Financial and Entrepreneurial Capability Configurations: An Exploratory Study of the Selective International Integration Among Thai Smes

1. Introduction

Small and medium-sized enterprises (SMEs) play a central role in economic development in emerging economies, and Thailand is no exception. SMEs contribute more than one-third of the national GDP and absorb a substantial share of the labor force, making their competitiveness critical for sustaining growth amid increasing globalization. Although Thailand’s SMEs have maintained a stable contribution to GDP, with an average share of 34.56% and an average growth rate of 4.5% between 2017 and 2022, domestic economic dynamism alone is insufficient for long-term resilience. As competitive pressures intensify, internationalization has become a strategic pathway for growth, risk diversification, and technological upgrading.

Figure 1. GDP of SMEs in Thailand GDP. (Source: Office of the National Economic and Social Development Council).

Figure 1. GDP of SMEs in Thailand GDP. (Source: Office of the National Economic and Social Development Council).

Despite the importance of this process, existing internationalization research has been disproportionately shaped by firms that have already succeeded in foreign markets (Khanthong, J., & Sakulkijkarn, W., 2024). Foundational theories—such as the Uppsala Internationalization Model, the Resource-Based View (RBV), and the Dynamic Capabilities Framework (DCF) primarily describe how firms accumulate knowledge, develop routines, and gradually increase foreign commitments (Johanson & Vahlne, 1977; Barney, 1991; Teece et al., 1997). While these perspectives have advanced understanding of the internationalization trajectory, they largely assume that expansion abroad is a natural or desirable progression. As a result, the sizable population of SMEs that remain exclusively domestic has received far less theoretical and empirical attention (Ferreira et al., 2020). The first gap in the literature, therefore, concerns asymmetry; we know considerably more about SMEs that internationalize than those that choose not to.

This omission is particularly consequential in emerging markets. Many Thai SMEs operate successfully at home yet hesitate to expand abroad due to capability shortages, limited resources, market uncertainty, or institutional weaknesses. Domestic orientation may reflect rational strategic choices rather than failure to grow (Tan & Sousa, 2015). However, the current literature does not clearly differentiate whether non-internationalization arises from financial constraints, insufficient entrepreneurial and organizational capabilities, or structural barriers embedded in the emerging-market context. This leads to a second gap: the domestic SME segment remains theoretically underexplored, with a limited understanding of the capability combinations that keep firms localized.

A third gap arises from the fragmented treatment of determinants of internationalization. Financial factors such as profitability, liquidity, leverage, and efficiency are typically analyzed separately from entrepreneurial, organizational, and relational capabilities. In practice, however, the strategic orientation of SMEs is shaped by the interaction between these financial and non-financial dimensions markets (Pascucci et al., 2021; Falavigna et al., 2024). Literature does not provide an integrated, empirically grounded framework that can capture how these multidimensional capability sets jointly determine whether SMEs operate internationally or domestically (Tukamuhabwa et al., 2021; Da Rocha et al., 2024).

Beyond these three classical gaps, a deeper conceptual gap underlies the SME internationalisation literature: the absence of a framework that explains why integration into wider economic space is selective, conditional, and configurational rather than uniform. Particularly, modern globalization is neither universally accessible nor uniform. Rather than being a linear expansion path that is accessible to every capable firm, cross-border participation is now increasingly perceived as contingent, selective, and irregular (Haddad, 2023; Gupta & Guidi, 2012). Recent research on global market interdependence indicates that integration and co-movement are dynamic and crisis-sensitive and are determined by the underlying institutional and geopolitical positioning rather than by straightforward market entry (Gupta, Haddad, & Selvanathan, 2024).

Globalization generates opportunities, but those opportunities are unevenly translated into firm-level participation because firms differ in the combinations of financial flexibility, organisational preparedness, and relational embeddedness they can mobilise. This logic resonates with the broader literature on global financial market integration, which treats integration as time-varying, structurally asymmetric, and shaped by underlying systemic positioning rather than as a stable convergence outcome (Haddad, 2023; Gupta, Haddad, & Selvanathan, 2024). Reframing SME participation in this way directs attention to how firm capability bundles condition systemic coupling, and away from a binary export-versus-domestic reading of internationalisation.

