Unequal Gains from Digital Transformation? Evidence on Firm Performance Heterogeneity and Endogeneity in Vietnamese Enterprises

Abstract

This study examines the drivers and heterogeneous impacts of platformization—a form of digital transformation involving systems such as supply chain management, product data management, and integrated information technology solutions—on firm performance in a developing economy. Drawing on the Resource-Based View and Dynamic Capabilities View, we analyze data from 5542 Vietnamese firms across four sectors using an endogenous switching regression model, complemented by quantile regression. Platformization decisions are shaped by firm resources and managerial expectations, with strong sectoral variation. In manufacturing and construction, larger assets and a lower leverage promote adoption, while, in wholesale and retail, workforce size and perceived competitiveness are key drivers. Platformization enhances the returns to assets and cash flow—especially among high-performing firms—while reducing the negative effects of high debt and geographic disadvantages. The findings offer three practical implications: (1) prioritize digital adoption in asset-heavy sectors when financial conditions are stable; (2) invest in coordination- and customer-focused platforms in labor-intensive sectors; and (3) use digital tools to convert liquidity into performance gains. These insights support inclusive digitalization policies and contribute to Sustainable Development Goals 8 and 9 by linking digital transformation to resilience, adaptability, and innovation-led growth in transitional economies.

1. Introduction

Digital transformation has become a critical driver of structural change and economic development in both developed and emerging economies. Defined as the integration of digital technologies into business processes, it reshapes traditional industrial structures, enables new forms of competition, and creates opportunities for innovation, operational efficiency, and improved financial performance [1]. In developing economies, it also offers a pathway to leapfrogging traditional stages of industrialization by reconfiguring production and value creation processes. Technologies such as cloud computing, e-commerce platforms, and supply chain management systems increasingly influence the distribution of economic power across firms and regions, with significant implications for productivity, competitiveness, and inclusive growth [2]. However, the uneven adoption of digital technologies risks reinforcing disparities between digitally advanced and lagging firms, particularly where access to infrastructure and digital literacy is limited. From an organizational economics perspective, digital transformation enables firms to mobilize and reconfigure resources, fostering adaptive capacity and resilience in volatile environments—capabilities emphasized in the Resource-Based View (RBV) [3,4] and the Dynamic Capabilities View (DCV) [5]. In this context, digital transformation is increasingly recognized not only as a tool for firm-level efficiency, but also as a driver of sustainable industrial upgrading, regional equity, and inclusive growth [6].

While digital transformation has gained prominence globally, its effects on firm performance remain complex and context-dependent, often influenced by regional economic structures, sectoral characteristics, and firm-specific factors. Existing studies predominantly focus on high-income economies, where firms typically have better access to digital infrastructure and resources, but less attention has been paid to how digital transformation influences firms operating in low- and middle-income economies, where resource constraints and market conditions pose unique challenges [7,8]. This is particularly evident in the financial sector, where digital transformation through FinTech has shown both promise and limitations for promoting financial inclusion in emerging markets [9].

Emerging economies, such as Vietnam, offer a compelling context for studying the impacts of digital transformation. Vietnam has experienced rapid digitalization, driven by both private sector initiatives and national policies aimed at enhancing competitiveness in the global market. Key government programs—such as the National Digital Transformation Program and project for developing human resources for national digital transformation—aim to promote the adoption of digital technologies across sectors, with specific support for small- and medium-sized enterprises (SMEs) through training, incentives, and infrastructure investment. As of 2023, Vietnam’s digital economy accounted for over 12% of its national GDP, with government targets set at 20% by 2025 and 30% by 2030 [10]. The government’s promotion of platformization has reshaped firm operations by expanding market access and improving efficiency, while also intensifying competition and putting downward pressure on margins. These changes are crucial for reducing regional productivity gaps and advancing more inclusive industrial development. Despite these shifts, research on digital transformation in Vietnam remains limited, with most studies focused on adoption rather than performance outcomes [11]. This gap highlights the need for context-sensitive studies that examine both the drivers and impacts of digital transformation in transitional economies.

In addition to contextual gaps, several methodological challenges continue to hinder progress in digital transformation research. A recurring issue is endogeneity: firms with stronger performance are often more likely to adopt digital transformation, which leads to biased estimates. While instrumental variable (IV) methods are commonly employed to address this problem, the difficulty of identifying valid instruments limits their reliability and broader applicability [12]. Moreover, many existing studies rely on interaction terms or moderation models to explain performance heterogeneity, which often oversimplifies how firms respond to digital transformation across different sectors and conditions. There remains a pressing need for more rigorous empirical approaches that can simultaneously address the self-selection bias and capture the outcome heterogeneity in firm performance [11]. Additionally, much of the literature focuses primarily on the adoption stage, offering limited insights into the performance implications of digital transformation post-adoption [13].

To address these limitations, this study examines the drivers and performance effects of platformization—a form of digital transformation involving integrated systems such as supply chain management and product data management—within the context of a developing and transitional economy. Specifically, we ask the following questions: (1) What factors influence firm-level platformization decisions across sectors in Vietnam? and (2) How does platformization affect firm performance, and how do these effects vary across the performance distribution?

This study focuses on Return on Sales (ROS) as the primary measure of firm performance. ROS captures both operational efficiency and pricing power, is consistently available in firm-level data, and enables a comparison across sectors with differing capital intensity. It provides a meaningful lens to assess the profitability implications of digital transformation in a context marked by structural and financial heterogeneity. The objective of this study is to develop a sector-sensitive, firm-level understanding of how platformization both emerges and affects financial outcomes in a resource-constrained environment. By combining an endogenous switching regression model with quantile analysis, we produce robust, context-aware evidence on the strategic and distributional effects of digital transformation. In doing so, the study extends existing methods by applying this combined approach to the context of digital transformation in a transitional economy, where structural heterogeneity and selection bias are central concerns.

Beyond its empirical contributions, this study is conceptually grounded in the RBV and the DCV, which guide our understanding of how digital transformation interacts with firm-level resources and constraints. RBV and DCV offer complementary insights into how firms derive competitive advantage—through the possession of valuable and scarce resources (RBV) and through the ability to adapt and reconfigure these resources in dynamic environments (DCV). We focus, in particular, on platformization as a mechanism that enhances resource utilization and mitigates disadvantages such as high leverage or locational remoteness. Sector-specific findings highlight that digital transformation’s performance effects are not uniform, reinforcing the importance of contextualizing digital strategies within the industrial structure.

This study makes three contributions. First, it provides rare empirical evidence on the digital transformation from a transitional economy where national digital strategies are reshaping industrial development but firm-level outcomes remain underexplored. Second, it links adoption decisions with performance effects using firm-level data across four sectors, offering a nuanced view of how platformization translates into financial outcomes. Third, it applies an endogenous switching regression model—complemented with quantile regression—to jointly estimate selection and outcome heterogeneity. This approach captures both causal effects and distributional differences, offering methodological value for studying digital transformation in resource-constrained settings.

This study contributes to empirical and theoretical knowledge, and the development of analytical frameworks that can inform corporate strategy and governance. Our methodological design supports more accurate decision-making for both policymakers and corporate leaders seeking to promote inclusive and sustainable digital development. In this light, digital transformation should not be viewed solely as a technological upgrade, but as a strategic and fiduciary responsibility. As argued by [14], integrating innovation and sustainability into governance falls within the evolving interpretation of fiduciary duty, particularly in contexts where digital innovation is central to competitiveness and resilience. Our findings support this perspective, showing that uneven gains from digitalization can be addressed through informed, context-sensitive strategies.

Finally, this study aligns with the broader agenda of sustainable development by analyzing how digital transformation contributes to economic sustainability and resilience in a transitional economy. As digital technologies reshape firm behavior, understanding their impact on profitability and competitiveness is essential for promoting inclusive and long-term economic growth, especially in resource-constrained environments like Vietnam. By examining the heterogeneous effects of digitalization across sectors and firm types, the study informs policies aimed at fostering sustainable industrial development, reducing productivity gaps, and supporting digital inclusion. These insights contribute to Sustainable Development Goal (SDG) 8 (Decent Work and Economic Growth) and SDG 9 (Industry, Innovation, and Infrastructure), and are especially relevant for policymakers balancing technological progress with equitable economic transformation.

The remainder of this paper is structured as follows: Section 2 reviews the relevant literature on digital transformation, focusing on its impact on firm performance, key measurement approaches, and methodological challenges. Section 3 outlines the research methodology, including the data, variables, and the endogenous switching regression model used to estimate the determinants of digital transformation adoption and firm performance. Section 4 presents the empirical results and discusses the findings in the context of existing theories and the economic environment in Vietnam. Finally, Section 5 concludes by summarizing the key contributions, offering managerial and policy implications, and suggesting avenues for future research.

2. Literature Review

Digital transformation has become a central focus in organizational and strategy research, especially as firms increasingly adopt digital technologies to reshape business models, improve efficiency, and respond to volatile environments. In both developed and developing economies, the integration of platforms, including the supply chain management (SCM), product data management (PDM), and information systems, has introduced new pathways for firms to enhance performance. To understand the strategic drivers and consequences of such transformations, scholars have drawn on foundational theories in strategic management. Among these, the RBV and the DCV have emerged as two dominant frameworks. This section reviews these perspectives to frame our conceptual and empirical approach to platformization and its performance implications.

