Strategic partnership maintenance under geopolitical friction: a composite-indicator analysis of China–Malaysia relations, 2015–2024

1 Introduction

Strategic partnerships in international relations increasingly operate under conditions of geopolitical friction, domestic contestation, and asymmetric interdependence. These pressures do not necessarily terminate cooperation. A bilateral relationship may continue to trade, exchange diplomatic courtesies, and sign cooperation documents while the institutional channels that keep the relationship usable become weaker. The relevant problem is therefore not only whether cooperation exists, but whether the working channels that sustain it remain active or are reactivated after disruption.

This article is positioned primarily in the literature on strategic partnerships. Its central theoretical problem is that a partnership may persist as a formal diplomatic category even while the institutional channels that make it workable vary substantially over time. Hedging explains why a secondary state may sustain engagement without alignment; strategic reassurance and resilience mark adjacent concepts that should not be inferred from the index; and composite-indicator methodology supplies the means of observing change across heterogeneous channels. These literatures therefore play different roles in the argument rather than constituting parallel contributions. Partnership maintenance is defined as the observable continuity and reactivation of the economic, political, security, and societal channels through which cooperation is conducted. It does not imply harmony, strategic convergence, or full alignment.

China–Malaysia relations provide a demanding empirical application. Malaysia has maintained dense economic and diplomatic engagement with China while preserving strategic ambiguity, managing South China Sea concerns, reassessing Belt and Road Initiative (BRI) projects, and navigating major domestic political turnover. The case is therefore useful for examining whether economic density alone tracks partnership maintenance, or whether political, security, and societal channels reveal a more uneven pattern of institutional usability.

The governance relevance is institutional rather than environmental. Development-oriented cooperation depends not only on material connectivity, but also on channels that keep coordination operational during stress, clarify expectations, and prevent disagreement from overwhelming wider cooperation. SDG 16 and SDG 17 provide only reference points for effective institutions, policy coherence, and partnerships; they are not used here as the object of measurement (Tosun and Leininger, 2017; United Nations, 2015).

This combination creates a measurement problem. If a partnership can remain economically dense while becoming politically fragile, partnership maintenance cannot be inferred from trade volume alone. Nor can it be read directly from official language about strategic mutual trust. The more relevant question is whether relationship-maintenance channels remain active across economic, political, security, and societal domains, and whether those channels are reactivated after disruption.

Research on Malaysia’s relations with China has developed along several separate tracks. Kuik (2008) established the hedging framework, while Gerstl (2020) examined its operation under Mahathir. Ngeow and Jamil (2022) analyzed the renegotiation of China-linked projects. Storey (2020) traced South China Sea policy continuity, and Yeoh (2019) examined Malaysian perceptions of contemporary China. Official statements document the relationship’s formal symbolism (Ministry of Foreign Affairs, Malaysia, 2024).

These literatures generally treat economic exchange, diplomatic engagement, security friction, and societal links separately. Composite-indicator scholarship identifies normalization, weighting, and aggregation as core design choices (OECD/European Union/EC-JRC, 2008; Saltelli et al., 2007). Separate work emphasizes indicator relevance and the need to expose uncertainty through sensitivity analysis (Hák et al., 2016; Saisana et al., 2005). The empirical and methodological gap is therefore clear: a partnership may remain economically valuable while becoming politically brittle, yet existing approaches provide limited guidance for organizing annual evidence across these domains.

Three questions follow. First, how can strategic partnership maintenance be operationalized as observable institutional activity in an asymmetric bilateral relationship? Second, how did China–Malaysia cooperation vary across economic, political, security, and societal channels from 2015 to 2024? Third, which dimensions co-moved most closely with the wider relationship environment, including South China Sea friction, BRI project health, and diplomatic language?

The theoretical claim is that formal partnership status and realized maintenance activity should be kept analytically separate. The SPMCI records annual variation in the economic, political, security, and societal channels through which cooperation is conducted; formal diplomatic status is treated as context rather than as a constituent of the measure. The partnership label itself therefore does not count as evidence of maintenance. Hedging and strategic reassurance help situate why those channels matter under asymmetry and friction, but the article does not seek to revise either literature.

The same distinction defines the measurement task. The SPMCI is a formative descriptive index of realized maintenance activity, not a latent-variable measure, forecast, or causal model. It organizes auditable annual evidence and tests how the resulting trajectory changes under alternative coding, indicator-selection, weighting, and scaling choices. The China–Malaysia application examines that variation across political turnover, BRI reassessment, pandemic disruption, maritime friction, and renewed high-level engagement.

2 Literature review and theoretical framework

The framework is organized hierarchically. Strategic-partnership scholarship supplies the primary object of explanation: the persistence and practical usability of a formally designated bilateral relationship. Hedging provides the foreign-policy context in which engagement is maintained without alignment. Strategic reassurance and resilience clarify adjacent but unobserved qualities that should not be equated with the index. Governance and composite-indicator research then provide the operational vocabulary and design principles used to measure observable maintenance channels. This ordering keeps the theoretical claim distinct from the measurement strategy.

2.1 Hedging as context for asymmetric partnership management

Hedging is a central concept in international-relations scholarship on how secondary and middle powers respond to great-power pressure under uncertainty. Rather than choosing between balancing and bandwagoning, hedging combines engagement, risk diversification, insurance, and the avoidance of irreversible alignment (Kuik, 2008). This logic is especially relevant in Southeast Asia, where states often seek economic benefits from China while preserving strategic autonomy and maintaining external options.

Malaysia is an instructive case because it has cultivated dense economic and diplomatic ties with China while preserving strategic ambiguity over sensitive security questions (Gerstl, 2020). Its approach has not been one of simple alignment. It has involved pragmatic engagement, caution over the South China Sea, and the management of domestic political concerns about dependence, sovereignty, and the terms of China-linked infrastructure cooperation. Ngeow and Jamil (2022) document the domestic and economic reassessment under Mahathir 2.0, while Storey (2020) shows the continuity of Malaysia’s maritime caution.

Hedging is politically sustainable only when China-related engagement can be domestically legitimated. Malaysian leaders must present cooperation with China as economically beneficial, institutionally manageable, and compatible with sovereignty protection. The 2018 electoral transition and the Mahathir administration’s reassessment of major China-linked projects showed that domestic contestation can unsettle the legitimacy of engagement even when economic interdependence remains strong (Gerstl, 2020; Ngeow and Jamil, 2022). This domestic dimension matters for partnership maintenance: if engagement cannot be framed as useful and controllable, the working relationship becomes politically vulnerable despite material exchange. Figure 1 summarizes this framework.

