Digital Governance Readiness and Pentahelix Collaboration in Sustainable Tourism Governance: A Serial Mediation Model in Rural Indonesia

3.1 Demographics respondents profile

The demographic characteristics of the 400 respondents from Kutagugung Village are summarized in Table 3. These details provide contextual insights into the composition of the sample, ensuring that the perspectives analyzed in this study reflect the diversity of the local population in terms of gender, age, education, occupation, and length of residence. Such demographic diversity is essential for evaluating the dynamics of pentahelix collaboration and digital governance readiness in the context of sustainable tourism governance.

Table 3. Respondents profile

The data in Table 3 show a relatively balanced gender distribution, with male respondents slightly outnumbering females. The age profile is concentrated in the productive range of 18-45 years (88%), which is significant for implementing tourism digitalization programs, as this age group is typically more adaptable to technological adoption. Education levels are varied, but the majority of respondents have completed secondary education, indicating a moderate level of formal learning that can support governance participation. Occupation data highlight that farmers and tourism workers dominate the sample, reflecting the village’s economic reliance on agriculture and tourism. Furthermore, nearly half of the respondents have resided in the village for more than 10 years, suggesting a strong familiarity with local governance issues and tourism practices.

3.2 Measurement model results

The measurement model assessment was conducted to evaluate the reliability and validity of the latent constructs used in the study. The evaluation included indicator reliability, internal consistency reliability, convergent validity, discriminant validity, and explanatory power of the structural model. The assessment followed the standard procedures commonly applied in PLS-SEM research. Indicator reliability was evaluated using outer loading values. All indicators exceeded the recommended threshold of 0.70, indicating that the measurement items adequately represented their respective constructs. Internal consistency reliability was assessed using Cronbach’s alpha and CR. As presented in Table 4, every construct returned a single final value for each reliability criterion: Cronbach’s alpha from 0.929 (M2) to 0.963 (X5), ρA from 0.930 (M2) to 0.964 (X5), and CR from 0.955 (M2) to 0.976 (X5), all above the acceptable threshold of 0.70, confirming satisfactory internal consistency reliability.

Table 4. Measurement model assessment: internal consistency, convergent validity and discriminant validity

Convergent validity was examined using the AVE. The results showed that all constructs achieved AVE values greater than 0.50, indicating that each construct explained more than half of the variance of its indicators. These findings confirm that the measurement model demonstrates adequate convergent validity. Discriminant validity was assessed using the HTMT. All HTMT ratios were below the conservative threshold of 0.85 (maximum HTMT = 0.501, Table 4), indicating that the constructs were empirically distinct from one another despite their high internal consistency. The explanatory power of the structural model, evaluated using the coefficient of determination (R²), followed a two-tier pattern rather than a uniform level across endogenous constructs: PT and Digital Governance Readiness reached the moderate threshold (R² = 0.333 and 0.332), while TS, PTE, Sustainable Tourism Governance, and Perceived Need for Digitalization fell in the weak-to-moderate range (R² = 0.198–0.247). This pattern, and its implications for how the model's contribution should be interpreted, is examined in detail in the structural model assessment reported below. The measurement model otherwise satisfies the recommended reliability and validity criteria, supporting the suitability of the constructs for subsequent structural analysis.

The results reported in Table 4 derived from this analysis not only offer theoretical contributions to the understanding of digital governance in rural tourism but also serve as practical guidance for policymakers seeking to strengthen transparency and accountability in rural tourism administration. It should be emphasised, however, that the reduction of illegal levies or revenue leakage was not measured in this study; the constructs assessed here capture the institutional and perceptual conditions under which such practices are theorised to become more difficult, not their actual incidence.

To test this directly, a robustness check was performed in which the single most redundant indicator in each construct — the item with the highest correlation to another item in the same block — was deleted and the model re-estimated on the reduced 26-item instrument (X1_2, X2_1, X3_2, X4_1, X5_2, PT_2, TS_2, PTE_2, M1_1, M2_2 and Y_2 removed). CR fell as expected (0.952–0.971) while AVE rose (0.880–0.944), confirming that the high CR values reflect indicator homogeneity and block size rather than a validity problem. More importantly, the structural results were substantively unchanged: the largest shift in any of the twenty-one path coefficients was 0.039 (TS → M1, from 0.237 to 0.276), no path changed sign, no supported hypothesis became non-supported and no rejected hypothesis became supported, and the explanatory power of the outcome construct was stable (R² for Y = 0.219 in the full model and 0.221 in the reduced model). The full instrument is therefore retained for the results reported below, since removing items would sacrifice content coverage of each construct without altering the conclusions, while the stability of the estimates under item deletion indicates that the reported relationships are not artefacts of near-duplicate indicators.

