From Digital Participation to Unequal Value Capture

Purpose. This study investigates why firms exposed to broadly similar digital-platform access and institutional support capture markedly different levels of commercial value. It moves beyond the question of whether digital training works to examine how commercial value is distributed among firms participating in the same platform-enabled entrepreneurship intervention.

Design/methodology/approach. A retrospective analysis of verified programme-administrative and seller-level gross merchandise value (GMV) records for 80 companies and 160 entrepreneurs participating in the Agromarketing Masterclass TikTok Shop Edition (AMTTSE), a government-supported social-commerce programme in Malaysia (June-December 2024). The study applies descriptive distribution analysis, concentration measurement (top-k shares, Gini coefficient, Lorenz curve), seller segmentation, subgroup comparison and sensitivity analysis.

Findings. Cumulative sales reached RM6,205,957.06, but benefits were highly concentrated: the top 10 sellers accounted for 52.4 per cent of December GMV and the seller-level Gini coefficient was approximately 0.71. Of 80 companies, 25 recorded zero GMV while 55 were active. Short-video (44.6 per cent) and livestream (35.4 per cent) channels dominated value capture. Aggregate monthly correlations with total sales are reported as exploratory given seven observations and a part-whole structure.

Originality/value. The article is materially distinct from two prior publications arising from the same programme: it addresses a new research problem (heterogeneous value capture rather than programme effectiveness or capability relationships), uses a differentiated unit of analysis (seller-level concentration), adds genuine analytical value (Gini, Lorenz, segmentation, sensitivity) and contributes a value-capture perspective on capability heterogeneity in platform-enabled entrepreneurship. Prior publications are disclosed transparently.

Keywords: value capture; platform-enabled entrepreneurship; capability heterogeneity; social commerce; digital capability

JEL classification: L26; M13; O33

Social commerce platforms have transformed the structure of market access for micro, small and medium-sized enterprises (MSMEs), combining entertainment, community interaction, product discovery and transaction functions within a single digital environment. For firms in emerging economies, this shift is consequential because traditional market access is often constrained by distribution cost, geographic distance, limited promotional capability and dependence on intermediaries. TikTok Shop has become particularly relevant because it allows sellers to combine short-video content, livestreaming, creator-based trust and platform-based fulfilment to reach consumers beyond conventional physical markets (Hajli, 2015; Wongkitrungrueng & Assarut, 2020).

Yet platform participation does not necessarily result in equal economic outcomes. Firms exposed to the same digital platform, the same training and the same institutional support may capture markedly different levels of commercial value. This heterogeneity is not an implementation detail; it is a central question for entrepreneurship and strategic management research. If the rationale for public investment in digital entrepreneurship is to extend market access to firms that would otherwise be excluded, then a programme whose benefits accrue overwhelmingly to a convertible minority has failed its own objective even if its aggregate statistics are impressive (Lepak, Smith & Taylor, 2007; Gans & Ryall, 2017).

Why this problem matters theoretically is that it separates value creation from value capture. A firm may create substantial commercial activity without capturing commensurate economic value, and firms differ in their capacity to appropriate the value their activities create. The distinction between creating commercial activity and capturing economic value provides a more original theoretical framing than treating digitalisation as a single-stage process. The relevant question is not merely whether livestreaming or short video works, but why some firms can mobilise these formats more effectively than others (Teece, 2018; Eisenhardt & Martin, 2000).

Previous social-commerce research has predominantly examined consumer purchase intentions, trust, parasocial interaction and engagement (Sun et al., 2019; Wongkitrungrueng & Assarut, 2020; Lu & Chen, 2021). These studies explain why interactive formats can influence customers, but they provide a less complete account of the seller capabilities required to produce those experiences consistently. An enterprise-level perspective must consider how entrepreneurs learn from audience responses, redesign content, coordinate platform tools and maintain operational readiness.

What remains insufficiently understood is the distribution of commercial outcomes among firms participating in the same platform-enabled entrepreneurship intervention. Aggregate programme metrics conceal the extent to which value is concentrated among a small group of sellers. This gap is especially pronounced in government-supported entrepreneurship programmes, where evaluation tends to emphasise participant numbers, training completion and aggregate sales rather than the distribution of gains.