2.2 Strategic reassurance and trust as adjacent constructs

Strategic reassurance and trust are adjacent constructs rather than the direct object of measurement. Strategic trust in international relations concerns structured expectations about restraint, reciprocity, and the reliability of future cooperation under uncertainty (Hoffman, 2002). While realist approaches emphasize constraints on durable trust, constructivist and institutionalist approaches show how repeated interaction, signaling, and institutional embeddedness can stabilize expectations (Wendt, 1999). In China’s diplomatic vocabulary, strategic mutual trust generally refers less to private sentiment than to the publicly manifested expectation that neither side will deliberately undermine the other’s core interests and that cooperation can continue despite asymmetry or friction (Goh, 2007). These ideas clarify why maintenance channels matter, but they do not turn the SPMCI into a measure of latent trust.

The measurement problem is that such reassurance cannot be observed directly. Previous studies have often relied on single indicators such as United Nations General Assembly voting similarity (Bailey et al., 2017), bilateral trade volumes (Kastner, 2007), or public opinion surveys (Pew Research Center, 2024). These indicators are useful, but each captures only one part of the relationship. Trade may expand while maritime suspicion increases; public sentiment may decline while official diplomacy remains active; and positive diplomatic language may coexist with operational friction. This article therefore focuses on the observable channels through which a relationship is maintained. Media can function as an instrument of soft power (Alketbi and Shariffuddin, 2025). Official diplomatic language is therefore treated here as a contextual signal rather than an independent measure of relationship quality.

2.3 Resilience as a boundary concept

Resilience is treated here as a boundary concept rather than the construct directly measured by the SPMCI. In International Relations, resilience refers to a relationship’s capacity to absorb disruption, adapt, and continue functioning while preserving core structures (Bourbeau, 2015). Adger (2000) distinguishes social resilience from ecological resilience while examining their relation. Folke (2006) emphasizes adaptation without loss of core functions. Work in public administration adds the importance of institutional flexibility under disruption (Duit, 2016). Community-level research also highlights networked capacities in recovery (Norris et al., 2008). Partnership maintenance is narrower: it records whether observable channels remain active or are reactivated without claiming that the relationship as a whole is resilient.

In the China–Malaysia case, the relevant evidence is the continued operation and recovery of economic, political, security, and societal channels while maritime friction, domestic contestation, and project-level disputes persist. Such evidence is consistent with resilience, but it does not establish resilience directly. The article therefore reports partnership-maintenance activity and treats resilience as an interpretive question requiring broader evidence.

2.4 Institutional usability and strategic partnership maintenance

Strategic-partnership research treats these arrangements as a distinctive form of bilateral cooperation whose purposes, scope, and institutional properties require specification (Wilkins, 2008). Research on China’s partnership diplomacy likewise shows that differentiated partnership categories can serve its broader alignment strategy (Strüver, 2017). Building on this literature, the present framework shifts attention from the existence or rank of a partnership label to annual variation in the working channels associated with it. A strategic partnership may remain officially designated even when the routines that sustain access, coordination, and friction management weaken. Conversely, working channels may recover without implying full alignment or the disappearance of conflict. The object of interest is therefore not whether a partnership label exists, but whether the institutional channels associated with that label remain usable under stress.

The operational distinction is between formal status and realized channel usability. Signed agreements and declared partnerships matter, but they do not by themselves establish that cooperation remains operational under stress. Formal relationship status is therefore kept outside the composite and used only as contextual information.

SDG 16 and SDG 17 are used as governance reference points rather than as measurement targets. SDG 16 emphasizes effective institutions and peaceful governance, while SDG 17 emphasizes partnerships, policy coherence, and cooperation as conditions for sustainable development (Tosun and Leininger, 2017; United Nations, 2015). The SPMCI is not an official SDG indicator and does not measure national SDG performance. It uses this governance vocabulary only to clarify why the institutional maintenance of partnerships matters under stress.

Cross-border infrastructure can create environmental risks that project volume alone does not capture (Ascensão et al., 2018). Research on BRI governance shows that institutional arrangements shape how those risks are managed (Coenen et al., 2021). In the present framework, the empirical question is whether a relationship retains enough economic exchange, political coordination, security communication, and societal connectivity to remain workable when disagreement persists.

2.5 Operationalizing strategic partnership maintenance

Governance measurement provides the operational vocabulary for translating institutional usability into observable dimensions. Indicator-based research makes complex institutional problems more visible and comparable. Le Blanc (2015) focuses on target integration, Stafford-Smith et al. (2017) examine coordinated implementation, and Tosun and Leininger (2017) place policy interlinkages at the center of SDG governance. Hák et al. (2016) address indicator relevance, while Saisana et al. (2005) examine uncertainty and composite-measurement quality. These works support the measurement logic used here: multidimensional institutional conditions require explicit design choices rather than reliance on a single proxy.

The gap is especially visible in relation to international partnerships. Partnership is not only a policy slogan; it is also a governance condition. International partnerships can support development and stability only when their working arrangements remain usable and institutionally durable under political and social stress (Leal Filho et al., 2022). A relational view of partnerships further suggests that the durability of cooperation depends on the quality of working channels, not merely on formal cooperation labels (Stott and Murphy, 2020).

A partnership-governance measurement problem therefore remains. In development-oriented bilateral relations, cooperation can be economically dense but institutionally brittle, politically active but socially contested, or formally cordial while exposed to unmanaged security friction. Measuring such conditions requires indicators that capture relationship-maintenance activity across several channels rather than relying on a single proxy such as trade, official language, or public opinion.

The SPMCI addresses this narrower measurement problem by organizing annual evidence on economic exchange, political coordination, security communication, and societal interaction within a bounded and auditable indicator structure.