This study addresses these gaps by examining the capability configurations that distinguish internationalized Thai SMEs from those that remain localized. Drawing on financial ratios and non-financial indicators related to international experience, team readiness, operational adaptability, risk tolerance, and institutional connectivity, the paper employs an inductive, machine learning–based approach to identify the capability patterns associated with each group. Extreme Gradient Boosting (XGBoost) facilitates the analysis of non-linear interactions among financial and entrepreneurial dimensions, revealing structures that traditional linear methods may overlook (Chen et al., 2022). This study extends understanding of SME internationalization by shifting attention to configurational patterns rather than isolated variables (Keelson et al., 2024; Yildiz et al., 2023; Szkudlarek et al., 2020).

2. Literature Review

2.1. Theoretical Framework

SME internationalization has long been interpreted through several foundational theories that emphasize different dimensions of firm behavior. The Resource-Based View (RBV) argues that sustainable competitive advantage arises from valuable, rare, inimitable, and non-substitutable resources (Barney, 1991). Within this framework, profitability and resource deployment reflect managerial competence and the firm’s ability to finance international risk. However, the RBV assumes relative resource stability, limiting its explanatory power in volatile environments.

To address dynamic contexts, the Dynamic Capabilities Framework (DCF) conceptualizes advantage as the firm’s ability to integrate, reconfigure, and renew its resource base in response to environmental change (Teece et al., 1997; Teece, 2023). Dynamic capabilities such as opportunity sensing, resource orchestration, and strategic realignment are central to competing in foreign markets where uncertainty, regulatory heterogeneity, and institutional voids are common. For SMEs, dynamic capabilities translate into operational adaptability, organizational learning, and proactive opportunity development.

Financial theories complement these strategic perspectives. Pecking Order Theory (POT) suggests that firms prioritize internal financing and resort to external debt only when necessary, implying that liquidity constraints limit high-risk expansion into foreign markets (Myers, 1984; Myers & Majluf, 1984). Agency Theory (AGT), by contrast, highlights the disciplining role of moderate leverage in enhancing efficiency and growth potential (Jensen & Meckling, 1976; Jensen, 1986).

POT posits that firms prefer internal financing, implying liquidity constraints may limit internationalization (Myers, 1984; Myers & Majluf, 1984). Conversely, AGT argues that moderate debt enhances discipline and efficiency, potentially fostering international expansion (Jensen & Meckling, 1976; Jensen, 1986). Hence, while financial constraints hinder globalization, prudent leverage can promote it. From a process perspective, the UPM and Organizational Learning Theory (OLT) emphasize experiential learning and gradual internationalization (Johanson & Vahlne, 1977; Lyles, 1990). Yet, “born global” SMEs can bypass this incremental path through digitalization and dynamic capabilities. The revised UPM reflects globalization’s realities, highlighting non-linear, network-based internationalization where strong relationships enable market “leapfrogging” (Johanson & Vahlne, 2009).

Together, these theories present a multidimensional framework: RBV and DCF explain internal and adaptive strengths; POT and AGT address financial discipline and constraints; UPM and OLT illuminate experiential and learning dynamics. The integration of these perspectives shows that SMEs’ international or domestic orientation depends on both financial foundations and non-financial capabilities, particularly resources, experience, networks, and adaptability. The model below proposes a combined theoretical framework to explain SMEs’ internationalization or localization orientation.

Figure 1. Integrated Research Framework.

Figure 1. Integrated Research Framework.

2.2. Conditional Globalisation, Capability Bundles, and Selective Integration

The theories surveyed above (RBV, DCF, POT, AGT, UPM, and OLT) collectively explain why firms differ in their willingness and capacity to engage foreign markets. Yet, taken together, they speak to a deeper question that frames this study: why do firms vary in the degree to which they become coupled with broader economic systems? Recent work on global financial market integration shows that integration is best understood as an evolving, conditional process rather than a fixed state, with co-movement and interdependence varying across time, regimes, and structural positions (Haddad, 2023). In a similar vein, evidence from Asian and developed equity markets demonstrates that interdependence is dynamic, time-varying, and crisis-sensitive rather than a one-directional convergence story (Gupta & Guidi, 2012). Translated to the SME context, these insights suggest that internationalisation is not a stable binary outcome but a contingent positioning outcome shaped by underlying capability bundles.