2.1. Digital Transformations: Theoretical Foundation

The decision to adopt digital transformation, including platformization, is commonly analyzed through the lens of strategic management, particularly the RBV and the DCV. These frameworks, widely used in organizational economics and management theory, provide insights into how firms develop and leverage internal assets to gain and sustain competitive advantage. The RBV, originating with Wernerfelt [3] and formalized by Barney [4], emphasizes the strategic role of tangible and intangible resources—such as financial assets, human capital, and technological capabilities—in shaping firm performance. From this perspective, resource-rich firms are better positioned to adopt digital transformation, as they can more easily mobilize the capital and knowledge needed to implement complex digital systems. However, the RBV often assumes relatively stable environments and resource abundance, which limits its applicability to developing economies.

To address the limitations of the RBV in dynamic environments, the DCV—developed by Teece, Pisano, and Shuen [5]—emphasizes a firm’s ability to sense opportunities, seize them, and reconfigure resources to maintain competitiveness [15]. This framework extends the RBV by accounting for the need for continuous adaptation in volatile and resource-constrained settings, making it particularly relevant for transitional economies. As noted by [16], dynamic capabilities include both organizational agility and strategic foresight, which are crucial for firms navigating institutional instability and market turbulence. However, existing applications of the DCV often underemphasize the role of subjective managerial perceptions and contextual uncertainties in shaping digital adoption. By incorporating firm-level perceptions into the analysis, this study builds on and extends the DCV logic to better capture the strategic decision-making processes in digitally transforming firms.

Firms undergoing digital transformation face a dual imperative: to leverage digital tools to enhance performance and competitiveness, and to adapt organizational structures, resources, and capabilities to rapidly evolving technological and market environments. The RBV offers a foundational lens to understand how firm-specific assets, including digital infrastructure and human capital, can generate a sustained competitive advantage when they are valuable, rare, inimitable, and non-substitutable [3,4]. However, RBV’s relative emphasis on static resource possession has been critiqued for not adequately capturing the dynamic nature of digital competition [5]. In response, the DCV has emerged to emphasize the importance of integrating, reconfiguring, and renewing internal and external competences in rapidly changing environments [6]. Digital transformation, in this view, reflects a firm’s capacity to adapt through learning, experimentation, and the reorganization of routines [7]. Prior studies have shown that dynamic capabilities such as sensing, seizing, and transforming can mediate the effect of digital tools on firm outcomes, particularly in uncertain or competitive markets [8,9].

Firms adopt digital transformation not only to automate processes but also to reorganize production, integrate supply chains, enhance market reach, and create new value through data-driven services. As the RBV suggests, firms that possess superior digital capabilities can leverage them for competitive advantage, especially when such capabilities are valuable, rare, inimitable, and non-substitutable [17,18]. The DCV further refines this logic, arguing that firms need the ability to integrate, build, and reconfigure internal and external competences to address rapidly changing environments [19,20]. Thus, digital transformation is not only an outcome of digital investment, but also of organizational adaptation and strategic resource alignment [21].

In this study, the RBV framework informs our interpretation of platformization as a strategic resource whose value depends on the firm’s internal assets and routines. The DCV perspective guides our empirical approach by emphasizing that digital transformation outcomes depend on a firm’s ability to reconfigure resources—an ability that varies across firms and influences both the decision to adopt platformization and the performance gains that follow. These theoretical lenses directly motivate our use of endogenous switching regression and quantile methods to capture the heterogeneity in both adoption and outcomes.

2.2. Digital Transformations and Firm Performance: Empirical Evidence

A growing body of empirical research has examined the link between digital transformation and firm performance across diverse settings, technologies, and sectors. These studies consistently highlight that, while digital transformation is generally associated with improved performance, the effects are far from uniform and depend on both firm-level capabilities and contextual enablers.

Recent meta-analytic evidence confirms a significant positive relationship between digital transformation and performance, with stronger effects observed in manufacturing sectors and developing economies [22]. This reinforces the relevance of digital transformation as a driver of economic upgrading in contexts like Vietnam. At the firm level, [23] uses a text-mining-based digital transformation index to show that digital initiatives in Chinese manufacturing firms improve profitability, particularly via cost reductions and innovation. Similar findings are found in [24]: digital transformation enhances cost efficiency and innovation, which translate into gains in return on assets (ROA) and equity (ROE), especially for firms at mature stages.

Other studies have explored the mechanisms and heterogeneity more explicitly. One line of research, using quantile regression, finds that digital transformation improves innovation output and efficiency, with stronger gains among firms closer to the technological frontier [25]. Another study identifies organizational capital as a key moderator, showing that digital transformation enhances firm value only when supported by strong internal coordination and governance structures [26]. A similar conclusion emerges in the ESG context, where digital transformation alone has a limited impact on firm valuation, but its interaction with ESG practices leads to substantial performance gains [27].

Studies using structural modeling approaches highlight the importance of sequencing and the depth of digital adoption. Evidence from an instance of sequential modelling with small firms in South Africa shows that simple digital tools such as mobile technologies and social media enhance innovation and productivity by improving information access [28]. Similarly, research based on a Difference-in-Differences with Propensity Score Matching (DiD-PSM) framework in Italy finds that productivity gains rise with both the number and sophistication of digital technologies adopted, with artificial intelligence (AI) tools yielding the strongest effects [29].

A consistent insight across these studies is the presence of threshold effects: performance improvements from digital transformation often materialize only when firms possess complementary capabilities or reach a certain scale. This pattern is further supported by [30], who report a J-shaped relationship between digitalization and firm performance in U.S. firms, with internationalization and FDI inflows amplifying the gains. These findings suggest that the benefits of digital transformation depend on firms’ internal resources, absorptive capacity, and alignment with contextual conditions—factors that directly inform this study’s empirical strategy.

2.3. Digital Transformation in Developing and Transitional Economies

Digital transformation is increasingly recognized as a catalyst for economic restructuring in developing economies. By enabling platformization and enhancing access to broader markets, digital transformation fosters structural changes across industries, contributing to improved productivity and competitiveness. In countries like Vietnam, where digitalization is a national priority, platform-based transformation has reshaped traditional business models as part of a broader strategy to modernize the industry and strengthen global integration.

Digital innovation offers the potential for firms in developing economies to leapfrog the traditional stages of industrialization by improving operational efficiency, expanding market access, and upgrading product and service quality [7,11]. However, these benefits are often constrained by systemic barriers such as inadequate infrastructure, limited access to capital, and pronounced regional disparities in digital readiness [7,8]. For instance, rural SMEs frequently lack the digital infrastructure available in urban centers, hindering their competitiveness. Successful transformation in such contexts requires coordinated efforts across policy, infrastructure, and firm-level capabilities.

Importantly, digital transformation is not just a matter of technological adoption but involves deeper shifts in organizational routines, workforce capabilities, and strategic orientation. Effective digital innovation, particularly in SMEs, depends on integrating individual, organizational, and environmental factors [11]. Firms must develop dynamic capabilities to sense digital opportunities, seize them, and reconfigure resources to sustain advantage in volatile markets. When supported by inclusive digital policy and infrastructure investment, this transformation enhances both firm resilience and equitable industrial upgrading.

The institutional and environmental conditions in transitional economies significantly shape digital transformation outcomes. Unlike high-income countries where digitalization is largely market-driven, digital adoption in Vietnam is shaped by structural constraints and guided by government-led initiatives. National programs—such as the country’s digital transformation strategy and its long-term vision for innovation and industrial upgrading—reflect a strong policy commitment to integrating digital technologies into the broader development agenda. These initiatives emphasize state support for digital infrastructure, capacity building, and technology diffusion across sectors [13,14,15].

Policymakers play a vital role in addressing persistent digital divides. Financial incentives, digital infrastructure investment, technology adoption programs, and sectoral knowledge-sharing networks are increasingly used to accelerate adoption across firm sizes and sectors [24]. Policies targeting infrastructure gaps and digital literacy—especially in rural and underserved areas—are essential for ensuring that smaller firms can benefit alongside larger enterprises. In this regard, digital transformation should be seen not only as a tool for efficiency but also as a mechanism for long-term economic sustainability and inclusion.

Vietnam’s experience reflects broader regional trends. Evidence from China shows that the Big Data Pilot Zone (BDP) policy significantly improved firm-level productivity, especially among non-state firms with a strong managerial capability [31]. Another study finds a U-shaped relationship between digital economy development and productivity, with mid-sized and centrally located firms benefiting the most—highlighting the importance of contextual factors such as geography and firm characteristics [32].

Similar findings highlight the enabling conditions for digital transformation. One study shows that digital finance can alleviate credit constraints, thereby facilitating adoption—particularly among smaller, private firms [33]. Additional evidence emphasizes the role of outward internationalization and foreign direct investment in enhancing digital transformation outcomes in U.S. firms, offering relevant insights for transitional economies participating in global value chains [30].

At the firm level, outcomes vary according to internal capacity and sectoral conditions. Evidence from China shows that platform-based technologies—such as supply chain and PDM systems—enhance manufacturing performance by boosting process innovation and reducing costs [23]. In South Africa, even informal micro-enterprises have been shown to benefit from digital communication technologies, which stimulate innovation and productivity in resource-constrained settings [28]. Together, these findings highlight the importance of firm capabilities, institutional quality, and sectoral linkages in shaping the outcomes of digital transformation [16,34,35].