2.6 Composite indicators as a measurement strategy

Composite indicators are used here as a measurement strategy rather than as the article’s primary theoretical contribution. They can summarize a multidimensional construct, but they also compress judgments about what belongs in the construct and how unlike observations should be combined. Standard guidance therefore emphasizes explicit choices about indicator selection, normalization, weighting, and aggregation (OECD/European Union/EC-JRC, 2008). Sensitivity analysis is then used to show how those choices affect the result (Saltelli et al., 2007). Critical scholarship asks a prior question: whether the resulting number has a sufficiently clear conceptual meaning. Ravallion (2012) cautions that composite indices can appear more precise than the theory behind their components and weights. Munck and Verkuilen (2002) treat conceptualization, measurement, and aggregation as distinct stages, each with its own standards. In research on interstate interaction, Goldstein (1992) shows that converting heterogeneous diplomatic events into a conflict–cooperation scale requires contestable classification and weighting decisions. Schrodt (2012) points to unresolved problems in political event data and the need for more standardized coding taxonomies.

These objections set limits on the SPMCI rather than making the exercise pointless. It is a formative, within-dyad index, not a universal ranking, and its event universes, coding rules, weights, and alternative specifications are reported so that readers can see where judgment enters. Sensitivity analysis can show whether the broad trajectory survives plausible alternatives, but it cannot establish that partnership maintenance has one uniquely correct numerical representation. The annual scores are therefore interpreted as a structured summary of the stated evidence base, with more weight placed on phase patterns and dimension-level movement than on small differences between adjacent years.

3 Materials and methods

3.1 Research design and case scope

This article uses a single-case, annual time-series measurement design to assess strategic partnership maintenance in China–Malaysia relations from 2015 to 2024. The period covers the late Najib Razak administration, Malaysia’s 2018 electoral turnover, the Mahathir 2.0 reassessment of China-linked infrastructure projects, the COVID-19 shock, intensified maritime friction, and renewed high-level engagement under Anwar Ibrahim. The 2015–2024 window is the core analysis period.

The case is analytically useful because it combines dense economic engagement with persistent political and security uncertainty. China remained Malaysia’s largest trading partner during the observation period, while Malaysia simultaneously managed South China Sea concerns, BRI-related domestic debates, and leadership changes. This combination allows the article to examine whether economic density alone tracks partnership maintenance, or whether political, security, and societal channels reveal a more uneven pattern of institutional usability.

The design is descriptive and measurement-oriented. Its purpose is to make visible the annual pattern of relationship-maintenance activity across multiple domains and to test whether that pattern behaves in substantively expected directions against plausibility-check measures. The short annual series and the institutional character of the construct make causal estimation and psychological inference inappropriate.

Accordingly, the empirical strategy is evaluated by transparency, construct coherence, and robustness to alternative specifications. The annual observations support case-specific trajectory tracing and dimension-level comparison, not population-level generalization.

3.2 Construct definition and Indicator selection

The SPMCI is a descriptive composite of realized activity in the channels that keep a strategic partnership workable. It does not estimate an underlying maintenance capacity. The four dimensions represent distinct maintenance functions: economic exchange provides material continuity; political coordination preserves authoritative access and formal cooperation; security communication records channels used for friction management; and societal interaction captures cross-border social connectivity. The index therefore summarizes observed maintenance activity within a defined evidence base rather than a latent attribute of the relationship.

The selection protocol applied four criteria. A constituent had to (1) map directly onto one of these maintenance functions, (2) be observable annually or codable under a stable rule across 2015–2024, (3) belong to a bounded source universe that could be audited and replicated, and (4) add non-redundant information relative to the other constituents. Applying these criteria produced an eight-indicator baseline, with two indicators in each dimension. Formal relationship status is retained in Supplementary File S1 for audit and comparison but excluded from the composite because a formal partnership label is conceptually distinct from realized maintenance activity.

Eight constituent indicators enter the baseline. The economic dimension contains Malaysia-reported bilateral merchandise trade and a China-related investment-flow proxy to Malaysia. The political dimension contains verified direct executive bilateral sessions and strict-core verified bilateral instruments. The security dimension contains annual activation of the Aman Youyi/Peace and Friendship exercise series and a defense-cooperation institutionalization score. The societal dimension contains Chinese tourist arrivals in Malaysia and Malaysian students studying in China.

The selection strategy prioritizes conceptual coverage within bounded and auditable institutional universes. A direct executive bilateral session is a separately documented substantive in-person meeting, telephone call, or video meeting between the Malaysian prime minister and the Chinese president or premier. Separate sessions with the president and premier during one trip are counted separately because they constitute distinct executive-access channels; multilateral co-attendance, ceremonial encounters, messages, and congratulatory letters are excluded.

A strict-core bilateral instrument is a formally signed or exchanged treaty, agreement, memorandum, protocol, implementation plan, formal letter of intent, or equivalent instrument between the two governments, ministries, central banks, national regulators, or equivalent national executive agencies. Agreements involving public research bodies, industry bodies, state media, state-owned enterprises, or state-backed project entities are retained only in a broader sensitivity series. Private commercial contracts and multilateral documents are excluded.

The military-event indicator is deliberately narrower than an all-exercise count. It records whether an edition of the China–Malaysia-initiated Aman Youyi/Peace and Friendship exercise series was held in a given year. The series began as a bilateral activity and later expanded to trilateral and multilateral editions. This bounded institutional-series rule avoids implying that the public archive captures every exercise attended by both states.

Trade and investment are retained because they capture the material exposure and continuity that help keep cooperation usable, but they are not interpreted as political reassurance. UN voting similarity is excluded because it measures preference or alignment in a multilateral arena rather than the usability of bilateral maintenance channels (Bailey et al., 2017). Elite-perception studies remain contextual evidence because comparable annual observations are unavailable for the full period (Yeoh, 2019). The same limitation applies to available public-opinion surveys (Pew Research Center, 2024). Parliamentary diplomacy is excluded because a stable, itemized annual bilateral archive could not be established and because legislative contact is not interchangeable with executive coordination. Direct executive sessions and strict-core instruments represent authoritative access and formal coordination, while Aman Youyi activation records the use of an identifiable operational-contact mechanism without implying alliance. Tourism is retained only as an indicator of short-term public-facing cross-border interaction, while student mobility captures longer-term educational embeddedness; neither is interpreted as political approval, and their sensitivity to travel restrictions is tested through an exclusion specification. Ministerial exchanges and naval port visits remain excluded because the public record does not support a sufficiently complete and stable annual ledger (Table 1).

Table 1. Theoretical mapping of SPMCI dimensions and indicators.

The table reports the conceptual role of each constituent indicator in the baseline. Formal relationship status is retained in Supplementary File S1 for audit and comparison but does not enter the composite. Full annual values, coding rules, selected source ledgers, source-coverage notes, and calculation files are supplied in Supplementary File S1.