A second insight from this stream concerns the structured and asymmetric character of co-movement across units. Gupta, Haddad, and Selvanathan (2024) show that integration in the world economy is shaped by underlying institutional and geopolitical positioning, producing uneven, hierarchical patterns of co-movement rather than homogeneous convergence. By analogy, the alignment of SMEs with transnational economic structures is unlikely to be uniform; instead, it is mediated by firm-level capability bundles—financial agility, operational adaptability, human capital, market knowledge, and institutional embeddedness—that condition the depth and quality of systemic coupling.

A third implication concerns system-level positioning. Even though the work cited above is empirically located in financial markets and macro-level country studies (Paramati, Mo, & Gupta, 2017), the conceptual lesson translates well to SMEs in emerging economies: participation in globalisation is shaped by where actors sit within wider structures of opportunity and constraint. For Thai SMEs, this position is jointly determined by financial structure, entrepreneurial readiness, learning trajectories, and institutional ties. Internationalised SMEs are therefore better understood as more deeply coupled with cross-border information, relational, and capital networks, while domestically oriented SMEs may remain only partially or selectively connected—not necessarily as deficient firms, but as actors positioned differently within a conditionally globalised system. This conceptual repositioning provides the analytical bridge between the firm-level capability literatures reviewed in Section 2.1 and the empirical patterns identified by the machine-learning model in later sections.

2.3. Proposed Factors

This section identifies key factors, supported by prior studies, to be integrated into the analytical framework, reflecting the combined theoretical perspectives. The literature emphasizes that both financial and non-financial indicators are crucial in predicting and shaping SMEs’ strategic orientation toward either international or domestic markets.

• Financial factors (FR)

Financial factors are central to SMEs’ readiness for internationalization because they shape resource availability, capital access, and the capacity to absorb internationalization costs. Indicators of profitability, liquidity, leverage, asset utilization, and cost control capture a firm’s internal financial health and strategic flexibility. Evidence consistently shows that SMEs with stronger financial structures are more capable of sustaining global expansion, while domestically oriented firms often face inefficiency or liquidity pressures that restrict risk-taking.

Within the RBV, profitability ratios—return on assets (ROA, FR1), return on equity (ROE, FR2), gross profit margin (GPM, FR3), operating profit margin (OPM, FR4), and net profit margin (NPM, FR5)—reflect the firm’s ability to generate and mobilize internal resources (Ahmed et al., 2024; Daadmehr, 2024; Govindan et al., 2023; Hassan et al., 2023; Ahmad et al., 2022). High profitability signals efficient resource use and managerial competence (Barney, 1991), supports scale, and enables risk diversification in foreign markets (Hitt et al., 2006; Yildiz et al., 2023). Empirical evidence confirms that profitable SMEs more easily overcome entry barriers and adapt to global competition (Keelson et al., 2024; Buckley & Casson, 2020), whereas low-profit firms remain domestically constrained (Brozović et al., 2023; Falavigna et al., 2024). Profitability has long been shown to influence export intensity (Aybar & Thirunavukkarasu, 2005; Riahi-Belkaoui, 1998; Lu & Beamish, 2006), reaffirming the importance of FR1–FR5.

Liquidity and leverage jointly determine SMEs’ ability to sustain international commitments. The Pecking Order Theory argues that limited liquidity restricts foreign investment (Leong & Yang, 2021; Merkoulova & Zivanovic, 2022), with studies showing that financing constraints hinder both entry and persistence in foreign markets (Fauceglia, 2015; Pascucci et al., 2021). Key indicators include the current ratio (CRR, FR6), debt-to-equity (DTE, FR7), debt-to-assets (DTA, FR8), and equity multiplier (EM, FR9) (Rashid & Jabeen, 2018; Kabbach-de-Castro et al., 2022; Yang & Chen, 2023; Miranda-García & Segovia-Vargas, 2024). High liquidity reduces financing costs and increases resilience (Minetti & Zhu, 2011; Manova, 2012; Pascucci et al., 2021; Mansilla-Fernández & Milgram-Baleix, 2023), whereas, from an agency perspective, moderate leverage can enhance discipline, though excessive debt increases risk and limits export capacity (Guedhami et al., 2022; Hassen et al., 2024).