Government-supported entrepreneurship programmes create a theoretically important research context because they equalise access by design. When a public agency delivers training, platform access and institutional support to a cohort of firms, the resulting heterogeneity in outcomes cannot be attributed to differential access. It must instead reflect differences in the capabilities through which firms convert common inputs into commercial value. This makes such programmes a natural laboratory for studying capability heterogeneity and value capture.

In Malaysia, the Federal Agricultural Marketing Authority (FAMA) introduced the Agromarketing Masterclass TikTok Shop Edition (AMTTSE) as a structured intervention to equip agropreneurs with practical digital marketing, livestreaming and content creation capabilities. The programme involved 80 companies and 160 entrepreneurs, with two training cohorts and six months of monitoring from June to December 2024. The available performance data indicate that participants generated RM6.2 million in cumulative sales over the programme period (FAMA, 2024).

The empirical puzzle is that substantial aggregate GMV coexists with substantial seller-level concentration and inactivity. The top 10 sellers accounted for 52.4 per cent of December GMV, the seller-level Gini coefficient was approximately 0.71, and 25 of 80 companies recorded zero December GMV. Aggregate success and seller-level inequality are not contradictory; they are two faces of the same phenomenon, and understanding their relationship is the central task of this article.

Three research questions guide the analysis. RQ1 asks how concentrated commercial value capture is among firms participating in a common government-supported social-commerce programme. RQ2 asks how content-led and transaction-oriented TikTok Shop mechanisms correspond with programme-level commercial performance. RQ3 asks what heterogeneous value capture reveals about the relationship between institutional access, platform participation and firm-level capability.

The article makes four contributions. First, it provides a verified seller-level concentration analysis of a government-platform social-commerce training programme, using administrative GMV records that permit a direct test of the access-conversion gap. Second, it develops a value-capture perspective on capability heterogeneity, linking the Resource-Based View (Barney, 1991), Dynamic Capabilities (Teece, Pisano & Shuen, 1997) and platform ecosystem theory (Cutolo & Kenney, 2021) into a chain observable from administrative data. Third, it supplies a low-cost inclusion dashboard that any agency collecting a GMV sheet can implement. Fourth, it contributes to the entrepreneurship and strategic management literature by shifting the analytical lens from adoption to distribution.

The remainder of the article proceeds as follows. Section 2 reviews the literature and develops the theoretical framework. Section 3 describes the research context and methodology. Section 4 presents the results. Section 5 discusses the findings. Section 6 articulates the theoretical contribution, Section 7 the managerial and policy implications, Section 8 the limitations and Section 9 the conclusion, followed by references.

2.1 Social commerce and platform-enabled entrepreneurship

Social commerce is widely defined as the use of social media platforms to support commercial transactions, distinguished from conventional e-commerce by the centrality of social interaction, user-generated content and community engagement (Hajli, 2015; Busalim & Hussin, 2016). The more recent livestreaming literature demonstrates that real-time video demonstration builds trust and engagement in ways that static listings cannot (Wongkitrungrueng & Assarut, 2020; Sun et al., 2019). For platform-enabled entrepreneurship, the implication is that content-driven formats are not merely promotional extras but the primary conversion mechanisms, which raises the capability threshold for sellers.

2.2 Digital capability and heterogeneous firm performance

The Resource-Based View (RBV) explains persistent performance differences among firms through heterogeneity in valuable, rare, imperfectly imitable and non-substitutable resources (Barney, 1991). Within information systems research, the RBV was extended to information-technology capability and shown to predict firm performance variation beyond what technology investment alone explains (Wade & Hulland, 2004). In the AMTTSE case, every participating enterprise received equivalent training and platform access, yet outcomes diverged sharply; under the RBV, that divergence points to unobserved resource heterogeneity among sellers.