The full collinearity VIF test produced values ranging from 1.384 to 1.802 across all eleven constructs (Table 5), well below the conservative threshold of 3.3, indicating that CMB does not pose a serious threat to the validity of the structural results. The marker-variable technique corroborated this conclusion with an important qualification: after adjusting the inter-construct correlation matrix using the smallest marker correlation (r = 0.199), 8 of 55 pairwise correlations lost statistical significance, all of which involved direct, non-hypothesized associations between the Pentahelix constructs (X1–X5) and Perceived Need for Digitalization (M2) or Sustainable Tourism Governance (Y). Critically, every correlation corresponding to the model's hypothesized structural paths — X1–X5 → PT/TS/PTE, PT/TS/PTE → M1, M1 → M2, and M2 → Y — remained significant after adjustment. Taken together, these results indicate that CMB does not materially threaten the hypothesized relationships tested in this study, though it may mildly inflate weaker, non-hypothesized associations; this pattern is consistent with the uniformly high CR values reported in Table 4, which are addressed further.

Table 5. Full collinearity VIF test (common method bias assessment)

Figure 3. Structural model result

Figure 3 presents the structural model results examining the relationships among the five Pentahelix constructs, governance perceptions, digital governance readiness, perceived need for digitalization, and sustainable tourism governance. Of the twenty-one hypothesised paths, seventeen are statistically supported and four are not: two direct effects (H2d, Academic Contribution → TS; H3b, Private Sector Involvement → PTE) and two specific indirect effects (H11d, H12b), each discussed below. The five Pentahelix constructs contribute positively to the governance perception constructs, namely PT, TS, and PTE. Among the exogenous variables, government support and community participation exert the strongest effects on PT, while private sector involvement is the strongest single contributor to stakeholder trust, suggesting that institutional regulation, social oversight and business participation each strengthen a different dimension of accountability perception within rural tourism governance. Academic contribution and media influence also contribute positively, indicating that collaborative governance effectiveness depends on the interaction of multiple institutional actors rather than isolated sectoral interventions.

The structural pathways further reveal that governance perceptions significantly influence digital governance readiness (M1). PT and TS show positive effects on readiness for digital governance adoption, implying that stakeholders are more prepared to support digital governance systems when governance processes are considered open, accountable, and institutionally reliable. PTE also contributes positively to digital governance readiness, indicating that stakeholders with greater operational familiarity and tourism involvement are more adaptive toward governance innovation and digital transformation.

Digital governance readiness demonstrates a strong positive effect on the perceived need for digitalization (M2, β = 0.445), the largest single coefficient in the model. In turn, perceived need for digitalization significantly influences sustainable tourism governance (Y, β = 0.366), suggesting that stakeholders who recognise digital governance mechanisms as necessary are more likely to perceive governance outcomes as accountable and sustainable. Digital governance readiness additionally exerts a smaller but significant direct effect on sustainable tourism governance (β = 0.172, p = 0.002). Digital Governance Readiness achieved one of the higher explanatory values in the model (R² = 0.332), indicating that governance perceptions collectively play a meaningful role in shaping institutional preparedness for digital governance implementation. Sustainable Tourism Governance itself, however, achieved only weak-to-moderate explanatory power (R² = 0.219), indicating that Pentahelix collaboration and digital governance readiness are meaningful contributors to — but not the primary determinants of — governance sustainability in rural tourism destinations; this boundary is examined further in the structural model assessment below and in the Discussion.

3.3 Structural model assessment

The structural model assessment was conducted to evaluate the proposed hypotheses, examine the significance of the structural relationships, and assess the mediation effects within the governance framework. Hypothesis testing was performed using the bootstrapping procedure with 5,000 resamples to determine the statistical significance of direct and indirect effects among the latent constructs. The results indicate that Pentahelix collaboration significantly influences governance perception constructs consisting of PT, TS, and PTE. The positive path coefficients demonstrate that collaborative interactions among government, private sector, community, academia, and media actors contribute to more transparent governance processes, stronger institutional trust, and broader experiential learning within rural tourism governance systems. These findings support the argument that collaborative governance strengthens socio-institutional conditions necessary for governance transformation.