Each indicator is normalized to a 0–1 scale using min–max scaling based only on observations in the 2015–2024 analysis period:

x ∗ it = ( x it − min t x it ) / ( max it x it − min t x it )

Dimension scores are calculated as unweighted averages of normalized indicators within each dimension:The final composite score is:

D _ d t = ( 1 / n _ d ) Σ i ∈ d x ∗ it

SPMCI t = 100 ( 0.30 E t + 0.30 P t + 0.20 S t + 0.20 C t )

The baseline weights are theory-led rather than estimated from the 10 annual observations. Within each dimension, the two constituents receive equal weight. Across dimensions, economic exchange and political coordination receive 0.30 each, while security communication and societal interaction receive 0.20 each. Material continuity and authoritative coordination are treated as the two broadest anchors in this measurement design; the resulting vector is a baseline specification, not a uniquely correct weighting structure.

Weight sensitivity is evaluated in two ways. An equal-dimension specification assigns 0.25 to each of the four dimensions. In addition, 10,000 weight vectors are drawn from Dirichlet (30, 30, 20, 20), with total concentration α₀ = 100 and random seed 20,260,702. The distribution is centered on the baseline vector (0.30, 0.30, 0.20, 0.20). The concentration setting is a design choice rather than an estimate from the data: α₀ = 100 keeps the draws as local perturbations around the theory-led vector, while the equal-weight and deterministic specifications examine larger departures. The reported 2.5th and 97.5th percentiles are weight-sensitivity intervals rather than statistical confidence intervals. Data-driven weighting methods such as principal component analysis are not used because the series contains only 10 observations and combines continuous, counted, binary, ordinal, estimated, and proxied inputs. OECD/European Union/EC-JRC (2008) provide the general design framework, and the sensitivity analysis follows the logic set out by Saisana et al. (2005).

3.3 Data quality and coding transparency

The dataset combines four evidence types. Observed annual values are used for trade, tourism, and most student observations. Itemized event records support direct executive bilateral sessions, strict-core bilateral instruments, and editions of the Aman Youyi/Peace and Friendship exercise series. One baseline constituent—defense-cooperation institutionalization—is rubric-coded. Formal relationship status is also rubric-coded but is retained only as an audit variable and does not enter the SPMCI. The three relationship-environment comparison series are coded separately and remain outside the composite. Estimated values are limited principally to the pandemic-year student series and remain explicitly flagged.

The rubric-based coding used in the baseline was conducted by the lead researcher, and the full annual series was not independently recoded by a second coder. No inter-coder reliability statistic is therefore reported. Supplementary File S1 reports the operational definition, annual values, source classes, and rubric-coded status of defense institutionalization. The robustness analysis also recomputes the index after removing that indicator. This does not eliminate coder subjectivity, which remains a limitation of the present design.

The supplementary computation package supplies a separate item-level ledger for each retained event variable. The executive ledger records 20 direct bilateral sessions during 2015–2024 and applies the same unit to official visits, summit-side meetings, telephone calls, and video meetings. The instrument ledger separates strict-core documents from broader public-institution instruments, mixed public–commercial files, commercial contracts, multilateral documents, press statements, and unresolved package totals. The exercise ledger records the 2015, 2016, 2018, and 2023 editions of the Aman Youyi/Peace and Friendship series and documents its expansion from bilateral to trilateral and multilateral participation.

Annual event values remain conservative within their stated universes. A zero for direct executive sessions means that no qualifying direct session was verified under the rule; it does not imply that bilateral diplomacy ceased. A zero for the exercise-series variable means that no Aman Youyi edition was held, not that the two militaries had no contact in any multilateral setting. A zero for strict-core instruments means that no qualifying national governmental or regulatory instrument was verified, although broader state-backed or public-institution documents may appear in the sensitivity series.

The strict-core document rule is preferred because its institutional universe is bounded and more consistently searchable across years. The broader public-institution series cannot be presented as exhaustive unless every national university, public research body, industry organization, state media organization, state-owned enterprise, and relevant state-backed project entity is searched under a predefined registry. Treating opportunistically discovered public-body documents as an exhaustive annual count would create a visibility bias.

The trade series is retained on a consistent Malaysia-reporting basis. Annual merchandise trade rises from RM230.89 billion in 2015 to RM484.12 billion in 2024. The investment variable remains a curated proxy rather than a harmonized bilateral FDI series. Public-opinion observations are not interpolated into an annual benchmark; available survey points are retained only as contextual evidence in the Supplementary material and Table 2.

Table 2. Data-quality categories and analytical treatment.

3.4 Robustness and relationship-environment comparisons

Robustness is assessed against indicator choice, weighting, scaling, archival boundaries, and pandemic-sensitive mobility. The baseline excludes formal relationship status and is compared with a status-inclusive nine-indicator specification as a sensitivity check. A leave-one-indicator-out test then omits each of the eight retained constituents in turn and recomputes the affected dimension and composite. Weight sensitivity is examined with the equal-dimension specification and the Dirichlet simulation described above. Additional specifications exclude defense institutionalization, exclude the two pandemic-sensitive mobility indicators, log-transform the inputs, rank-normalize them, and apply a pre-2020 anchored z/logistic transformation. These tests are used to identify which features of the annual trajectory survive plausible changes in construction rather than to select a single preferred alternative.

The mobility-exclusion specification addresses a narrower problem. Tourism and student mobility were directly constrained by border controls in 2020–2022, so their collapse cannot be read straightforwardly as bilateral deterioration. Rather than impute counterfactual flows, the specification removes both series and recomputes the composite over the remaining three dimensions, with the remaining weights renormalized by their sum. Executive contact is treated separately. The direct-session rule counts substantive telephone and video meetings as well as in-person meetings, so a shift to remote diplomacy during the pandemic does not by itself produce a zero. This is a stress test of whether the low period persists without the two indicators most directly affected by pandemic mobility restrictions; it is not a replacement index.

Three document-count specifications address institutional-universe and package uncertainty. The preferred baseline uses the strict-core series of 6, 6, 4, 3, 0, 1, 3, 2, 4, and 13 documents for 2015–2024. A broad public-institution verified-minimum series uses 8, 6, 4, 5, 3, 1, 3, 2, 7, and 14. A 2016 upper-bound specification replaces the six itemized strict-core documents with the reported package total of 16. A separate 2023 boundary test adds signed ministerial meeting minutes, while a 2024 cap limits the strict-core instrument value to the prior observed maximum.