In Vietnam, digital transformation remains uneven across sectors and regions. Digital intensity is generally higher among exporters and state-owned enterprises [24,36], while private domestic SMEs often face greater challenges due to cost concerns, skill shortages, and uncertainty about digital returns. This aligns with the recent empirical research and case studies [37,38]. Additionally, concerns over labor displacement and digital inequality persist, particularly in sectors with large informal workforces.

In sum, digital transformation holds substantial promise for accelerating economic restructuring in transitional economies. Yet, its success depends on the interplay among national policy, infrastructure development, firm-level capabilities, and institutional context. This study contributes to the literature by examining how platformization—a core aspect of digital transformation—shapes firm performance within the context of Vietnam’s ongoing economic transition.

2.4. Methodological Advances in Estimating Digital Transformation Effects

Estimating the effects of digital transformation on firm performance presents two persistent methodological challenges: endogeneity in adoption decisions and heterogeneity in outcomes. Firms that adopt digital transformation are often systematically different from non-adopters—typically possessing greater resources, stronger management, or prior growth trajectories—leading to selection bias and complicating causal inference.

To address endogeneity, researchers have employed IV regression using regional or industry-level digital transformation averages as instruments [34,35]. However, the exogeneity of these instruments is frequently questioned, as they may also directly affect firm performance [24]. Other common strategies, such as lagged variables or proxy-based controls, often introduce additional bias or fail to capture the complexity of digital transformation adoption dynamics.

Quasi-experimental designs have gained traction as alternatives. DiD PSM, as used in studies like [29,31], help control for observable differences between treated and control firms. Yet, these approaches remain limited when the selection arises from unobservable firm characteristics. Structural models such as the sequential modelling approach [28] and system GMM have also been employed to capture feedback loops and innovation dynamics, though these models often require strong identifying assumptions.

Alongside endogeneity, researchers increasingly recognize that digital transformation effects are not uniform across firms. Recent studies emphasize that the performance effects of digital transformation depend on mediating mechanisms such as business model innovation [39], the firm’s distance to the technological frontier [40], and organizational alignment during digital mergers and acquisitions [41]. Contextual disruptions, such as pandemics or natural disasters, have also been shown to shape digitalization outcomes, particularly for non-technological firms [42].

Quantile regression, as applied in [25,43], reveals that high-performing firms tend to derive greater benefits from digital adoption—findings consistent with theories of absorptive capacity and resource-based heterogeneity. However, these approaches typically do not correct for endogeneity, limiting their causal interpretability.

Mediation and moderation analyses further illuminate how and under what conditions digital transformation affects firm outcomes. Mediation models identify mechanisms such as improved asset turnover, lower operating costs, and innovation success [24,34], while moderation analyses reveal that factors like environmental performance, organizational capital, or regional competition may amplify or attenuate digital transformation effects [34,35,36]. Yet, these analyses are often conducted separately from causal identification strategies, resulting in partial insights.

Moreover, few empirical approaches consider the role of managerial perceptions in shaping digital transformation decisions. In resource-constrained settings, strategic foresight—not just current capacity—can drive adoption. By incorporating managerial expectations about future competition and profitability [11,19], researchers can better explain the variation in adoption under uncertainty.

This study addresses these gaps by employing an Endogenous Switching Regression (ESR) model that simultaneously estimates the determinants of digital transformation adoption and its heterogeneous effects on performance. This method corrects for self-selection based on unobservables and allows for different outcome equations for adopters and non-adopters. Combined with quantile regression, our approach explores distributional effects, offering insights into which firms benefit most from digital transformation. This integration of causal inference with heterogeneity estimation enhances robustness and interpretive power.

To address these empirical challenges, this study adopts an ESR model combined with a quantile-based analysis to provide a more robust assessment of digital transformation effects. While ESR is a known method to account for selection bias, its application in digital transformation studies—particularly in transitional economies—remains limited. By simultaneously modeling adoption and outcomes for adopters and non-adopters, and allowing heterogeneous effects across the performance distribution, our approach captures both the selection and outcome heterogeneity. This integrated strategy offers practical value for empirical research on digital transformation, where both unobserved self-selection and uneven benefits are central concerns. Thus, our contribution lies in refining the identification of digital transformation effects by aligning the model choice with the theoretical complexity of the phenomenon.

3. Methodology

3.1. Data Source

This study uses data from the 2020 Vietnam Enterprise Survey (VES), conducted annually by the General Statistics Office of Vietnam (GSO). The VES is a nationwide census that collects comprehensive information on the operations, inputs, outputs, and financial indicators of all registered firms. The 2020 wave includes a special module on digital transformation, capturing responses from approximately 8000 randomly selected firms. While labeled as 2020, the data primarily reflect firm activities during fiscal year 2019 and early 2020—before Vietnam experienced widespread pandemic-related disruptions. The 2020 wave offers a rare opportunity to analyze firm-level digital transformation using nationally representative data, as it is the first to include detailed metrics on digital platform adoption across sectors.

To ensure analytical focus and comparability, we classify firms into four major sectors: (1) manufacturing; (2) wholesale and retail; (3) information and professional services; and (4) construction, logistics, and hospitality. These sectors are selected based on their strategic importance to Vietnam’s digitalization strategy, their varying levels of digital maturity, and their distinct production and service characteristics. Manufacturing (2370 firms) represents industrial automation and supply chain integration; wholesale and retail (946 firms) capture consumer-facing digital adoption; information and professional services (761 firms) are inherently digital-intensive; while construction/logistics/hospitality (1447 firms) represent traditional sectors undergoing digital catch-up. After excluding firms with missing data or inconsistent responses, the final sample includes 5542 firms, covering all regions and firm size categories, ensuring the robustness and representativeness of the analysis.

3.2. Model Specification and Variable Definitions

The model specification and variable selection in this study are informed by the RBV and the DCV. According to the RBV, platformization enhances the productivity of valuable firm-specific resources (e.g., assets and cash flow) by embedding them in digitally enabled processes. Meanwhile, the DCV emphasizes firms’ abilities to reconfigure and adapt these resources in response to dynamic environments, particularly in resource-constrained and transitional contexts. These perspectives guide our choice of performance measure, model structure, and the inclusion of firm-specific and contextual variables that capture both resource endowments and adaptive capacity.

Platformization in this study refers to the adoption of three specific digital technologies: Supply Chain Management (SCM) systems, Product Data Management (PDM) systems, and IT solutions for logistics and sales. Firms are classified as platformizing if they report using any one of these technologies. While this binary classification does not capture the intensity or scope of digital adoption, it is supported by prior literature that links these technologies directly to operational and financial performance. Integration of SCM and ERP systems has been shown to improve decision-making and enhance firm performance [19], while ERP systems positively influence performance through supply chain integration [20]. SCM also plays a mediating role in improving financial outcomes [21], and PDM systems contribute to better technical data management, product quality, and process efficiency [37,38,44]. These studies collectively support the use of SCM, PDM, and IT systems as valid proxies for digital platformization.

ROS is used as the proxy for firm performance, as it directly reflects operational efficiency and profitability relative to revenue—key dimensions influenced by digital platformization. ROS is particularly suitable in the context of heterogeneous firm sizes and sectoral structures, where profit margins offer a more normalized measure of firm outcomes than absolute profit or revenue. Platformization decisions and performance outcomes are jointly estimated using an endogenous switching regression framework [45], which corrects for potential selection bias by modeling the decision to platformize through a selection equation and estimating separate performance equations for platformized and non-platformized firms. The selection equation is given by

$$p_i^{*}=\gamma Z_i+u_i$$

(1)

where $p_i^{*}$ is a latent variable of which the realization $p_i=1$ if $p_i^{*}>0$ and $p_i=0$ otherwise, and $Z_i$ is the set of covariates explaining the decision to platformize, $\gamma$ the corresponding parameters, and $u_i$ the error term. The set of covariates $Z_i$ includes firm size, financial indicators, owner’s perception regarding digital transformation, and firm characteristics.

Firm size is a significant factor influencing platformization, as larger firms typically possess more resources to invest in digital technologies and face operational complexities that can benefit from platformization (RBV) [11]. In this study, firm size is measured using two proxies: total assets and the number of employees. These proxies capture firms’ financial and human resource capacity, which are critical for implementing digital platforms. Larger firms are generally better positioned to adopt complex digital systems due to their ability to allocate substantial resources toward innovation [4]. Labor, defined as the average number of employees, is included in the platformization decision equation but excluded from the performance equations. This approach is consistent with prior research, which suggests that, while labor is an important factor in strategic decisions, its influence on short-term financial outcomes is less direct [8].

Financial indicators used in this study include leverage, capital intensity, and net cash flow. These variables are essential for understanding a firm’s financial capacity and constraints in adopting digital platforms. Leverage, measured as the ratio of liabilities to total assets, reflects a firm’s financial obligations. High leverage can deter platformization due to the increased financial risk and reduced flexibility it imposes on firms (DCV) [34]. Capital intensity, defined as the ratio of assets to sales, indicates the extent to which a firm relies on fixed assets in its operations. Firms with high capital intensity may adopt platforms to better manage their asset base and enhance operational efficiency (RBV) [17]. Net cash flow, calculated as the ratio of cash to total assets, represents a firm’s liquidity. Strong cash flow enables firms to absorb the upfront costs of digital transformation and sustain ongoing investments required for successful digital adoption [11].