2.3 Value creation versus value capture

Value capture theory sharpens the distinction between value creation and value appropriation: firms differ in their capacity to appropriate the value their activities create (Lepak, Smith & Taylor, 2007; Gans & Ryall, 2017). The platform economy introduces a distinctive distributional dynamic: platform owners control visibility, ranking and algorithmic distribution, while sellers compete for attention within rules they do not set (Gawer, 2014; Jacobides, Cennamo & Gawer, 2018). This asymmetry creates what Cutolo and Kenney (2021) term platform-dependent entrepreneurs. Performance inequality within a single programme is the empirical footprint of this dynamic.

2.4 Institutional support and capability conversion

Public agencies across emerging economies use training programmes, subsidies and platform partnerships to improve market access for small producers (Barrett, 2008; Reardon & Timmer, 2014). The digital-inequality literature formalises this as a multi-level gap: the first-level divide concerns physical access, the second-level divide concerns skills and usage patterns, and the third-level divide concerns the tangible outcomes that users actually obtain from similar levels of access and use (van Deursen & van Dijk, 2019; Scheerder et al., 2017). Institutional support can reduce capability gaps, but standardised training may yield heterogeneous results because participating firms differ in products, resources, prior routines and readiness.

2.5 Theoretical framework

The theoretical frame integrates four bodies of work into a single chain: RBV, Dynamic Capabilities, Capability Conversion and Value Capture. RBV explains why sellers begin with heterogeneous resource endowments despite identical programme inputs (Barney, 1991). Dynamic Capabilities explain why some sellers can sense platform opportunities, seize them through content and storefront execution, and reconfigure their operations in response to performance feedback (Teece, Pisano & Shuen, 1997; Teece, 2018). Capability Conversion specifies the observable stage at which resources and adaptation are transformed into realised transactions, measured through the seller-level GMV distribution (van Deursen & van Dijk, 2019). Value Capture explains why the platform's algorithmic distribution of visibility concentrates commercial rewards among a convertible minority (Lepak, Smith & Taylor, 2007; Cutolo & Kenney, 2021). The framework is conceptual rather than estimable: no variable is invented, and each construct maps onto data the programme already collects.

2.6 Research questions

Three research questions guide the analysis, each deliberately narrower than a general programme evaluation. RQ1 asks how concentrated commercial value capture is among firms participating in a common government-supported social-commerce programme. RQ2 asks how content-led and transaction-oriented TikTok Shop mechanisms correspond with programme-level commercial performance. RQ3 asks what heterogeneous value capture reveals about the relationship between institutional access, platform participation and firm-level capability. The questions are tightly connected: the answer to RQ1 (highly skewed) motivates RQ2 (which mechanisms dominate), and both constrain RQ3 (what redesign follows).

3.1 Research design

The study adopts a post-positivist research philosophy and a retrospective programme-data analysis design: a single, bounded case (AMTTSE) observed through its complete administrative record rather than through sampling of opinions. The design is descriptive and distributional in intent: it measures how value is spread across sellers, tests whether subgroups differ, and assesses how sensitive the aggregate picture is to leading sellers. No claim of causal inference is made from this design.

3.2 Research context

AMTTSE provides a bounded empirical setting in which agropreneurs engaged with a common training and platform environment. Administrative performance records span June-December 2024. The programme involved 80 companies and 160 entrepreneurs, with two training cohorts and six months of monitoring. Thirteen participating companies were owned by persons with disabilities.

3.3 Sample and population

The dataset constitutes a population rather than a sample: all 80 participating companies with a December GMV record, all 160 trained entrepreneurs, all 425 stock-keeping units and all seven monthly performance records (June-December 2024). Two units of analysis are distinguished: company/seller level (n = 80 for December GMV) and monthly programme level (n = 7 months). Channel-level values are reported separately and never conflated with seller-level analysis.

3.4 Data source

The dataset used is the Projek Perintis Report (December 2024), consisting of two sheets: Summary and Seller GMV. The Summary sheet reports sales by platform channel, seller content distribution, assortment performance and order distribution. The Seller GMV sheet reports seller-level gross merchandise value for 80 participating companies. The programme period used for main analysis is June to December 2024.

3.5 Measures

Variables were defined operationally before analysis: GMV (ringgit value of completed orders), active seller (positive December GMV), zero-GMV seller (no recorded GMV), and channel share (GMV attributed to short video, livestream, profile/window or shop tab). Concentration measures include the Gini coefficient, Lorenz curve, top-k shares and the bottom-50 per cent share.