SMEs frequently face structural credit constraints that limit long-term investment (Forte & Salomé Moreira, 2018) and increase vulnerability to cash flow shocks (Rodeiro-Pazos et al., 2023; Al-Hyari et al., 2012). Although high liquidity supports survival, it often reduces strategic investment capacity. Evidence shows that SMEs with balanced leverage and sufficient liquidity are better positioned to exploit foreign opportunities (Fauceglia, 2015), while low liquidity combined with high leverage encourages defensive, domestic-focused strategies (Kim, 2016; Forte & Salomé Moreira, 2018). The joint assessment of FR6–FR9 thus provides insight into internationalization readiness.

Asset utilization efficiency and cost control also influence competitiveness. Total asset turnover (TAT, FR10), receivables turnover (ART, FR11), inventory turnover (INT, FR12), payables turnover (APT, FR13), and cost-to-revenue ratio (CRT, FR14) measure a firm’s ability to convert resources into outcomes (Kotane & Kuzmina-Merlino, 2017; Hyz et al., 2018; Breivik et al., 2023; Hossain & Sultana, 2024; Káčer & Alexy, 2025). High efficiency strengthens domestic competitiveness and supports export readiness (Fernández-López et al., 2020; Rodeiro-Pazos et al., 2023; Younis & Karmowska, 2025), whereas weak efficiency and long cash conversion cycles hinder international engagement (Kahiya & Dean, 2016; Mansilla-Fernández & Milgram-Baleix, 2023). While cost discipline is valuable, excessive rigidity may suppress innovation and slow foreign expansion (Rice et al., 2024; Hauser et al., 2013).

Taken together, the financial variables FR1–FR14, as shown in Table 1., provide a multidimensional assessment of SME financial health. Profitability supports resource mobilization, liquidity and leverage shape financial resilience, and efficiency enhances competitiveness. These factors collectively indicate whether SMEs possess the flexibility, stability, and resource strength needed to compete internationally.

• Non-financial factors (NF)

Non-financial factors shape the strategic and organizational foundations of SME internationalization by capturing elements of managerial mindset, operational adaptability, human capital, learning, and institutional engagement that do not appear in financial statements. These variables can be grouped into five integrated capability dimensions.

The first dimension, International Strategic Orientation, reflects firms’ forward-looking commitment to global markets. Expected International Earnings (EIE, NF1) indicate whether anticipated returns justify internationalization costs (Yi & Wang, 2012; Arbelo et al., 2024; Ramon-Jeronimo et al., 2019). Marketing Allocation Abroad (MAA, NF2) supports foreign visibility and positioning (Ishii & Yuki, 2025; Theingi & Purchase, 2011). International Risk Tolerance (IRT, NF3) facilitates early entry, network engagement, and experiential learning (Figueira-de-Lemos et al., 2011; Puthusserry et al., 2020; Vahlne, 2020). Long-term Global Focus (LGF, NF4) captures managerial vision (Mammadov & Wald, 2025; Torkkeli et al., 2018; Nummela et al., 2004), while Offshore Customer Knowledge (OCK, NF5) strengthens the firm’s understanding of foreign demand (Billore & Billore, 2019; Stoian et al., 2017). International Positioning Capability (IPC, NF6) reflects higher-order dynamic capabilities needed to sense, seize, and reconfigure opportunities abroad (Pfajfar et al., 2024).

The second dimension, Operational Readiness, reflects the extent to which SMEs can adapt production and supply-chain systems for cross-border operations. Production Scale-up Capacity (PSC, NF7) supports responsiveness to foreign market volume and variety (Koren et al., 2018; Herrigel, 2015; Ling-yee & Ogunmokun, 2008). Adaptive Supply Chain Capability (ASC, NF8) enhances resilience under structural shifts and disruptions (Tukamuhabwa et al., 2021; Richey et al., 2022; Mishra & Singh, 2022). Product Adaptation Capability (PAC, NF9) enables customization for foreign needs, while Logistics System Capacity (LSC, NF10) ensures end-to-end delivery reliability (Patrucco & Kähkönen, 2021). Together, these variables capture the operational flexibility highlighted by dynamic capabilities theory (Teece, 2023).