Managerial perceptions of digital transformation’s potential impact on profitability and competitiveness are critical determinants of platformization decisions (DCV). These perceptions are captured through two binary variables. The first variable indicates whether managers believe that digital transformation will increase profitability in the next three years, while the second reflects whether managers expect digital transformation to enhance competitiveness over the same period. These variables serve as proxies for strategic foresight, capturing the extent to which managers anticipate long-term benefits from digital transformation. Previous studies have highlighted the role of managerial expectations in shaping strategic decisions, particularly in uncertain environments where future returns on digital investments are difficult to predict [11,44].

The ownership structure is another key variable influencing platformization (DCV). This study categorizes ownership into three types: state-owned enterprises (SOEs), foreign-owned enterprises, and private domestic firms. SOEs often face bureaucratic inertia, which can slow down the adoption of new technologies despite their access to government resources [46]. In contrast, foreign-owned and private domestic firms tend to be more agile and innovation-driven, making them more likely to adopt digital platforms [8]. The ownership variable allows for an examination of how different ownership types influence the likelihood of adopting digital transformation, reflecting the heterogeneity in strategic behavior across firms.

Location is also a critical determinant of platformization (DCV). In this study, a dummy variable is used to identify firms located in Vietnam’s large cities, including Hanoi, Hai Phong, Da Nang, Ho Chi Minh City, and Can Tho. Firms in these cities benefit from better infrastructure, a more skilled workforce, and greater exposure to competitive pressures, all of which encourage digital adoption [11,47]. By including a location variable, the study accounts for regional disparities in infrastructure and market conditions, providing insights into how urban advantages influence platform adoption rates.

The institutional context is captured through the Provincial Competitiveness Index (PCI), a comprehensive measure of governance quality at the provincial level. Developed by the Vietnam Chamber of Commerce and Industry (VCCI) in collaboration with USAID, the PCI is based on annual surveys of approximately 10,000 businesses and incorporates additional data from government agencies. It evaluates ten key dimensions, including market entry, access to land, transparency, time costs of regulatory compliance, informal costs, and the proactivity of provincial leadership. Higher PCI scores indicate better governance and a more supportive business environment. By including the PCI in the analysis, this study assesses whether governance quality facilitates platformization by reducing institutional barriers and fostering innovation. This variable is particularly relevant in Vietnam, where significant regional differences in governance quality exist. Understanding how these differences affect digital adoption can provide valuable insights for policymakers aiming to create a more business-friendly environment across regions.

The use of the endogenous switching regression model aligns with the DCV by allowing us to estimate heterogeneous performance outcomes across firms with different adaptive capabilities—those who adopt platforms and those who do not—while correcting for selection bias in the platformization decision. The firm performance of the two groups of firms are given by the following:

$$\text{Non-platformizing firms}:{FP}_i^0={\beta }_0{X}_i+{ϵ}_i^0$$

(2)

$$\text{Platformizing firms}:{FP}_i^1={\beta }_1{X}_i+{ϵ}_i^1$$

(3)

In this setting, ${FP}_i^0$ and ${FP}_i^1$ are the ROS of non-platformizing and platformizing firms, $X_i$ is the set of covariates explaining firm performance, and ${\beta }_0$ and ${\beta }_1$ are corresponding coefficients of the two groups. The covariates for the firm performance (ROS) equations of both platformizing and non-platformizing firms are similar to those used in the platformization equation, with some adjustments. First, the assets and capital intensity entered as levels in the selection equation because the probit model is already non-linear, but as logarithm in the firm performance equation. Second, labor and the owner’s perception of digital transformation are excluded, as they are not expected to impact current firm performance directly. Additionally, assets and capital intensity are expressed in logarithmic form to capture diminishing returns as firm size or capital intensity increases. Finally, leverage is included in a quadratic form to account for a potential non-linear relationship, where moderate levels of leverage may enhance performance, but excessive leverage could have negative effects.

3.3. Estimation Method

The error terms ${ϵ}_i^0$ and ${ϵ}_i^1$, together with $u_i$, are assumed to have a trivariate normal distribution with zero mean and a covariance matrix

$$\mathsf{\Omega }=\left[\begin{array}{ccc}{\sigma }_u^2& {\sigma }_{1u}& {\sigma }_{0u}\\ {\sigma }_{1u}& {\sigma }_1^2& .\\ {\sigma }_{0u}& .& {\sigma }_0^2\end{array}\right]$$

where ${\sigma }_u^2$ is the variance of the error term of the platformization equation, which is assumed to be equal to 1, ${\sigma }_1^2$ and ${\sigma }_0^2$ are the variances of the error terms in the firm performance equations, ${\sigma }_{1u}$ is the covariance between ${ϵ}_i^1$ and $u_i$, and ${\sigma }_{0u}$ is covariance of ${ϵ}_i^0$ and $u_i.$ The covariance of ${ϵ}_i^1$ and ${ϵ}_i^0$ is unidentified, because we cannot observe the firm performance for the same firm in both states of platformizing and non-platformizing.

The coefficients of Equations (1)–(3) are estimated simultaneously in the endogenous switching model using the data for $FP$, $X,$ and $Z$ by maximizing the log-likelihood function

$$\mathit{ln}L=\sum _i\left(G_i\left[\mathit{ln}\left\{\Phi \left({\eta }_i^1\right)\right\}+\mathit{ln}\left\{{\displaystyle \frac{\varphi \left(\frac{{ϵ}_{1i}}{{\sigma }_1}\right)}{{\sigma }_1}}\right\}\right]\right)+\left(1-G_i\right)\left[\mathit{ln}\left\{1-\Phi \left({\eta }_i^0\right)\right\}+\mathit{ln}\left\{{\displaystyle \frac{\varphi \left(\frac{{ϵ}_i^0}{{\sigma }_0}\right)}{{\sigma }_0}}\right\}\right]$$

where $\mathsf{\Phi }\left(.\right)$ is a cumulative trivariate normal distribution function, $\varphi (.)$ normal density function, and

$${\eta }_i^j={\displaystyle \frac{\gamma Z_i+\frac{{\rho }_j{ϵ}_i^j}{{\sigma }_j}}{\sqrt{1-{\rho }_j^2}}}j=0,1$$

where ${\rho }_j={\displaystyle \frac{{\sigma }_{ju}^2}{{\sigma }_u{\sigma }_j}}$ is the correlation coefficient between ${ϵ}_i^j$ and $u_i$. The model can be implemented using the movestay command in Stata 15, which is specifically designed for estimating endogenous switching regression models.

3.4. Robustness Check

To examine the robustness and distributional heterogeneity of digital transformation impacts on firm performance, we complement the endogenous switching model with quantile regression analysis. Quantile regression, introduced by [48], estimates covariate effects at different points of the conditional distribution of the dependent variable, rather than focusing solely on the mean. This method is particularly suitable for identifying whether the determinants of firm performance vary across low-performing, median, and high-performing firms. The same set of covariates used in the ROS equations of the switching model is employed here, and estimations are conducted separately for platformized and non-platformized firms within each sector. Quantile regressions are estimated across a wide range of percentiles—from the 10th to the 90th—to capture the full distributional heterogeneity in firm performance, using bootstrapped standard errors to ensure robust inference. This approach allows us to validate the stability of our core findings and uncover performance heterogeneity that mean-based models may overlook.

4. Results

4.1. Descriptive Statistics and Sectoral Patterns

Table 1. Summary statistics.

	Manufacturing (N = 2370)			Wholesale and Retail (N = 946)			Information and Professional Services (N = 761)			Construction, Logistics, and Hospitality (N = 1447)
	Median	P10	P90	Median	P10	P90	Median	P10	P90	Median	P10	P90
ROS (%)	0.55	−12.98	9.02	0.26	−9.57	4.01	0.9	−20.46	17.24	0.38	−19.18	6.9
Assets (mil. US$)	1.42	0.08	15	0.37	0.05	4.75	0.16	0.02	2.2	0.42	0.05	4.76
Labor (workers)	51	5	424	6	2	64	8	2	79	12	2	120
Leverage (Liabilities/Assets)	0.6	0.1	0.96	0.59	0.01	0.94	0.36	0.01	0.94	0.49	0.01	0.9
Capital intensity (Assets/Revenue)	0.97	0.38	3.89	0.79	0.2	4.8	1.09	0.3	6.08	1.4	0.41	9.32
Net cash flow/Assets	0.08	0.01	0.53	0.12	0.01	0.81	0.42	0.04	0.98	0.15	0.01	0.88
Provincial Competitiveness Index (PCI)	63.91	62.23	69.27	63.91	62.34	66.93	63.91	62.78	66.93	63.98	62.3	69.27

presents summary statistics for four chosen sectors, Manufacturing, Wholesale and Retail, Information and Professional Services, and Construction, Logistics, and Hospitality, highlighting key financial and operational indicators. ROS varies across sectors, reflecting the differences in cost structures, pricing strategies, and competitive intensity. The median ROS ranges from 0.26% in Wholesale and Retail to 0.90% in Information and Professional Services, with Manufacturing and Construction, Logistics, and Hospitality reporting 0.55% and 0.38%, respectively.