The broad public-institution specification is a sensitivity test rather than an alternative claim of exhaustive coverage. Its purpose is to show whether the trajectory depends on excluding national public research bodies, statutory boards, industry bodies, state media, and state-backed project entities. The 2016 package specification tests unresolved undercount uncertainty without reclassifying 10 unidentified package items as verified baseline documents.

The SPMCI and its dimensions are compared with South China Sea friction intensity, BRI project health, and diplomatic-language intensity using Spearman rank correlations. Maritime friction is coded from documented incidents reported by the Asia Maritime Transparency Initiative (2020, 2021) and Reuters (2020). Ngeow and Jamil (2022) and Ngeow (2024) document project reassessment and resumption. Coenen et al. (2021) provide a broader governance frame for BRI projects. Diplomatic language is coded from official statements (Ministry of Foreign Affairs, Malaysia, 2024). These coefficients are used as descriptive relationship-environment checks only. They are not treated as external validation because none of the three comparison series is fully independent of the empirical environment from which the index is constructed. No p-values are used because the analysis contains 10 annual observations and several ordinal series.

A stronger external test would require evidence generated independently of the index and not assembled from the same diplomatic record. Future applications could pair annual SPMCI scores with expert assessments collected before the scores are disclosed, independent foreign-policy event or interaction datasets, and alternative partnership measures applied to the same dyads. Agreement should be assessed at the level of broad phases and relative ordering rather than exact annual values, because these benchmarks may capture preferences, perceptions, or conflict events rather than maintenance activity itself.

3.5 Measurement boundaries and interpretation limits

The SPMCI is a within-case relative composite for the 2015–2024 observation period. It summarizes observable maintenance activity within the defined indicator universe. It is not an absolute measure of bilateral trust, friendship, strategic alignment, latent capacity, or resilience. A higher score indicates more activity in the retained maintenance channels, not a general claim that the relationship is free from friction.

The design is not intended for causal inference. The series contains 10 annual observations and combines continuous, counted, binary, ordinal, estimated, and proxied inputs. The article therefore does not estimate population-level relationships, use significance tests as validation, or treat the index as a forecasting model. Its evidentiary standard is transparency, construct coherence, auditability, and robustness across plausible coding and weighting choices.

This boundary is central to interpretation. The index is useful when it exposes the assumptions required to combine heterogeneous evidence; it becomes misleading when its precision is overstated. The results are therefore reported as phase patterns, dimension-level movements, and sensitivity checks rather than as exact estimates of an underlying relationship condition.

4 Results

4.1 Baseline SPMCI trajectory

The strict-core baseline records 66.4 in 2015 and 65.6 in 2016. It falls to 48.5 in 2017, rises modestly to 51.7 in 2018, and declines again to 41.1 in 2019. The index reaches its minimum of 19.4 in 2020 and remains low in 2021 and 2022 at 23.8 and 23.0, before rising to 60.9 in 2023 and 71.5 in 2024. The main pattern is therefore a 2020–2022 trough followed by a strong 2023–2024 recovery.

Figure 2 is more useful as a sequence of phases than as a ranking of 10 isolated years. The high 2015–2016 starting level gives way to a lower 2017–2019 period, followed by the 2020–2022 trough and a sharp recovery after 2022. The recovery does not reproduce the earlier configuration. Security activity is important to the 2015–2016 high and the 2023 rebound, whereas the further increase in 2024 is carried mainly by political, economic, and societal gains despite a weaker security score.

Excluding formal relationship status from the political dimension does not overturn this trajectory. Compared with the status-inclusive specification, the baseline has a Spearman rank correlation of 0.952; both specifications identify 2020 as the minimum and 2024 as the maximum. The 2024 political score reflects three direct executive sessions and 13 strict-core instruments, rather than the partnership label itself.

The composition of the recovery is therefore uneven. Political coordination reaches its within-period maximum in 2024 and societal flows have largely returned, while the security score remains well below its 2015–2016 level. The composite records a broad restoration of working channels, not a return to the same institutional balance seen at the start of the series.

4.2 Dimension-level patterns

Figure 3 shows that the four dimensions do not trace a common bilateral cycle. Economic activity is the least synchronized with the political downturn. The investment proxy weakens early, but expanding trade later sustains a material floor even as political and security channels contract. Malaysia-reported bilateral merchandise trade rises from RM230.89 billion in 2015 to RM484.12 billion in 2024 (Malaysia External Trade Development Corporation, 2025). The divergence is substantively important: commercial density persists while several institutional maintenance channels weaken.

Political coordination is high in 2015–2016, declines broadly from 2017, reaches its minimum in 2022, and then reverses sharply in 2023–2024. The sequence spans Malaysia’s 2018 political transition, the reassessment of major China-linked projects, pandemic disruption, and renewed executive engagement under Anwar Ibrahim. In contrast with the economic series, trade continuity did not prevent a marked thinning of direct executive access and formal bilateral instruments. The 2024 maximum reflects three direct executive sessions and 13 strict-core instruments; the anniversary-cap test shows that the recovery is not solely an artifact of document concentration in that year.

Security activity follows a different rhythm. Aman Youyi activation and defense institutionalization keep the dimension high in 2015–2016, produce a partial return in 2018, disappear from the bounded series in 2020–2022, and revive in 2023 before falling again in 2024. The 2018 increase occurs while political coordination remains weak, whereas the 2023 revival coincides with a broader political recovery. By 2024, however, the political maximum is no longer matched by security. This unevenness is consistent with compartmentalized cooperation: professional contact can be activated episodically without resolving maritime caution or moving in lockstep with civilian diplomacy.

Societal connectivity moves on a third timetable. It rises through 2019 despite weakening political coordination, collapses when border controls interrupt travel and study, and rebounds after reopening. The pre-pandemic divergence suggests that social exchange can retain momentum after official channels have begun to thin; the pandemic convergence with the political and security low reflects different mechanisms rather than a single deterioration in bilateral relations. The dimension is therefore interpreted as mobility-based connectivity and tested separately through the pandemic-sensitive exclusion.