The third dimension, Human Capital and Team Readiness, covers leadership, skills, and coordination relevant for international work. Leadership Global Mindset (LGM, NF11) shapes strategic interpretation of global dynamics (Lyles, 1990; Augusto Felício et al., 2015; Miočević & Crnjak-Karanović, 2012). Cross-Cultural Communication Skills (CCS, NF12) facilitate negotiation and collaboration in diverse settings (Sabet & Chapman, 2023; Szkudlarek et al., 2020). International Task Coordination (ITC, NF13) supports cross-functional integration (Luo et al., 2010; Belderbos et al., 2024). Team International Readiness (TIR, NF14) strengthens export preparedness (Tate et al., 2021; Gerschewski et al., 2020), while International Marketing Personnel Access (IMA, NF15) ensures availability of staff with foreign-market knowledge (Kaleka & Morgan, 2019; Enderwick & Akoorie, 1994; Ipsmiller et al., 2021).

The fourth dimension, Market Experience and Knowledge, captures learning-based capabilities that reduce uncertainty in foreign environments. International Market Intelligence Access (IMI, NF16) supports data-driven decisions (İpek & Bıçakcıoğlu-Peynirci, 2020; Ishii & Kikumori, 2024). Trade Fair Participation (TFP, NF17) strengthens networking and signaling (Hansen, 1996; Rolf Seringhaus & Rosson, 1998). Experience Serving Foreign Clients (ESF, NF18) and Export Operation Experience (EOE, NF19) contribute to foreign market familiarity (Martí et al., 2017; Birru et al., 2019). Customs Clearance Experience (CCE, NF20) and Foreign Trade Regulation Knowledge (TRK, NF21) enhance compliance and help firms leverage trade agreements (Martí et al., 2017; Goel et al., 2021; Martín Martín et al., 2022; Mukunoki & Okoshi, 2025).

The fifth dimension, Digitalization and Institutional Connectivity, reflects the ability to leverage digital tools and external networks. Digital International Marketing (DIM, NF22) expands reach and reduces information asymmetries (Da Rocha et al., 2024). E-Commerce Channel Usage (ECU, NF23) facilitates rapid, cost-efficient international entry (Hånell et al., 2019; Vázquez-Martínez et al., 2021). Public and Private Export Assistance Participation (PPA, NF24) enhances export knowledge and commitment (Eun et al., 2011; Shamsuddoha et al., 2009). Finally, International Business Experience (IBE, NF25), Business Association Membership (institutional connectivity) (BAM, NF26), Past International Activity (PIA, NF27), Formal Association Membership (FAM, NF28), and International Network Involvement (INI, NF29) represent relational resources that strengthen opportunity recognition, legitimacy, and resource mobilization (Martín Martín et al., 2022; Yoon et al., 2020).

Collectively, these non-financial factors provide a comprehensive picture of SMEs’ internationalization capacity. They encompass strategic vision, operational flexibility, human and social capital, experiential learning, and digital and institutional embeddedness, forming the core non-financial foundations that complement financial indicators in shaping international or domestic strategic orientation. Table 2 summarizes these non-financial factor groups and their sources.

3. Data and Methodology

3.1. Data Collection and Sample

The empirical research is based on a sample of 179 Thai SMEs, which were surveyed over the period from 2021 to 2023. The database includes both financial and non-financial variables that are important in determining the capabilities of firms in both international and domestic markets. Financial information typically includes performance data such as profitability ratios, liquidity measures, and capital utilization measures. Non-financial information reflects entrepreneurial skills, e.g., international business experience, team preparedness, risk affinity, and institutional connectivity. This summary dataset enables us to compare the organization of processes in firms operating abroad and those active only on the home market from several perspectives. To ensure robustness, the dataset was partitioned into a training set (80%) and a testing set (20%) using random sampling. This stratification ensured that model training and validation were performed on disjoint subsets, thereby minimizing overfitting bias.