The structural relationships further reveal that governance perception variables significantly influence digital governance readiness (M1). PT exhibits a positive effect on digital governance readiness, indicating that stakeholders are more willing to support governance digitalization when administrative processes and financial flows are considered transparent and accountable. TS also demonstrates a significant positive relationship with digital governance readiness, suggesting that institutional trust reduces resistance toward governance innovation and digital transformation initiatives. Similarly, PTE positively contributes to governance readiness, implying that stakeholders with greater tourism familiarity and operational exposure possess stronger adaptive capacity toward digitally integrated governance systems.

The results additionally show that digital governance readiness significantly influences the perceived need for digitalization (M2). This finding suggests that institutional preparedness and technological capability strengthen stakeholders’ awareness regarding the importance of digital governance mechanisms in improving governance effectiveness and accountability. Both digital governance readiness and perceived need for digitalization further demonstrate significant positive effects on sustainable tourism governance (Y), confirming that governance sustainability depends substantially on institutional readiness and support for digitally mediated governance systems.

The mediation analysis further demonstrates that governance perceptions play important intermediary roles in linking Pentahelix collaboration and sustainable tourism governance outcomes. PT and TS indirectly strengthen governance sustainability through digital governance readiness, indicating that collaborative governance contributes to governance transformation by improving institutional legitimacy and governance preparedness. PTE also demonstrates a sequential mediation effect through digital governance readiness and perceived need for digitalization, suggesting that experiential familiarity supports stakeholders’ acceptance of governance innovation and digital accountability systems.

Table 6 shows that the five Pentahelix actors do not contribute uniformly to governance perceptions, and that their relative ranking shifts depending on which perception dimension is considered. For PT, Government Support exerts the strongest influence (β = 0.206), followed by Community Participation (β = 0.186) and Private Sector Involvement (β = 0.163), a pattern consistent with transparency being shaped most directly by the openness of formal authorities and by social oversight from the local community. For TS the ordering reverses at the top: Private Sector Involvement is the strongest contributor (β = 0.198), ahead of Government Support (β = 0.143), while Academic Contribution is the only Pentahelix actor whose effect is not statistically significant (H2d, β = 0.068, p = 0.163, 95% CI [−0.029, 0.160]). For PTE, Government Support is again the strongest predictor (β = 0.172), whereas Private Sector Involvement does not reach significance (H3b, β = 0.085, p = 0.089, 95% CI [−0.012, 0.184]). The two non-significant paths are informative rather than incidental: academic actors, whose contribution operates through training and evidence rather than through day-to-day interaction, do not measurably shape interpersonal trust, and the private sector's role in shaping experiential familiarity is weaker than its conventional association with direct visitor service delivery would suggest.

Table 6. Direct effects of the Pentahelix constructs on governance perceptions

Table 7 indicates that the governance-readiness chain is strongest at the M1→M2 stage (β = 0.445), substantially exceeding the strength of any other stage in the model. Digital governance readiness is therefore a far stronger predictor of perceived digitalization need than governance perceptions (PT, TS, PTE, β = 0.215–0.271) are of readiness itself. The effect of M1 on Y is transmitted through two routes of comparable magnitude: a direct effect (β = 0.172, p = 0.002) and an indirect effect through M2 (0.445 × 0.366 = 0.163), giving a total effect of 0.335 and a variance-accounted-for ratio of 48.6%. This corresponds to complementary partial mediation: perceived need for digitalization carries about half of the influence of readiness on governance outcomes, and the two routes reinforce rather than substitute for one another. The practical implication is direct: infrastructure and capability alone (the direct route) deliver only about half of the achievable governance impact, and the other half depends on whether stakeholders come to regard digitalization as necessary.