Across the series, divergence is pronounced in 2018–2019 and again in 2024. During 2020–2022, political, security, and mobility-based societal channels are all weak, while the economic dimension remains comparatively stronger. The convergence is therefore only partial and does not imply a single cause. In 2023, political and security activity rebound as societal connectivity begins to recover. These cross-dimension differences show when the relationship is sustained by several channels and when it rests more heavily on only part of its institutional structure.

4.3 Weight and specification sensitivity

The Dirichlet weight analysis leaves a wide separation between the low and recovery phases. The 95% weight-sensitivity interval is 16.7–22.2 in 2020 and 65.7–76.7 in 2024. These are simulation-based sensitivity bounds, not statistical confidence intervals. They show that the contrast between the two periods is not produced by the exact 0.30/0.30/0.20/0.20 weight vector.

The deterministic checks probe different sources of uncertainty. Reintroducing formal relationship status yields ρ = 0.952 with the baseline and leaves the minimum and maximum years unchanged. Changes to the bilateral-instrument universe, the unresolved 2016 package, and the 2024 anniversary cap alter annual levels and, in some cases, the identity of the highest year, but the broader trough-and-recovery sequence remains.

Other checks are more demanding. Excluding Chinese tourist arrivals and Malaysian students raises the 2020 score from 19.4 to 22.0; 2020 remains the minimum and 2024 the maximum (ρ = 0.988). Excluding defense institutionalization, the only rubric-coded constituent retained in the baseline, gives ρ = 0.964 and preserves the same phase pattern, although the highest year shifts to 2015. Equal dimension weights produce ρ = 0.939; 2020 remains the minimum, but the maximum shifts to 2016. The precise high-year ranking is therefore sensitive to some defensible specifications (Table 3).

Table 3. Selected robustness specifications for the 8-indicator SPMCI.

Leave-one-indicator-out tests give the same general result. Omitting each of the eight baseline indicators in turn produces correlations with the baseline ranging from 0.879 to 0.988. Every omission preserves a lower 2020–2022 phase and a higher 2023–2024 recovery phase, although the lowest or highest individual year changes in several cases (Table 4). Figure 4 therefore supports a bounded robustness claim: the phase structure is more stable than exact annual ordering or amplitude.

Table 4. Leave-one-indicator-out robustness tests.

4.4 Relationship-environment comparisons

The SPMCI has a strong negative rank association with South China Sea friction intensity (ρ = −0.835), a positive association with BRI project health (ρ = 0.610), and a positive association with diplomatic-language intensity (ρ = 0.826). These coefficients are descriptive co-movement checks. They show that the index tends to be lower when maritime friction is higher and higher when project and diplomatic conditions are more favorable, but they do not constitute independent validation.

The political dimension has the strongest and most consistent relationship-environment pattern: −0.746 with maritime friction, 0.709 with project health, and 0.825 with diplomatic language. The security dimension correlates −0.699, 0.509, and 0.561 with the same three series. The economic dimension is much less closely associated (−0.079, 0.250, and 0.434), consistent with the persistence of commercial exchange through periods of political and security strain.

Independence remains limited across these comparisons. Diplomatic language is generated within the same political relationship, BRI project health overlaps with the project environment and some document activity, and the maritime-friction series is itself a rubric-coded reconstruction of reported incidents. The coefficients are therefore used only to assess whether the trajectory moves in substantively plausible directions; they are not external criterion tests or causal estimates (Table 5).

Table 5. Descriptive Spearman rank associations with relationship-environment indicators, 2015–2024.

N = 10 annual observations. Coefficients are descriptive; no population-level significance, independent validation, or causal claim is made.

5 Discussion

The China–Malaysia case supports a theoretical distinction within strategic-partnership scholarship: formal persistence and economic density do not guarantee that a partnership’s working channels remain equally usable. Economic interdependence can sustain a material floor, while political coordination, security communication, and societal interaction weaken or recover on different timelines. The SPMCI makes this within-partnership variation visible.

5.1 Economic interdependence as a material floor, not partnership maintenance

The economic dimension confirms that trade and investment matter, but it also qualifies a deterministic reading of interdependence (Keohane and Nye, 1977). Political-economy research likewise shows that trade is conditioned by domestic institutions, state intervention, and power relations (Islam et al., 2019). Bilateral trade expanded substantially during the observation period, yet this material density did not prevent the SPMCI from falling sharply during the late 2010s and pandemic period.

The international-governance implication is that economic exchange can support material continuity without preserving the wider set of channels required for partnership maintenance. A relationship may remain commercially dense while losing the political routines, security channels, and domestic legitimacy needed to keep cooperation workable under stress.

5.2 Political coordination as governance infrastructure

The political dimension is built from direct executive sessions and strict-core bilateral instruments. These are not treated as diplomatic ceremony alone. They record two forms of observable coordination: access between national leaders and formalized cooperation among national governmental and regulatory institutions. Collaborative-governance research similarly emphasizes repeated interaction, institutional design, and shared procedures rather than episodic goodwill (Ansell and Gash, 2008; Emerson et al., 2012).

The contrast between the 2018–2022 downturn and the 2023–2024 recovery illustrates this governance function. Political turnover in Malaysia, the reassessment of China-linked infrastructure projects, and pandemic-era disruption weakened the routines through which the relationship had previously been sustained (Ngeow and Jamil, 2022; Ngeow, 2024). Under Anwar Ibrahim, renewed high-level engagement and new agreements coincided with a marked recovery in the SPMCI. This pattern does not show that political visits caused the recovery. It shows that the observable recovery of relationship-maintenance activity moved closely with the restoration of political coordination. Political coordination thus functions as governance infrastructure through which cross-border cooperation can remain usable after domestic or geopolitical shocks.

5.3 Political and security channels as maintenance infrastructure

Direct executive access and strict-core bilateral instruments appear to be the most visible political maintenance infrastructure in the empirical application. The audited session rule treats a separately documented meeting with the Chinese president and a separately documented meeting with the Chinese premier as distinct channels even when they occur during one Malaysian prime ministerial visit. This is more defensible than counting an entire trip once while simultaneously counting remote contact or summit-side meetings under a different unit.

Strict-core instruments stabilize activity across bounded national governmental and regulatory institutions. The distinction from the broad public-institution series matters substantively. A treaty between governments, a central-bank swap agreement, a regulator-to-regulator memorandum, a state-media agreement, and a state-backed commercial-project framework may all be politically relevant, but they do not carry the same institutional form. The baseline therefore avoids converting heterogeneous public-sector visibility into an undifferentiated government-agreement count.