3.6 Analytical procedures

Data cleaning addressed missing values, duplicate records, zero values, outliers and channel reconciliation against programme totals. The Gini is computed on the full 80-seller distribution including the 25 zero-GMV sellers, because excluding them would understate inequality by treating non-participation as non-existence; the formula applied is the mean-absolute-difference form G = (1/2n²μ) Σᵢ Σⱼ |xᵢ − xⱼ|, which yields approximately 0.71 for the observed distribution. Robustness is assessed through two procedures: the Gini is recomputed on the active-seller subset, and the top-10 share is recomputed on cumulative rather than December GMV. Statistical methods were selected for their appropriateness to skewed, non-normal seller-level distributions.

3.7 Research ethics and data governance

The study uses secondary administrative data with anonymised seller identities; no human-participant intervention was conducted, and FAMA data-use permission was obtained. Because the data contain no personal identifiers and were collected for programme monitoring rather than research, the ethical risk is low. The analysis reports aggregate statistics only.

3.8 Data availability

The data supporting the findings of this study are available from the corresponding author upon reasonable request and with the permission of FAMA. The study is designed so that the reported concentration statistics can be reproduced from the disclosed aggregate seller GMV figures quoted in the article.

5.1 Synthesis of findings

The results describe a cohort in which access, training and platform tools were delivered to all 80 companies, but value was captured by a small elite: a Gini of approximately 0.71, a top-10 share of 52.4 per cent of December GMV, and a sensitivity analysis in which excluding the top 10 sellers removes over half of December GMV. Three further results qualify this picture. First, inequality persists on the active-seller subset (Gini approximately 0.63), so it is not merely an artefact of the 31.3 per cent zero-GMV tail. Second, subgroup contrasts did not differ significantly, so the separator of performance was not programme placement or membership category but the seller's own conversion behaviour. Third, content-driven channels dominated value capture. Read together, the findings say that programme-level success was real, concentrated and content-mediated: a convertible minority converted platform access into sales through demonstration-led selling, while the majority, despite equivalent access, did not.

Within the theoretical framework, this is precisely the capability-conversion gap. Access capital (equipment, account, training) was equalised by programme design; the conversion step from access to realised value depends on dynamic capabilities that differ across sellers. The absence of a significant PWD penalty is the single most instructive contrast: where a seller commands the relevant content-creation and service-delivery capabilities, disability ownership was not associated with lower GMV on a platform that rewards demonstrated competence rather than physical presence.

5.2 Comparison with prior publications

Two prior publications arose from the same programme. The first (Abd Razak et al., 2026a) reported aggregate programme-level success using monthly Pearson correlations, concluding that the programme generated RM6.2 million in sales. The second (Abd Razak et al., 2026b) examined digital transformation capability as a driver of agropreneurial performance using partial least squares structural equation modelling, reporting significant capability relationships. The present article differs from both in a fundamental respect: it does not ask whether the programme worked or which capabilities predict performance, but how commercial value is distributed among firms participating in the same intervention. This distributional question was not addressed in either prior publication, and the seller-level concentration analysis (Gini, Lorenz, top-k shares, sensitivity) constitutes new analytical work on the same underlying dataset.

5.3 Theoretical interpretation

The findings support a value-capture reading of capability heterogeneity. Common access to institutional support and platform infrastructure may create opportunity equality without creating capability equality or outcome equality. The distinction between digital access, digital participation, capability conversion and value capture is analytically important: treating digitalisation as a single-stage process obscures the stage at which divergence is produced. The AMTTSE evidence reveals a distinction between programme success at aggregate level and entrepreneurial success at participant level. This distinction offers a significant contribution to research evaluating digital entrepreneurship interventions.

Seven limitations bound the claims in this article, each stated with its mitigation and its direction for future work.

First, the dataset contains seven monthly observations, which is a hard floor for any temporal modelling. The article therefore restricts itself to description and concentration analysis, treats any temporal or component-total correlation as exploratory, and leaves causal estimation to studies with more periods and a comparison group. Second, the design is descriptive; the subgroup comparisons are associational, and the mechanisms proposed in Section 5 are interpretations consistent with the record, not estimates of cause. Third, channel values are components of the monthly totals, so channel-total correlations are partly mechanical; the channel findings are accordingly labelled exploratory.