Firms in Manufacturing are generally larger in scale, with the highest median assets ($1.42 million) and labor force (51 workers), while firms in other sectors operate at smaller scales. Information and Professional Services exhibit the highest liquidity (net cash flow/assets of 0.42) and the lowest leverage (0.36), indicating more stable financial positions. Capital intensity is the highest in Construction, Logistics, and Hospitality (1.4), suggesting heavy investment relative to revenue.

Figure 1 presents the adoption rates of various digital technologies—SCM platforms, PDM platforms, IT in sales, and IT in services—across four major sectors. Platformization levels vary considerably by sector and technology type. Manufacturing leads in SCM (12.15%) and PDM (19.45%) adoption, and also shows the highest usage of IT in sales (48.90%), reflecting its emphasis on integrated production and distribution systems. In contrast, Construction, Logistics, and Hospitality lag behind in all four categories, especially in PDM (8.98%) and IT in sales (29.09%), indicating the slower adoption of core digital infrastructure. Information and Professional Services firms show the highest rate of IT usage in services (50.72%), aligning with the sector’s digital intensity and service-oriented nature. Wholesale and Retail maintain moderate adoption across all technologies, with relatively strong engagement in IT for sales (46.72%) but lower rates in back-end systems like PDM.

4.2. Endogenous Switching Regression Results

Table 2. Estimates of the endogenous switching regression (selection and equation and ROS equations for two groups of firms) by sectors.

	Manufacturing (N = 2370)			Wholesale and Retail (N = 946)			Information and Professional Services (N = 761)			Construction, Logistics, and Hospitality (N = 1447)
	Selection	Firm Performance		Selection	Firm Performance		Selection	Firm Performance		Selection	Firm Performance
		Not Platformized	Platformized		Not Platformized	Platformized		Not Platformized	Platformized		Not Platformized	Platformized
Assets (mil. VND, in logarithm in FP equations)	0.013 *** (0.002)	2.206 *** (0.294)	1.787 *** (0.312)	0.003 (0.002)	1.176 ** (0.47)	2.000 *** (0.504)	0.02 (0.012)	5.01 *** (1.03)	4.952 *** (0.798)	0.003 * (0.002)	1.989 *** (0.428)	1.428 *** (0.444)
Labor (workers)	0.0001 ** (0.0001)			0.003 *** (0.001)			0.0003 (0.0004)			0.0002 (0.0002)
Leverage	−0.082 ** (0.041)	−8.456 *** (1.526)	−12.97 *** (1.572)	0.05 4(0.114)	−1.248 (4.412)	−3.588 (3.71)	−0.057 (0.053)	−12.409 *** (3.039)	−21.54 *** (5.386)	−0.152 ** (0.067)	−8.069 *** (2.302)	−2.696 (3.844)
Leverage squared		0.375 ** (0.177)	0.546 *** (0.146)		−4.916 * (2.51)	−4.98 *** (1.228)		0.69 *** (0.218)	8.035 *** (1.815)		0.266 *** (0.094)	−3.366 ** (1.562)
Capital intensity (assets/revenue, in logarithm in PF equations)	−0.014 *** (0.003)	−2.81 *** (0.486)	−8.053 *** (0.531)	−0.004 (0.003)	−2.735 *** (0.519)	−6.25 *** (0.671)	−0.001 (0.002)	−9.263 *** (1.24)	−8.11 *** (1.18)	−0.008 *** (0.003)	−2.548 *** (0.633)	−1.196 * (0.706)
Net cash flow/Assets	−0.521 *** (0.109)	−3.482 (2.627)	7.391 *** (2.821)	−0.405 *** (0.145)	−5.197 * (2.8)	2.835 (3.219)	−0.247 * (0.136)	13.214 *** (4.802)	21.101 *** (4.282)	−0.243 *** (0.093)	−0.887 (2.648)	10.257 *** (3.055)
Ownership (base = private domestic firm)
State-owned	−0.125 (0.158)	−2.654 (3.795)	−6.823 ** (3.458)	0.331 (0.326)	−5.042 (6.867)	0.742 (5.107)	−0.211 (0.452)	−10.146 (13.432)	−4.417 (12.629)	0.059 (0.188)	4.954 (5.308)	−0.706 (5.481)
Foreign-owned	0.074 (0.059)	3.493 ** (1.574)	−3.147 *** (1.088)	0.15 (0.189)	−3.162 (3.814)	−3.245 (3.285)	0.108 (0.137)	1.076 (4.814)	−1.822 (3.534)	0.268 * (0.144)	4.615 (4.362)	3.672 (3.967)
Located in large cities (1 = Yes)	−0.039 (0.058)	−0.283 (1.437)	−0.394 (1.144)	0.14 (0.087)	2.742 * (1.625)	−1.118 (1.757)	0.259 ** (0.109)	0.281 (3.572)	−9.765 *** (3.413)	0.088 (0.066)	−0.365 (1.839)	−2.73 (2.009)
Provincial competitiveness index	0.003 (0.01)	0.315 (0.246)	0.06 (0.196)	−0.021 (0.015)	−0.493 * (0.275)	−0.056 (0.344)	0.000 (0.021)	−0.047 (0.614)	−0.021 (0.598)	0.002 (0.011)	0.141 (0.316)	0.151 (0.35)
Perception of the impact of digital transformation
on future competition	0.147 *** (0.047)			0.156 * (0.085)			0.434 *** (0.147)			0.104 ** (0.05)
on future profit	0.062 (0.044)			0.178 ** (0.082)			0.144 (0.134)			0.08 * (0.048)
Constant	−0.06 (0.641)	1.573 (15.898)	0.236 (12.704)	1.194 (1.002)	46.721 ** (18.028)	3.927 (21.961)	−0.204 (1.326)	10.293 (40.232)	22.754 (38.832)	−0.278 (0.742)	8.787 (20.531)	10.014 (22.584)
Sigma (${\mathit{\sigma }}_0\mathrm{a}\mathrm{n}\mathrm{d}{\mathit{\sigma }}_1$)		25.402 *** (0.681)	17.49 *** (0.355)		18.836 *** (0.842)	17.269 *** (0.563)		25.742 *** (1.203)	28.108 *** (1.91)		29.308 *** (0.848)	28.51 *** (1.049)
Rho (${\mathit{\rho }}_0\mathrm{a}\mathrm{n}\mathrm{d}{\mathit{\rho }}_1$)		0.958 *** (0.005)	0.114 *** (0.079)		0.89 *** (0.02)	0.084 *** (0.13)		−0.093 *** (0.396)	−0.463 *** (0.209)		0.951 *** (0.006)	−0.944 *** (0.008)
Log-likelihood	−11,562			−4525			−4045			−7215
p-value LR test for overall significance	0.000			0.000			0.000			0.000
LR test of independence of equations	${\chi }_{df=2}^2$ = 371.99, p-value = 0.00			${\chi }_{df=2}^2$ = 58.58, p-value = 0.00			${\chi }_{df=2}^2$= 1.52, p-value = 0.47			${\chi }_{df=2}^2$ = 422.99, p-value = 0.000

Note: Standard errors are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. The dependent variable in the selection equation is binary, where 1 indicates a platformizing firm. The dependent variable in the firm performance equations is ROS expressed as a percentage.

presents the estimates from the endogenous switching regression model for the wholesale, retail, and other sectors. For each sector, the estimates are provided for three equations: the selection equation and the firm performance equations for both platformizing and non-platformized firms. The dependent variable in the selection equation is binary, where a value of 1 indicates a platformizing firm, while the dependent variable in the firm performance equations is ROS. The coefficients from the selection equation are interpreted as a probit regression, while the firm performance equations are interpreted as OLS regressions. The LR test for overall significance rejects the null hypothesis that all coefficients are equal to zero for all four sectors. However, the LR test for the independence of equations fails to reject the null hypothesis (${\rho }_0={\rho }_1=0$) for the Information and Professional Services sector, indicating that the three equations can be estimated separately for this sector. In contrast, this null hypothesis is rejected for other sectors, suggesting that the three equations must be jointly estimated for these sectors.

4.2.1. The Selection Equations

The selection equation estimates the probability that a firm adopts platformization, conditional on firm-level characteristics. The results show that these factors vary in significance and direction across sectors.

In the manufacturing and construction/logistics/hospitality sectors, total assets are positively and significantly associated with platformization, indicating that larger firms in these capital-intensive sectors are more likely to adopt digital platforms. However, asset size is not a significant predictor in the wholesale/retail or information and professional services sectors.

Labor size has a positive and significant effect on platformization in manufacturing and wholesale/retail, but is not significant in the other two sectors. In these labor-intensive sectors, a larger workforce size appears to correspond with a higher likelihood of digital adoption.

Leverage is negatively associated with platformization in manufacturing and construction/logistics/hospitality, suggesting that firms with higher debt levels are less likely to adopt platforms. This relationship is not significant in wholesale/retail or information/professional services.

Capital intensity has a significant negative effect in manufacturing and construction/logistics/hospitality but no significant relationship in the other sectors. Similarly, net cash flow negatively predicts platformization across all four sectors, with the strongest effect observed in manufacturing.