Table 7. Direct effects along the digital governance readiness chain

Table 8 confirms this mediation pattern at the level of individual actors. Because the sampling distribution of a product of coefficients is skewed, significance is assessed by whether the 95% bootstrap confidence interval excludes zero rather than by the normal-approximation p-value. On that criterion 13 of the 15 specific indirect effects are significant. The two exceptions complement the direct-effect findings above: H11d (Academic Contribution → TS → M1 → Y, 95% CI [−0.0009, 0.0090]) and H12b (Private Sector Involvement → PTE → M1 → Y, 95% CI [−0.0005, 0.0087]) are not significant, mirroring the non-significant direct paths H2d and H3b from which they are constructed. Each actor therefore has a pathway of influence that fits and one that does not: academic contribution is better directed toward strengthening transparency (via PT, where H10d is significant) than toward trust, while the private sector is more effective through the transparency and trust pathways than through PTE. The generally small magnitude of the indirect effects (0.003–0.010) is mathematically expected, since each is the product of three sequential coefficients, and is consistent with the weak-to-moderate R² for Y (0.219), reinforcing that Pentahelix collaboration and digital readiness are meaningful but not singularly dominant contributors to sustainable governance outcomes.

Table 8. Indirect (serial mediation) effects of the Pentahelix constructs on sustainable tourism governance

Interpretation of Table 9, The R² values reported in Table 9 exhibit a consistent two-tier pattern rather than a uniform level of explanatory power across the model. PT (R² = 0.333) and Digital Governance Readiness (R² = 0.332) fall at the moderate threshold, whereas TS, PTE, Sustainable Tourism Governance and Perceived Need for Digitalization (R² = 0.198–0.247) cluster in the weak-to-moderate range. This pattern is not incidental. PT and Digital Governance Readiness occupy the first receiving position at each stage of the model: transparency is predicted simultaneously by all five Pentahelix actors operating in a domain closely tied to their visible collaborative behaviour, and digital governance readiness aggregates three converging governance-perception inputs at once. Constructs positioned further along the causal chain consistently explain less variance, indicating that explanatory power attenuates as the model moves from proximal collaborative inputs toward distal governance outcomes.

Table 9. Coefficient of determination (R²) for the endogenous constructs

This attenuation carries a substantive implication rather than being a limitation to note only in passing: it indicates that Pentahelix collaboration and digital governance readiness function as necessary enabling conditions for sustainable tourism governance rather than as sufficient determinants of it. For the outcome construct specifically, an R² of 0.219 means that approximately 78 percent of the variance in governance outcomes is attributable to factors outside the present model, most plausibly enforcement capacity, budget allocation, regulatory consistency, and local political dynamics, none of which the survey instrument captured. This value is interpreted as weak-to-moderate throughout the manuscript, and the same Chin (1998) thresholds are applied consistently to every endogenous construct in Table 9. The value supports the claim that Pentahelix collaboration and digital governance readiness are meaningful contributors to sustainable tourism governance. It does not support the claim that they are its primary drivers, and we do not make that claim in the discussion that follows.

Two further results in this manuscript corroborate the modest explanatory power for the outcome construct. First, the partial-mediation pattern reported in Table 7, in which roughly half of the effect of digital governance readiness on sustainable tourism governance is transmitted through perceived digitalization need rather than directly, implies that variance carried forward from the original predictors is attenuated at each successive mediating stage; a low R² for the final construct in a two-step mediation chain is therefore expected on structural grounds alone. Second, the uniformly high CR values reported in Table 4 (0.955–0.976) indicate that the indicators within each construct are highly homogeneous; this narrows the measured variance available to each construct and may itself constrain how much of that variance downstream predictors can explain, meaning the modest R² for the outcome construct likely reflects both the absence of unmeasured governance drivers and characteristics of the measurement instrument.

These results define the boundary of what this study can claim. The model demonstrates that Pentahelix collaboration builds governance perceptions and digital governance readiness effectively, and that readiness translates efficiently into perceived digitalization need. It demonstrates less completely how that readiness translates into actual governance outcomes on the ground, and we treat that gap as a substantive finding rather than a shortcoming to be minimized. It is also the empirical basis for the framing adopted throughout this manuscript: the evidence supports Pentahelix collaboration and digital governance as mechanisms that strengthen the institutional conditions for accountable tourism governance, not as mechanisms shown here to eliminate extortion or revenue leakage outright.

3.4 Discussions

The structural results indicate that governance perceptions mediate the link between Pentahelix collaboration and digital governance readiness. Fourteen of the fifteen paths from the Pentahelix constructs (X1–X5) to PT, TS and PTE are positive and significant (β = 0.068–0.206; R² = 0.333, 0.247 and 0.224 respectively), and these three perception constructs in turn explain a moderate share of Digital Governance Readiness (R² = 0.332), with PT (β = 0.271), TS (β = 0.237) and PTE (β = 0.215) each contributing independently. This pattern is consistent with governance literature describing rural tourism destinations as prone to “shadow governance” — informal coordination and undocumented transactions operating alongside formal structures [1, 5]. In Kutagugung Village, fragmented monitoring and limited digital oversight plausibly sustain this vulnerability; the moderate rather than strong R² values suggest that collaboration alone does not fully account for governance perceptions, and that institutional control mechanisms outside the model likely also matter [8].