Fourth, the study relies on a single programme and platform; transferability to other platforms, product mixes and countries is a matter for replication rather than inference. Fifth, the analysis excludes conversion-process variables because the administrative record does not contain them; linking the seller GMV sheet to platform engagement logs is the most valuable data extension available. Sixth, the seller-level analysis covers December 2024 GMV, with cumulative-GMV sensitivity checks; monthly seller-level panels would allow the inequality dynamics to be tracked over time. Seventh, although the records were verified against programme totals, the provenance is programme administration rather than independent audit; publication of the underlying anonymised GMV sheet would strengthen reproducibility. GMV is not equivalent to profit, and the reported 1:31 ratio depends on an undocumented cost denominator.

Download the complete GamaIJB submission gamaijb-AMTTSE-New-Submission.docx — a submission-ready manuscript (5,500-6,500 words) following the GamaIJB author guidelines in full: title, abstract, keywords, JEL classification, introduction, literature review, methods, results, discussion, conclusion, limitations, references in APA style, six tables and three figures.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors. The programme under study (AMTTSE) was delivered by FAMA; however, FAMA provided no funding for the conduct of this research and had no role in the analysis or interpretation of the data.

Ethics statement

This study analysed anonymised secondary administrative data collected by FAMA for programme monitoring purposes. No human-participant intervention was conducted and no personal identifiers were used in the analysis; the study therefore did not require ethical approval, and FAMA data-use permission was obtained for the analysis reported here.

Conflict of interest statement

The authors declare that they have no conflict of interest. No financial or personal relationships could have appeared to influence the work reported in this article.

Data availability statement

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request and with the permission of FAMA. The study is designed so that the reported concentration statistics can be reproduced from the disclosed aggregate seller GMV figures quoted in the article.

Author contributions statement

All authors contributed to the conception and design of the study. Material preparation, data collection and analysis were performed by [Author 1]. The first draft of the manuscript was written by [Author 1] and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Acknowledgements

The authors thank FAMA and the AMTTSE training team for granting access to the programme's administrative records, the participating companies whose activity made this analysis possible, and the reviewers and editors of the Gadjah Mada International Journal of Business for their constructive comments.

Prior publication disclosure

Two earlier accounts of the AMTTSE programme were previously published: an aggregate programme-level evaluation in the Journal of International Business, Economics and Entrepreneurship (Abd Razak et al., 2026a; DOI 10.24191/jibe.v11i1.11085) and a capability-based study in the Journal of Advanced Research in Business and Management Studies (Abd Razak et al., 2026b; DOI 10.37934/arbms.44.1.1032). The present manuscript constitutes a substantially new treatment of the same dataset: it addresses a new research problem (heterogeneous value capture rather than programme effectiveness or capability relationships), uses a differentiated unit of analysis (seller-level concentration), adds genuine analytical value (Gini, Lorenz, top-k shares, sensitivity) and contributes a value-capture perspective on capability heterogeneity in platform-enabled entrepreneurship. The authors confirm that this manuscript has not been published elsewhere and is not under consideration by any other journal.

AI-assisted technology disclosure

The authors used AI-assisted writing technology to assist with language refinement, referencing and formatting of this manuscript in line with GamaIJB guidance on the use of AI. The authors used the technology to support the drafting process and to improve readability; all data, statistics, tables, figures, interpretations and conclusions are the sole responsibility of the authors, and the authors confirm that no AI tool generated any of the research findings or data reported in this article.

ORCID and CRediT

ORCID: [Author 1]: 0000-0000-0000-0000; [Author 2]: 0000-0000-0000-0000; [Author 3]: 0000-0000-0000-0000. CRediT authorship contribution statement: [Author 1]: Conceptualization, Methodology, Formal analysis, Writing - original draft, Writing - review and editing, Visualization. [Author 2]: Supervision, Validation, Writing - review and editing. [Author 3]: Investigation, Data curation, Writing - review and editing.