Urban location is positively associated with platformization only in the information and professional services sector, while it is not a significant predictor in the other three sectors. PCI scores are not significantly related to platformization decisions in any sector.

Finally, the perceived importance of digital transformation for future competition has a positive and statistically significant effect on platformization across all sectors. Perceptions of profitability are significant only in the wholesale and retail, and construction, logistics, and hospitality sectors.

4.2.2. The Firm Performance Equations

The results from the firm performance equations reveal distinct patterns across platformized and non-platformized firms in all four sectors. The dependent variable is ROS, and the estimated coefficients capture the marginal effects of explanatory variables on firm profitability. These differences reflect the structural heterogeneity captured through separate outcome equations, rather than formal interaction terms. The model structure allows firm characteristics such as assets, leverage, and cash flow to influence performance differently across digital transformation regimes.

Assets (in logarithmic form) are positively associated with the ROS in all sectors and for both platformized and non-platformized firms. The effect is stronger for non-platformized firms in manufacturing and construction/logistics/hospitality, while platformized firms exhibit higher returns to assets in wholesale and retail. In information and professional services, asset effects are similar across both groups.

Leverage enters the models as a quadratic term. In most sectors, both platformized and non-platformized firms exhibit U-shaped relationships between leverage and ROS, though the turning points and effect sizes vary. In construction/logistics/hospitality, platformized firms show an inverse U-shape, suggesting different leverage dynamics across digital adoption groups.

Capital intensity is negatively associated with ROS in all sectors and both groups. The magnitude of the effect differs by sector and platformization status. In most cases, platformized firms experience less severe performance penalties from capital intensity, although the effect remains negative.

Net cash flow to assets has a positive and significant impact on ROS for platformized firms in manufacturing and construction/logistics/hospitality. Among non-platformized firms, the effect is generally insignificant or negative. In information and professional services, the sign of the effect differs between groups. In wholesale and retail, net cash flow negatively affects ROS for non-platformized firms and is insignificant among platformized ones.

An urban location shows mixed effects. In wholesale and retail, a location in a large city significantly improves ROS for non-platformized firms but not for platformized ones. In information and professional services, a city location negatively affects ROS among platformized firms. In manufacturing and construction/logistics/hospitality, the location variable is not significant. PCI scores are not significantly associated with ROS in any sector or group.

4.3. Robustness Check with Quantile Regression

To assess the robustness of the endogenous switching regression results and explore the distributional heterogeneity in the determinants of firm performance, we estimate quantile regressions for platformized and non-platformized firms separately. Figure 2 presents the estimated coefficients across quantiles (τ = 0.1 to 0.9) for key explanatory variables, with 95% confidence bands. The dashed line represents the OLS benchmark, allowing for a visual comparison between the average and conditional effects.

Across all sectors, the marginal effect of assets on ROS varies across quantiles and platformization status. Among platformized firms in manufacturing- and logistics-related sectors, the asset coefficient is relatively stable across quantiles. For non-platformized firms, the effect of assets on ROS is stronger at higher quantiles. Leverage exhibits nonlinear effects. In manufacturing and services, non-platformized firms show a more pronounced U-shaped relationship across quantiles, while platformized firms demonstrate a flatter pattern. In logistics-related sectors, platformized firms display an inverted U-shape with diminishing effects at the upper quantile.

Capital intensity negatively affects ROS across all quantiles, but the effect is generally smaller for platformized firms, especially at the median and upper quantiles. This pattern is most consistent in the manufacturing and construction/logistics/hospitality sectors. Net cash flow has a consistently positive effect on ROS among platformized firms, particularly in manufacturing and logistics, with the strongest effects at the 75th percentile. For non-platformized firms, cash flow effects are generally weak or negative across quantiles.

City location and PCI have minimal or inconsistent effects across quantiles and sectors. In some cases, the location variable has a weak positive effect at lower quantiles but is not significant at the median or upper ends of the distribution. These quantile-specific results complement the ESR estimates by revealing how the impact of firm characteristics on profitability varies across the performance distribution and by platformization status.

5. Discussions and Implications

5.1. Discussions

Descriptive patterns reveal remarkable differences in size, capital structure, and technology adoption across sectors, reinforcing the need to consider the structural heterogeneity when evaluating digital transformation outcomes. The data show that firms in capital-intensive sectors face high barriers to platformization, while liquidity-rich service firms often delay adoption—suggesting that financial readiness does not always translate into transformation initiatives. These asymmetries justify the use of sector-specific models and have clear policy implications for targeted support mechanisms. These findings resonate with Vietnam’s national digital agenda, which emphasizes sector-specific digital adoption through coordinated state intervention. Strategic policies such as the National Digital Transformation Program and recent initiatives to promote digital human capital development underscore the need for tailored support across industries, especially for capital- or labor-intensive sectors facing high transformation costs.

5.1.1. The Drivers of Platformization Adoption

The results reveal that firm assets influence platformization decisions in a sector-specific manner, reinforcing the importance of resource availability in digital transformation. In manufacturing and construction, logistics, and hospitality, larger asset bases significantly increase the likelihood of platformization, consistent with the RBV, which posits that resource-abundant firms are better positioned to adopt complex technologies [24]. These sectors often require substantial capital investments to integrate digital systems across supply chains and operations. In contrast, wholesale and retail firms show no significant relationship between assets and platformization, possibly because platform adoption in these sectors is driven more by market dynamics and customer-facing strategies than internal resource endowments [49]. Surprisingly, in information and professional services, assets also do not significantly predict digital adoption, suggesting that, in knowledge-intensive sectors, intangible capabilities may matter more than the physical scale [50]. These findings highlight the role of firm assets and call for sector-specific digital strategies that address distinct adoption barriers and resource dependencies.

Labor size influences platformization decisions in a differentiated manner across sectors. It has a positive and significant effect in manufacturing and wholesale and retail, where larger workforces create operational complexity and coordination challenges that digital platforms help manage. In the wholesale and retail sector, this effect is particularly strong, consistent with prior literature emphasizing the need for digital tools to handle sales operations, supply chains, and customer engagement in labor-intensive environments [24]. In the wholesale sector, a larger labor size similarly encourages digital adoption, reflecting the importance of streamlining distribution and inventory systems through platformization [51]. However, in information and professional services and construction, logistics, and hospitality, labor size does not significantly predict platformization, suggesting that other factors—such as digital skill intensity or capital investment—play a more central role in adoption. The positive effects observed in manufacturing and commerce-oriented sectors support the RBV, which argues that leveraging human resources through digital technologies can yield competitive advantages [4], and align with the DCV, which emphasizes adapting internal resources to operational demands in dynamic environments [5]. These results highlight the strategic value of digital platforms in labor-intensive sectors while calling attention to the varying drivers of adoption in service-oriented and project-based industries.

The effect of leverage on platformization is highly sector-dependent. In manufacturing and construction, logistics, and hospitality, higher leverage significantly reduces the likelihood of platformization, indicating that financial constraints deter firms from investing in digital technologies. This finding supports the previous research showing that firms with limited financial flexibility are less likely to undertake transformative investments due to the perceived risk and upfront costs associated with digital adoption [30,35]. In contrast, in wholesale and retail and information and professional services, leverage does not significantly influence platformization decisions. This may suggest that firms in these sectors are more responsive to operational demands or competitive pressures than to capital structure considerations when pursuing digital transformation [1]. In particular, wholesale firms may prioritize supply chain integration and external collaboration—objectives that can override the limiting effects of financial leverage [51]. Overall, these results emphasize the importance of financial health in enabling digital adoption in asset-heavy sectors and point to the need for targeted financial support policies to reduce leverage-related barriers.

Capital intensity significantly and negatively affects platformization in the manufacturing and construction, logistics, and hospitality sectors. Firms in these capital-intensive industries often face difficulties reallocating financial and managerial resources toward digital initiatives, as existing physical assets demand ongoing investment and attention. This reflects the “legacy burden” described in the literature, whereby firms with entrenched capital structures may struggle to pivot toward disruptive technologies [50]. In contrast, capital intensity does not significantly influence platformization in the wholesale and retail or information and professional services sectors, where digital adoption is more closely tied to operational drivers such as customer engagement and supply chain efficiency [1].

Across all four sectors, net cash flow exerts a negative effect on platformization, with the most pronounced impact observed in the manufacturing sector. This suggests that firms with stronger internal liquidity may prefer to invest in conventional or short-term improvements rather than in uncertain or disruptive digital initiatives. The result aligns with the so-called “Digital Transformation Paradox”, which posits that financially flexible firms are not always the earliest adopters of digital technologies, particularly when transformation entails risk or organizational disruption [46,51]. Together, these findings highlight that both the composition of firm assets and internal cash availability shape digital investment decisions, emphasizing the need for sector-sensitive incentives to address structural and behavioral barriers to platformization.