3.2. Methodology

The objective of this study is to identify the configurational differences between internationalized and localized SMEs in Thailand by examining how financial and entrepreneurial capabilities interact. Because SME internationalization is shaped by complex, non-linear, and interdependent relationships, a traditional hypothesis-testing approach would be inadequate. Instead, an inductive, machine learning–based method is employed, allowing patterns to emerge from the data without imposing restrictive linear assumptions (Mullainathan & Spiess, 2017; Shmueli, 2010). Extreme Gradient Boosting (XGBoost) is selected as the core analytical tool due to its predictive accuracy, efficiency, and interpretability. As an optimized implementation of gradient boosting, XGBoost builds ensembles of decision trees to minimize a regularized loss function (Chen & Guestrin, 2016). Several features make it well suited for this study.

First, robustness with heterogeneous, high-dimensional data: XGBoost can process continuous variables such as financial ratios alongside categorical or ordinal indicators of entrepreneurial orientation and institutional connectivity with minimal preprocessing—an advantage in SME research that often integrates accounting and survey-based data (Hastie et al., 2009).

Second, ability to model non-linearities and interactions: SME internationalization reflects the interplay of financial agility, risk orientation, team readiness, and institutional ties. XGBoost’s tree-based structure automatically captures non-linear effects and higher-order interactions that conventional regression models may miss (James et al., 2021).

Third, regularization and overfitting prevention: XGBoost incorporates L1 (Lasso) and L2 (Ridge) penalties, shrinkage, subsampling, and column sampling to control variance and improve generalization—important for studies with moderate sample sizes typical of SME datasets (Natekin & Knoll, 2013).

Fourth, interpretability: Despite being a machine learning model, XGBoost provides feature importance metrics—gain, cover, and frequency—that reveal the relative influence of financial and entrepreneurial variables on classification outcomes (Lundberg & Lee, 2017). This ensures the model contributes not only predictive accuracy but also theoretical insight.

Fifth, empirical validation in related fields: Gradient boosting has been successfully applied to corporate default prediction (Lessmann et al., 2015), SME credit scoring (Khandani et al., 2010), and entrepreneurial performance modeling (Cho & Kim, 2020), demonstrating its suitability for financial and organizational research.

In sum, XGBoost is used in this study as an instrumental tool: its capacity to handle non-linear interactions, control overfitting, and generate interpretable feature importance scores makes it well suited for surfacing the latent capability bundles associated with selective integration. The contribution of the paper, however, is not the predictive performance of the model. The conceptual contribution lies in interpreting the patterns the algorithm reveals through a conditional-globalization lens that treats firm-level capabilities as conditioning factors for systemic coupling with wider economic structures (Gupta & Guidi, 2012; Gupta, Haddad, & Selvanathan, 2024; Haddad, 2023; Paramati, Mo, & Gupta, 2017). The algorithm identifies the pattern; the theoretical contribution explains what the pattern means.

3.3. Handling Class Imbalance

A methodological challenge was the imbalance of firm categories: 68.72% were internationalized, 6.70% were potentially internationalizing, and 24.58% were localized. To mitigate this imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was employed to generate synthetic observations for the minority classes. This resumption ensures more stable classification by preventing bias toward the majority class.

3.4. Analytical Approach: Extreme Gradient Boosting (XGBoost)

Key analytic instrument used is Extreme Gradient Boosing (XGB), a machine learning algorithm that is proven to yield great performance in classification tasks dealing with high-dimensional and mixed homogeneous and heterogeneous data. We selected XGBoost because it can handle missing values, prevents overfitting, and also provides human-interpretable estimates of feature importances. In the research, this has led to the development of 47 financial and non-financial factors. FR indicators, such as ROA, ROE, CRR, DTE, and TAT were also added. Beyond financial characteristics, strategic features, namely international experience, readiness of operational team, risk preference toward international investment and institutional linkage were included. XGBoost is a scalable tree-boosting algorithm that constructs an additive ensemble of regression trees. Formally, let the training dataset be