The post-2022 recovery coincides with renewed high-level engagement under Anwar Ibrahim. Political activity recovered strongly, while the security dimension remained uneven. Maintenance activity recovered across several channels, but maritime caution and intermittent defense activation persisted.

The Aman Youyi series identifies a recurring operational-contact mechanism whose institutional form changed over time. The bilateral 2015 and 2016 editions were followed by a trilateral edition in 2018 and a six-country edition in 2023. Recording series activation does not imply alliance or resolve sovereignty disputes. Its value lies in showing whether an identifiable mechanism for professional contact was used. The security–development trade-off is also visible within Malaysia: research on ESSZONE in Tawau shows that measures designed to address non-traditional maritime threats can constrain local maritime economic activity (Anuar et al., 2018).

5.4 Domestic legitimation, pandemic shock, and anniversary effects

Domestic legitimation is a condition of sustainable foreign-policy engagement. China-related cooperation must be framed as economically useful, institutionally manageable, and compatible with sovereignty protection. In Malaysia, the 2018 electoral transition and the reassessment of China-linked projects did not end cooperation, but they unsettled the narratives and administrative routines through which cooperation had been justified (Ngeow and Jamil, 2022).

The baseline retains a pronounced 2020–2022 trough, but its depth should not be read as a pure measure of bilateral deterioration. Border restrictions directly suppressed tourist and student flows, while some diplomatic interaction moved to remote formats. Telephone and video meetings remain countable under the direct-session rule, so the political series does not automatically treat remote executive contact as absent. Within-period min–max scaling also accentuates exceptional lows. The dedicated exclusion test removes both mobility series rather than constructing counterfactual pandemic-year values. Under this specification, the 2020 score rises from 19.4 to 22.0 and remains the minimum, while 2024 remains the maximum; the alternative trajectory is closely associated with the baseline (ρ = 0.988). The trough is therefore best read as a compound low period. Pandemic restrictions deepened it, but they do not by themselves account for the wider phase pattern. A shock-adjusted normalization is not used because it would require unverifiable assumptions about mobility in the absence of border controls.

The strict-core audit also qualifies the interpretation of the 2024 anniversary effect. The baseline contains 13 qualifying national governmental and regulatory instruments in 2024, not an undifferentiated total of 14 government agreements; the additional verified item belongs to the broader public-institution universe. Capping the 2024 strict-core count at the prior-period maximum lowers the SPMCI from 71.5 to 63.4 and makes 2015 marginally the highest year. The post-2022 recovery nevertheless remains. The broad-public specification likewise preserves the trough-and-recovery pattern. This sensitivity is one reason to emphasize phases rather than the exact ordering of high years.

The 2016 package creates the opposite uncertainty. Six strict-core documents are individually recoverable, whereas public reporting gives a package total of 16. The baseline therefore retains six and treats 16 as an upper-bound sensitivity. Under that specification, 2016 becomes the highest year, but the 2020–2022 low period and post-2022 recovery remain. The result reinforces the same interpretive boundary: document-count uncertainty can change exact annual rankings without erasing the broader phase structure.

5.5 Comparative implications for China–ASEAN partnership research

The comparative claim is deliberately limited. Neither the Malaysia-specific indicator list nor its baseline weights should be treated as a model for other cases. The potentially transferable element is the distinction between a partnership’s formal status and the continued usability of the channels that sustain it, together with a procedure for defining bounded annual evidence in economic, political, security, and societal domains. This distinction may be useful for asymmetric relationships that combine substantial cooperation with recurring political or security strain. ASEAN states differ in the mix of engagement, caution, and external options through which they manage China’s power (Kuik, 2008). Those differences call for adapted indicators rather than a mechanically uniform index.

An application elsewhere in Southeast Asia should therefore combine a common conceptual core with a dyad-specific module. Trade and high-level political access may be observable across most relationships, but the relevant security channel could be a coast-guard hotline, a border committee, a defense dialogue, or a recurring exercise. Tourism and student mobility may capture societal connectivity in one case, while labor mobility, cross-border communities, or professional exchange may be more appropriate in another. Relationships involving South China Sea claimants would also require a more detailed account of friction-management institutions than dyads without an active territorial dispute. Using identical indicators everywhere would produce formal comparability at the expense of construct validity.

Absolute cross-dyad score comparison would require a different design. The present min–max normalization is internal to the China–Malaysia series, and the weights are justified for this case; a score of 70 in another partnership would not be commensurable with 70 here. A comparative study would need pre-specified common definitions, pooled or externally anchored normalization, comparable source-coverage audits, and a weighting rule fixed before the results are inspected. Without those steps, the defensible comparison is between phase patterns and dimension profiles, not numerical levels.

5.6 Policy implications and comparative propositions

The policy implications should be read first at the level of the China–Malaysia case. Here, the results show that economic density did not move in step with the political, security, and societal channels captured by the index. One practical implication is that regular political communication can be treated as part of the infrastructure of partnership maintenance rather than as diplomatic symbolism alone. Summit and ministerial channels preserve access and provide routines through which disagreements can be managed after domestic or external shocks. The analysis does not establish that such contact causes a stronger partnership; it shows that political coordination is one of the channels whose weakening and recovery are visible in the Malaysian sequence.

Where maritime friction is recurrent, maintaining practical conflict-management channels may also help prevent operational disputes from spilling across the wider relationship. Defense dialogues, coast-guard communication, incident-notification routines, and professional military exchanges are possible mechanisms. The present study does not test their causal effects, but the uneven security trajectory cautions against assuming that dense economic ties will supply these channels automatically.

For Malaysia, the BRI episode also points to the importance of project governance. Transparent review, fiscal clarity, local consultation, and public communication can make large infrastructure projects easier to defend domestically when their terms become contested. This implication follows from the political setting of the case rather than from a cross-national test, and should be treated accordingly.

Societal connectivity deserves similar caution. Tourism and education recovered after border reopening, but the pandemic shows how quickly mobility-based indicators can collapse for reasons that are not specific to bilateral politics. These channels may still matter for the social reach of a partnership, yet their contribution should be assessed alongside the external shocks that shape them.