Only firms in the information and professional services sector are more likely to adopt digital platforms when located in large urban areas. This finding reflects the urban concentration of digital infrastructure, high-skill labor, and innovation ecosystems—factors particularly critical for service-oriented and knowledge-intensive firms. In contrast, location does not significantly influence platformization in the manufacturing, retail, or wholesale sectors, where adoption decisions may depend more on operational needs or internal firm characteristics than on urban proximity. Retail and wholesale firms, in particular, may focus more on streamlining logistics, improving customer experience, or managing supply chains, regardless of location. These findings suggest that urban advantages matter most for sectors that depend on information flows and human capital, supporting the view that locational effects are contingent on sectoral demands and business models [50]. This finding aligns with Vietnam’s digital transformation strategy, which has prioritized the expansion of urban infrastructure and high-skill digital ecosystems in major cities. However, complementary policies—such as the national plan to reduce regional disparities in digital literacy and access—are necessary in order to ensure that rural and industrial regions can also benefit from platform adoption.

The PCI, which reflects the quality of the institutional and business environment, does not significantly influence platformization decisions across any sector. This finding aligns with the literature suggesting that firm-level characteristics and strategic priorities, rather than local institutional factors, are the primary drivers of platform adoption [49]. These findings suggest that sector-sensitive digital policies are necessary in order to address financial and structural constraints, particularly in lagging sectors—supporting more equitable and sustainable industrial upgrading.

Firm owners’ perception of digital transformation’s impact on future competition significantly increases the likelihood of platformization across all sectors, highlighting the role of managerial foresight in digital adoption [44]. This aligns with the DCV, which emphasizes sensing and responding to environmental shifts as critical to strategic adaptation [8]. In contrast, perceptions of future profitability influence platformization only in wholesale/retail and construction/logistics/hospitality, where firms may weigh short-term financial returns more heavily. In manufacturing and information/professional services, this effect is not significant, possibly due to the longer adoption horizons or different performance expectations. These findings emphasize the consistent role of competitive expectations in driving platformization and suggest the importance of policies that enhance business leaders’ understanding of digital transformation’s strategic value. This highlights the relevance of recent national initiatives to raise awareness and foster digital foresight among business leaders—especially in the SME sector—through government-backed training and skill development programs.

5.1.2. Platformization and Firm Performance

The firm performance equations show that assets (log) significantly improve ROS across all sectors, though the magnitude differs by platformization status. In manufacturing, the effect is stronger for non-platformized firms (2.206) than for platformized ones (1.787). This implies that a 1% increase in assets is associated with a 0.022 percentage point increase in ROS, reflecting the high capital productivity in traditional operations. However, the smaller coefficient for platformized firms suggests that, while digital transformation enhances operational coordination, its marginal gains on asset returns may taper off in capital-intensive sectors.

In wholesale and retail, the pattern reverses: platformized firms experience a higher asset–ROS elasticity (2.000) compared to non-platformized firms (1.176), likely due to the enhanced efficiency from the digital inventory and sales platforms [44,46]. In information and professional services, assets are strongly associated with ROS in both groups (5.01 and 4.952), indicating that digital adoption does not significantly alter the asset–performance relationship in knowledge-intensive firms. In construction, logistics, and hospitality, platformization moderately reduces the asset–ROS effect (from 1.989 to 1.428), suggesting that the benefits of digital systems may be dampened by integration lags or fragmented operations. These findings align with the RBV [3,4], which emphasizes that firm resources—augmented by digital tools—can yield superior performance, while also highlighting sectoral variations in the return on asset utilization through platformization.

The relationship between leverage and ROS varies across sectors and follows a quadratic pattern, with distinct shapes depending on platformization. In manufacturing, both platformized and non-platformized firms exhibit U-shaped relationships, with leverage initially decreasing the ROS and then increasing it at higher levels. This suggests that, while moderate debt burdens reduce profitability, firms may become more efficient or disciplined at higher leverage levels—particularly when supported by digital systems that improve resource utilization [24]. The turning point occurs at approximately 11.3% for non-platformized and 11.9% for platformized firms, with stronger negative effects observed among platformized firms, possibly reflecting riskier investment behavior enabled by technology [50].

In wholesale and retail, the relationship is also U-shaped for both groups, but only marginally significant. The effect is modest and statistically weaker than in other sectors, suggesting that leverage plays a limited role in explaining profitability, likely due to the thinner margins and higher operational volatility. The information and professional services sector shows a pronounced U-shaped pattern for both groups, especially for platformized firms, where leverage initially suppresses ROS but sharply increases beyond the turning point. This reflects the sector’s sensitivity to the financial structure and the benefits of digital adoption in managing cost structures [1]. In construction, logistics, and hospitality, the patterns diverge: non-platformized firms show a mild U-shape, while platformized firms exhibit an inverse U-shape, with performance peaking at moderate leverage levels before declining. This suggests that digital tools may help firms optimize resource use only up to a point, beyond which excessive debt burdens outweigh efficiency gains. These findings highlight that the leverage–performance relationship is highly non-linear and sector-specific. Moreover, platformization modifies this relationship, either by amplifying performance gains at higher debt levels or by mitigating risks through efficiency improvements. This supports the view that digital transformation can enhance financial flexibility but also introduces new risk–return trade-offs [1,50].

The impact of capital intensity (log) on ROS is consistently negative across all sectors, with varying magnitudes depending on the platformization status. In manufacturing, a 1% increase in capital intensity reduces the ROS by 0.028 percentage points for non-platformized firms and by 0.081 percentage points for platformized firms, indicating that digital adoption does not offset the performance costs of asset-heavy operations. In wholesale and retail, the effect is similarly negative, with coefficients of −0.027 and −0.0625 for non-platformized and platformized firms, respectively. These findings suggest that, while platformization may improve process efficiency, it does not fully mitigate the strain imposed by capital-intensive models—particularly when digital systems themselves require costly integration and maintenance [24].

In information and professional services, capital intensity also reduces the ROS significantly, with effects of −0.093 for non-platformized firms and −0.081 for platformized ones. This is notable given the sector’s relatively low physical asset base, implying that excess investment in fixed assets may crowd out more flexible, knowledge-based expenditures. In construction, logistics, and hospitality, capital intensity negatively affects the ROS in both groups, though less severely among platformized firms (−0.025 vs. −0.12), possibly due to the operational gains from digital tools that improve scheduling and resource coordination. Overall, these results show that capital intensity reduces profitability across sectors, and that platformization does not fully counteract this effect. This aligns with the literature noting that digital transformation, while improving resource utilization, can introduce additional cost burdens in asset-heavy environments [24]. The findings highlight the need for firms to align platformization with strategic capital management, particularly in sectors where asset rigidity risks undermining the efficiency gains from digital adoption [1].

The impact of net cash flow (as a ratio to assets) on the ROS is both sector-specific and platformization-dependent, with notable differences in direction and magnitude. In manufacturing and construction, logistics, and hospitality, cash flow significantly improves the ROS only for platformized firms, suggesting that digital adoption enhances firms’ ability to convert liquidity into performance gains through improved allocation, real-time tracking, and process integration [50]. For non-platformized firms in these sectors, the effect is statistically insignificant, indicating a limited capacity to translate liquidity into strategic investments or operational improvements. In information and professional services, cash flow significantly affects the ROS in both platformized and non-platformized firms, but with opposite signs: positive for platformized firms and negative for non-platformized ones. This pattern suggests that, in the absence of digital systems, firms may misallocate liquidity or face inefficiencies that erode performance, whereas platformization enhances financial agility and precision [50]. Interestingly, in wholesale and retail, cash flow significantly reduces ROS for non-platformized firms, possibly due to liquidity being tied up in inventory or inefficient operations [24]. In contrast, platformized firms in this sector show no significant relationship, implying that digital adoption may neutralize, but not fully leverage, liquidity advantages. These findings highlight the strategic role of platformization in enhancing the productivity of cash flow and mitigating inefficiencies. They also point to the risk that, without digital integration, even strong liquidity positions may fail to improve—or may even harm—profitability. This reinforces the need for digital tools to unlock the performance potential of financial flexibility [24,50].

The effect of a location in large cities on the ROS is sector- and platform-dependent, with mixed and sometimes counterintuitive results. In wholesale and retail, a location in a large city significantly increases ROS for non-platformized firms by approximately 2.74 percentage points, suggesting that urban advantages—such as access to infrastructure, labor, and dense markets—play a key role in enhancing performance for firms not yet digitally transformed [24,50]. However, for platformized firms in the same sector, the effect is not significant, implying that digital platforms may substitute for some locational advantages by enabling a broader market access and operational efficiency. In information and professional services, the relationship reverses: being located in a large city reduces the ROS by 9.77 percentage points for platformized firms, while the effect is not significant for non-platformized ones. This may reflect the intensified competition and higher operating costs in urban professional markets, which digital tools alone may not offset [50]. In manufacturing and construction/logistics/hospitality, a city location has no significant impact on the ROS, regardless of platformization, suggesting that urban proximity plays a limited role once sectoral operations are standardized or distributed. Overall, these findings indicate that the performance benefits of an urban location diminish—or even reverse—once firms adopt digital platforms, highlighting the potential of digital transformation to either mitigate or reshape geographic advantages.

The PCI, however, does not significantly influence the ROS in any sector, aligning with findings that institutional quality affects broader economic readiness rather than directly driving firm-level performance metrics like sales [51]. These results emphasize digital transformation’s role in reducing urban–rural disparities while highlighting the primacy of firm-level strategies, such as platformization, over institutional factors in enhancing sales performance. Although institutional quality, as measured by the PCI, does not show significant effects, this may reflect the relatively recent rollout of Vietnam’s digital governance reforms. Emerging policy frameworks—such as regulatory incentives for science, innovation, and digital integration—are likely to shape firm performance more substantially over time as implementation deepens across provinces.