Digital Governance Readiness shows a strong effect on Perceived Need for Digitalization (β = 0.445, R² = 0.198), and Perceived Need for Digitalization is the stronger of the two direct predictors of Sustainable Tourism Governance (β = 0.366) compared with the direct effect of Digital Governance Readiness (β = 0.172; R² = 0.219). The weak-to-moderate R² for sustainable tourism governance indicates that readiness and digitalization need jointly explain only about one-fifth of the variance in governance outcomes, with the remainder driven by factors outside the model such as policy enforcement intensity, budget constraints, and political will. This sequencing — readiness must precede perceived necessity, which then precedes governance outcomes — echoes technology-readiness findings elsewhere [29, 30, 32] and reinforces that governance digitalization in Kutagugung Village is a socio-institutional process rather than a purely technological one.

The results also caution against treating digitalization as automatically sufficient for governance legitimacy. Because Perceived Need for Digitalization (β = 0.366) outweighs the direct effect of Digital Governance Readiness (β = 0.172) as a predictor of sustainable governance outcomes, and carries 48.6% of the total effect of readiness on that outcome, stakeholders' perception that digital tools are necessary appears to matter at least as much as institutional readiness itself. Digital systems introduced without stakeholder buy-in therefore risk underutilization or resistance, consistent with prior socio-technical governance findings [6, 23, 24, 28]. Policymakers should accordingly prioritise building perceived necessity through transparent communication of governance gaps alongside technical readiness, rather than assuming infrastructure investment alone will shift outcomes.

The five Pentahelix actors contribute complementary governance functions: policy and regulatory support from government, investment and service delivery from the private sector, local legitimacy from community participation, knowledge transfer from academia, and information dissemination and public accountability from media and digital actors (Figure 4). Read alongside the path results above, these roles operate as enabling conditions for governance perceptions rather than as direct guarantees of reduced extortion or leakage, underscoring that Pentahelix collaboration strengthens the institutional groundwork for accountability without, on its own, eliminating informal revenue practices.

Theoretically, this study extends the sustainable tourism governance literature by testing collaborative governance and digital governance readiness within a single serial-mediation model, rather than treating them separately as most prior studies have done. Empirically, the comparatively low R² for the outcome construct suggests the model captures the socio-institutional antecedents of governance sustainability more successfully than it captures the outcome itself, a boundary condition that future models should address by incorporating additional predictors such as enforcement intensity, budget allocation, and political turnover.

Practically, four implications follow directly from the path results. First, government–private sector coordination, which showed the relatively strongest contribution to transparency and trust, should be prioritized. Second, community and academic involvement should be expanded to strengthen legitimacy and institutional learning. Third, media and digital actors should be integrated systematically into governance strategy, given their role in shaping PT. Fourth, because perceived need for digitalization outpredicts readiness in explaining governance outcomes, communication and change-management efforts, not infrastructure spending alone, should accompany any digital governance rollout.

This study has several limitations. The cross-sectional, single-destination design limits generalizability, and governance perceptions and digital readiness may shift as tourism systems mature and technology adoption deepens. Extortion and revenue leakage were addressed conceptually as the motivating governance vulnerability rather than measured through dedicated behavioural items, and respondent role was not recorded as a discrete Pentahelix category, which precluded testing whether these structural relationships differ across stakeholder types. A further limitation concerns the measurement instrument itself: each construct was measured with three or four closely specified items drawn from a single response format, producing CR values at the upper end of the acceptable range. Although discriminant validity was confirmed, such homogeneous item sets narrow the variance available within each construct and may inflate the estimated associations between them, and future replications would benefit from more heterogeneously worded and behaviourally anchored item pools. Future research should incorporate direct indicators of illegal-levy frequency, adopt stratified sampling by Pentahelix role to enable multi-group comparison, and employ longitudinal and multi-destination designs to test whether the governance-readiness sequence identified here holds as tourism systems and digital infrastructure mature.