For other ASEAN or middle-power settings, these points are better framed as propositions for comparative research than as findings established here. Future work can test whether political access, conflict-management routines, project governance, and societal connectivity play similar roles across dyads with different security exposure, domestic institutions, and patterns of dependence. The present study supplies a case-based framework for asking those questions; it does not establish a general model of middle-power behavior.

5.7 Scope conditions and limitations

First, the analysis concerns one dyad and 10 annual observations. It cannot establish general regularities across strategic partnerships, and its within-period scaling does not permit direct numerical comparison with another dyad.

Second, the SPMCI is formative and theory-led. Its content validity depends on indicator selection, bounded coding rules, and transparent interpretation rather than internal-consistency statistics.

Third, the evidence types remain heterogeneous. Trade and tourism are observed series, investment remains a proxy, 2020–2021 student values remain estimated, and defense institutionalization is rubric-coded. Formal relationship status is retained in the data package for audit and comparison but does not enter the baseline. Coding was conducted by the lead researcher, and no independent second coding was carried out for the full annual series. The operational definition, annual values, and source classes for defense institutionalization are disclosed; excluding that indicator leaves the broad phase pattern intact (ρ = 0.964). That sensitivity check does not substitute for an inter-coder reliability statistic, and single-coder judgment remains a limitation. The two mobility indicators also exhaust the societal dimension. Excluding them is therefore a transparent stress test of pandemic exposure, but it cannot show how a broader set of social ties would have behaved under the same shock.

Fourth, the event audit improves reproducibility but does not abolish archival uncertainty. The direct-session and Aman Youyi universes are explicit and itemized. The strict-core instrument series is also itemized, but 10 items in the reported 2016 package remain unresolved, and the broader public-institution universe is not claimed to be exhaustive. Event counts should therefore be interpreted as verified values within their stated coding universes, not as proof that no other politically relevant contact or document existed.

Fifth, the relationship-environment measures are not fully independent. Diplomatic language is proximate to political activity, BRI health overlaps with the project environment, and maritime friction is rubric-coded. The removal of the stitched public-opinion series avoids manufacturing an annual benchmark from non-comparable surveys, but it leaves the article without a continuous independent criterion series. These correlations are therefore descriptive co-movement checks. They can show whether the SPMCI moves in substantively plausible directions, but they do not establish external validity or criterion validity.

External validation should be designed prospectively. A stronger study would combine independently collected annual expert assessments, an established foreign-policy event or interaction dataset, and at least one alternative partnership measure across several dyads. Where possible, index coders should be unaware of benchmark values, and benchmark assessors should not see the SPMCI results. This would permit separate tests of convergent validity, discriminant validity, and transferability rather than relying on one politically proximate comparison.

Sixth, the dimensional weights remain normative design choices. The equal-weight specification is closely related to the baseline (ρ = 0.939) and retains the 2020 minimum and post-2022 recovery, but it shifts the highest year from 2024 to 2016. This is direct evidence that exact annual rankings are more weight-sensitive than the broad phase pattern. Cross-dyad applications would therefore need an ex ante weighting rationale and sensitivity analysis rather than automatic reuse of the China–Malaysia vector.

Seventh, transparency and sensitivity analysis do not make a formative index theory-free. In the leave-one-indicator-out tests, all eight omissions preserve the broad phase pattern, but correlations with the baseline range from 0.879 to 0.988 and some specifications alter the identity of the lowest or highest year. These results support claims about broad phases more strongly than fine annual rankings. They cannot prove that the construct has a unique numerical form, and small differences between adjacent annual scores should not be treated as substantively decisive.

6 Conclusion

This article treats strategic partnership maintenance as variation in the observable channels that keep a formally designated relationship workable. The SPMCI summarizes eight indicators across economic, political, security, and societal dimensions; formal relationship status is retained only as contextual information and does not enter the composite. Applied to China–Malaysia relations from 2015 to 2024, the index shows a broad 2020–2022 low period followed by a marked recovery in 2023–2024. Economic exchange remains comparatively persistent, whereas political coordination and security communication vary more sharply with the broader relationship environment. The scores describe maintenance activity within the stated indicator universe; they are not estimates of latent trust, alignment, or resilience.

The evidentiary design is based on three bounded and itemized event variables. Direct executive sessions are counted under one consistent session rule; bilateral instruments are separated into a bounded strict-core baseline and a broad public-institution sensitivity series; and the military-event variable is restricted to activation of the Aman Youyi/Peace and Friendship series. These definitions make the indicator universes contestable and replicable instead of presenting web-visible public activity as an exhaustive diplomatic record.

The trajectory is robust in the limited sense tested here. A status-inclusive comparison produces a closely related series (ρ = 0.952) while leaving the broad phase structure intact. Excluding pandemic-sensitive mobility indicators (ρ = 0.988), excluding defense institutionalization (ρ = 0.964), and each of the eight leave-one-indicator-out tests also retain the 2020–2022 low phase and the 2023–2024 recovery. Equal dimension weights remain closely related to the baseline (ρ = 0.939) but move the highest year to 2016, and several other specifications alter the precise low or high year. What survives most consistently is the phase pattern and dimension-level variation, not the exact ranking of individual years.

The results also clarify the limits of what the index can show. The SPMCI is a formative, within-dyad composite whose value depends on transparent coding rules, visible assumptions, and sensitivity analysis. The present series covers only 10 annual observations; one baseline constituent remains rubric coded, the investment measure is a proxy, the 2020–2021 student values are estimated, and no independent second coding was conducted for the full series. Future applications should extend the observation period, use independent coding where feasible, and test the framework across additional dyads with common definitions and externally defined normalization. Independent expert assessments or established foreign-policy event datasets would provide a stronger basis for validation than the politically proximate comparison series used here. For now, the SPMCI is best used as an auditable summary of partnership-maintenance activity rather than a definitive score of strategic relationship quality.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. OpenAI ChatGPT (web version accessed 28 August 2026; GPT-5.6 Sol model; OpenAI, https://chatgpt.com/) was used to assist with language polishing, consistency checks, and formatting review. The authors independently selected the sources, coded the corpus, checked the evidence, performed the calculations, interpreted the results, and approved the final manuscript. The authors take full responsibility for the work.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpos.2026.1930858/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Azhari Setiawan, Raja Ali Haji Maritime University, Indonesia

Reviewed by: Paijie Wan, Shanghai University of Finance and Economics, China

Rizki Damayanti, Paramadina University, Indonesia