5.1.3. Distributional Effects of Platformization

The results reveal substantial heterogeneity across the ROS distribution, and,, importantly, support the main findings while offering additional insights. Assets consistently exhibit a positive impact on the ROS across quantiles in both groups. However, the effect is larger and steeper in lower quantiles, especially among non-platformized firms, indicating that asset accumulation has diminishing marginal returns. In platformized firms, the curve is flatter, suggesting that digital systems may standardize asset efficiency and reduce performance dispersion. This pattern reinforces the RBV-based insight that platformization enhances the productivity of physical resources but may also flatten performance gradients.

Leverage and leverage squared show distinct curvature in platformized firms, especially at the upper quantiles. The U-shaped relationship becomes more pronounced, with the ROS initially declining and then increasing with higher leverage—suggesting that digital capabilities help high-leverage firms manage financial stress more effectively. This non-linearity is muted among non-platformized firms, where leverage shows weak or inconsistent effects. These patterns are consistent with our switching model results and reinforce the importance of the financial structure in shaping digital returns.

Capital intensity has a negative effect throughout the distribution in both groups, more severe at the lower quantiles and especially pronounced in non-platformized firms. This suggests that capital-heavy firms without digital integration suffer greater inefficiencies, whereas platformization partially mitigates these performance penalties—aligning with the previous findings on legacy burden and digital inefficiencies. Net cash flow over assets displays divergent patterns: it has no effect or negative effects among non-platformized firms but becomes increasingly positive and significant among platformized firms at higher quantiles. This finding strengthens the argument that digital systems enhance liquidity utilization and operational agility, especially in top-performing firms. Other variables, such as foreign ownership, state ownership, PCI, and urban location, show limited and statistically inconsistent effects across quantiles. However, the broad stability of these coefficients provides reassurance that our baseline results are not driven by outliers or distributional skews.

Overall, the quantile regression analysis confirms the robustness of our main findings and further reveals that digital transformation alters not only the mean but the distributional structure of firm performance. Platformization systematically reshapes how firm-level factors translate into profitability, particularly at the tails of the performance distribution. These results strengthen the theoretical and empirical case for adopting nuanced, distribution-aware methods when evaluating digital transformation impacts. These distributional insights are essential for policy frameworks aiming to ensure that the benefits of digital transformation are not limited to top-performing firms, but are extended across the performance spectrum to support inclusive and sustainable economic growth.

5.2. Academic Contributions

This study contributes to the growing literature on digital transformation and firm performance in several important ways. First, by focusing on platformization as a specific digital strategy, it deepens our understanding of how digital transformation manifests in transitional economies. While prior research often considers digital transformation as a broad or aggregated concept [24,32], our analysis disaggregates platformization as a distinct form of digital upgrade that restructures value chains, operational processes, and customer engagement. This micro-level view adds nuance to studies that document the strategic and performance implications of digital adoption.

Second, we integrate RBV and DCV frameworks to explain both the decision to adopt platformization and the heterogeneous performance outcomes that follow. This dual-theory approach provides a systematic foundation for modeling strategic digital adoption not just as a technology choice but as a capability-driven transformation. While RBV has been frequently referenced in digital transformation literature [27,30], few studies combine it with DCV in a way that is explicitly operationalized in the empirical strategy. Our use of ESR to model this joint decision–performance structure is theoretically motivated and empirically grounded.

Third, our findings highlight significant heterogeneity in digital transformation outcomes, especially in the interaction with firm characteristics such as cash flow, leverage, and asset base. Unlike many previous studies that focus on the average treatment effects [28,33], we identify how firms at different performance quantiles experience differential gains from platformization. This reinforces the idea that digital transformation is not universally beneficial and supports the emerging literature emphasizing firm-specific thresholds, absorptive capacity, and structural complementarity [25,26].

Fourth, by applying the ESR model in combination with quantile regression, this study advances methodological application in estimating DT effects. Many empirical studies rely on OLS or DiD designs, which may not fully address self-selection or performance heterogeneity. Our approach directly responds to recent meta-analytical critiques [22] that call for causal identification strategies suited to the dual challenge of endogeneity and asymmetric outcomes—especially in emerging and transitional contexts.

Finally, the Vietnamese context adds empirical richness to the global digital transformation literature, which remains disproportionately focused on China, the U.S., and the EU. Our micro-level evidence complements national policy efforts such as the National Digital Transformation Program (2020–2025) and the country’s science, technology, and innovation strategy, illustrating how government priorities for digital upgrading interact with firm-level strategies and performance outcomes. By situating our findings within Vietnam’s socioeconomic and policy context, this study offers grounded insights with broader relevance for other developing economies undergoing similar transitions.

5.3. Managerial Implications

The findings of this study offer several actionable insights for managers considering digital transformation through platformization. First, firms should assess their organizational and financial readiness before adopting platform-based solutions, as performance gains are contingent on internal capabilities such as asset strength, liquidity, and leverage management. Second, managers can leverage digital platforms to unlock the performance potential of existing liquidity, particularly in manufacturing, construction, and logistics-related sectors. However, in capital-intensive firms, digital transformation should be pursued strategically—not as an additional investment layer, but as a means to optimize asset utilization. Third, the benefits of platformization vary by sector and location: while rural and less urban firms may use digital tools to overcome locational disadvantages, professional service firms in urban centers may face diminishing returns due to the high competition and cost pressures. Finally, firms with a moderate to high leverage may find performance improvement through digital transformation, but must balance digital investments carefully to avoid financial overextension. Overall, platformization should be integrated into a firm’s broader resource strategy and adapted to its specific sectoral conditions, financial constraints, and operational priorities.

6. Conclusions

This study provides a new understanding of the drivers and impacts of platformization on firm performance and makes both theoretical and methodological contributions, while offering evidence from the context of a developing economy. Theoretically, it advances the RBV and DCV by illustrating how platformization interacts with tangible and intangible resources to optimize firm performance. The research highlights that platformization amplifies the positive effects of assets and liquidity on sales performance, particularly in capital-intensive sectors, while mitigating the adverse impacts of a high leverage and locational disadvantages. Sector-specific findings from manufacturing, wholesale and retail, information and professional services, and construction/logistics/hospitality extend the literature by revealing that the effects of digital adoption on resource utilization and profitability are highly heterogeneous. These results emphasize the importance of aligning digital transformation strategies with sectoral and operational realities to support sustainable and inclusive industrial upgrading.

Methodologically, the study demonstrates the advantages of the ESR over a traditional moderation analysis and IV regressions. By simultaneously estimating firm performance equations for platformized and non-platformized firms, this approach captures the structural heterogeneity and selection bias, offering a more nuanced understanding of platformization’s effects. Additionally, the inclusion of quantile regressions reveals distributional differences in performance impacts across the sales performance spectrum, showing that platformization has more pronounced benefits for firms at the median and upper quantiles. These findings address the gaps in prior research that focused only on mean effects and uniform interaction terms, offering a more robust framework for evaluating digital transformation. By focusing on Vietnam—a rapidly developing economy undergoing significant digital and structural transformation—the study contributes context-sensitive insights into how platformization supports the broader processes of market reform and economic upgrading. This contextual relevance enhances its applicability to other transitional economies, offering a more robust framework for evaluating digital transformation within the context of equitable economic resilience and structural reform.

From a sustainability perspective, the study contributes to the understanding of how digital transformation can support economic and social sustainability in emerging markets. Platformization enhances resource efficiency, reduces the spatial inequalities in performance, and promotes the broader access to digital infrastructure and tools. These outcomes are especially relevant to SDG 8 (Decent Work and Economic Growth) and SDG 9 (Industry, Innovation, and Infrastructure), where platformization plays a strategic role in improving productivity, expanding access to digital infrastructure, and reducing firm-level disparities. By identifying heterogeneous effects across sectors and firm types, the study supports the design of more inclusive and equitable digitalization strategies.

From a policy standpoint, the findings provide actionable guidance. First, firms with a high capital intensity or leverage should integrate platformization with effective cost and financial management to fully realize performance gains. Second, the influence of managerial perceptions on platformization underscores the need for programs that develop strategic foresight among business leaders. Training and awareness initiatives can help managers better understand the long-term value of digital transformation, especially in under-digitized sectors. Third, while platformization reduces firms’ dependence on urban advantages, continued investments in digital infrastructure and talent development in non-urban areas remain critical for equitable transformation. By identifying the conditions under which digital adoption leads to more equitable and efficient outcomes, this study contributes to the broader sustainability agenda and provides evidence for designing inclusive digital development strategies in transitional economies.

Despite its contributions, the study has limitations that suggest directions for future research. The analysis is based on data from a single national context, which may affect the generalizability of the results. Comparative studies across countries with varying institutional and technological landscapes could offer deeper insights into the contextual determinants of platformization. Furthermore, the binary treatment of platformization does not capture the full range of digital adoption. Future work should consider more granular, continuous measures to assess the intensity and sophistication of platform use. Lastly, while this study emphasizes managerial perceptions, further research could explore how leadership traits, organizational readiness, and environmental pressures interact to shape digital transformation trajectories.