552, p = 0.000) compared to offline purchasing (INT → OFF, β = 0.487, p = 0.000). This suggests that young consumers exhibit a greater tendency to engage in online secondhand shopping, indicating a preference for this channel over in-store purchases. Consequently, Hypothesis H6 is supported.

Furthermore, in both models (online and offline), the control variable EXP and the moderating variable PC do not demonstrate significant regulatory or control effects (PC*INT → ONL, p = 0.400; PC*INT → OFF, p = 0.808; EXP → ONL, p = 0.845; EXP → OFF, p = 0.929). Thus, PC does not moderate the intention-behavior relationship, and EXP does not exert a controlling influence on young consumers’ secondhand purchasing behavior.

Figure 3. Structural model for ONL

Table 7. Path coefficients results of ONL

Source: Compiled by the author from survey data

This study conducted a comprehensive survey of the factors influencing secondhand purchasing behavior of Vietnamese youth, which compared the differences between two consumption channels: online and offline, based on the TPB theory. The empirical results confirm that while PEB remain the primary driver, factors such as PBC, SN, and EC also play significant roles in shaping purchase intentions. Notably, the study demonstrates that the link between intention and behavior is more robust in online channels than in offline channels, reflecting the rapid digitalization of sustainable consumption. The importance of the study is clearly demonstrated by shedding light on the consumption dynamics of the young population – which accounts for a large proportion and has an increasingly strong influence in the market structure, especially in the context of the shift to sustainable consumption and the development of the circular economy.

Academically, research contributes in four main areas. Firstly, the study expands the TPB theoretical framework by integrating economic and environmental factors – which has not been explored much in previous studies in the Vietnamese context. Secondly, the study establishes the intermediary role of purchase intention in the expanded TPB model, contributing to clarifying the psycho-behavioral mechanism in sustainable consumption. Third, the study provides empirical evidence from Vietnam, a growing market that lacks in-depth research on sustainable consumer behavior. Finally, the study conducts a comparative analysis of consumer behavior by distribution channel, thereby providing a practical perspective on the characteristics of the digital and traditional consumer environment.

The application of the research is expressed on many levels. For businesses, the results of the research can be used to design marketing campaigns that emphasize the economic value of second-hand goods, and integrate social and environmental factors to improve communication efficiency. For e-commerce platforms and second-hand markets, the study suggests improving user experience, trust, and control, thereby driving actual buying behavior. For regulators, the study provides a basis for developing policies to support the formal development of the second-hand market, through a transparent legal framework, quality standards, and digital support.

However, research still has some limitations. Firstly, the data was collected using a convenient sample selection method, so it may not be fully representative of all young people in Vietnam. Secondly, the study uses a cross-sectional design, so it is not possible to determine the causal relationship over time. Third, the study only focused on the age group of 18–35 and did not consider behavioral differentiation factors by region, income, or education level. From these limitations, future studies may expand in the following directions: (1) Adopt a longitudinal study design to test the sustainability of consumer intent and behavior over time; (2) Incorporate qualitative methods such as in-depth interviews or behavioral diaries to gain a deeper understanding of second-hand consumption motivations and barriers; (3) Compare second-hand consumption behavior between different generation groups or national contexts, thereby determining how cultural factors affect consumption decisions; (4) Explore the role of emotional factors, personal identity, and the influence of the online community in second-hand consumption – especially in the rapidly evolving digital environment.

This study not only fills the gap in the literature on second-hand consumer behavior in Vietnam but also contributes to expanding the TPB theory in the new context. The experimental results and policy recommendations from the study are an important basis for promoting sustainable consumption and developing the second-hand market in a more efficient, formal, and responsible way in the near future.