$$D={\left\{\left(x_i,y_i\right)\right\}}_{i=1}^n,x_iϵR^m,y_iϵ\left\{\mathrm{0,1}\right\},$$

where $x_i$ denotes the feature vector of SME $i,y_i$ is the binary class label (1 = internationalized, 0 = localized), $n$ is the sample size and $m$ is the number of features. XGBoost builds an additive ensemble of classification trees:

$${{\mathsf{ŷ}}_{\mathrm{i}}}^{\left(\mathrm{t}\right)}=\sum _{k=1}^{t}fₖ\left(xᵢ\right),fₖ\in F$$

where $F$ is the space of regression trees.

The objective function at iteration $t$ is:

$$L^{\left(\mathrm{t}\right)}={{\Sigma }_{\mathrm{i}=1}}^{\mathrm{n}}l(y_{\mathrm{i}},{{\mathrm{ŷ}}_{\mathrm{i}}}^{(\mathrm{t}-1)}+fₜ(xᵢ\left)\right)+\Omega (fₜ)$$

where l is the differentiable convex loss (logistic loss for binary classification), and Ω(ft) regularizes tree complexity:

$$\Omega \left(f\right)=\gamma T+½\lambda {\Vert w\Vert }^2,$$

with T as the number of leaves, w as the leaf weights, and γ, λ as regularization parameters. To optimize efficiently, XGBoost applies a second-order Taylor expansion of the loss:

$$L^{\left(\mathrm{t}\right)}={{\Sigma }_{\mathrm{i}}}^{=1\mathrm{n}}\left[gᵢfₜ\left(xᵢ\right)+½hᵢfₜ{\left(xᵢ\right)}^2\right]+\Omega \left(fₜ\right)$$

where

$$gᵢ={\displaystyle \frac{\partial l\left(yᵢ,{{\mathrm{ŷ}}_{\mathrm{i}}}^{(\mathrm{t}-1)}\right)}{\partial {{\mathrm{ŷ}}_{\mathrm{i}}}^{(\mathrm{t}-1)}}},hᵢ={\displaystyle \frac{\partial ²l\left(yᵢ,{{\mathrm{ŷ}}_{\mathrm{i}}}^{(\mathrm{t}-1)}\right)}{\partial {\left({{\mathrm{ŷ}}_{\mathrm{i}}}^{(\mathrm{t}-1)}\right)}^2}},$$

This formulation allows for efficient computation of optimal splits at each node, making XGBoost scalable and accurate.

The predictive ability of the model was tested by a number of normal metrics which are calculated based on the concept of a confusion matrix. These statistics include accuracy, precision (or the fraction of positive identifications that were correct), recall (which is the proportion of actual positives that are correctly identified), and F1-score (which is the harmonic mean of precision and recall). Model performance was evaluated using standard classification metrics:

$$Accuracy={\displaystyle \frac{(TP+TN)}{(TP+TN+FP+FN)}}$$

$$Precision={\displaystyle \frac{TP}{(TP+FP)}},Recall={\displaystyle \frac{TP}{(TP+FN)}}$$

$$F1=2.{\displaystyle \frac{\left(Precision\times Recall\right)}{\left(Precision+Recall\right)}}$$

where TP, TN, FP, and FN represent true positives, true negatives, false positives, and false negatives, respectively.

The XGBoost classifier in this study achieved a test accuracy of 92.59%, with precision = 0.95 and recall = 0.86, demonstrating strong predictive ability in distinguishing internationalized from localized SMEs.

To interpret the model, gain-based feature importance scores from XGBoost were used to rank predictors. Financial indicators such as ROA, ROE, DTE Ratio, Total Asset Turnover, and CRR emerged as dominant. Non-financial features, including international experience, team readiness, and association membership, were also highly influential.

Subsequently, confirmatory statistical analyses (Welch’s t-test and Mann–Whitney U-test) were applied to verify significant differences in financial ratios between the two groups, ensuring robustness against heteroskedasticity and non-normality.

4. Results and Discussions

The use of the Extreme Gradient Boosting (XGBoost) machine learning approach enabled the systematic analysis of financial and operational attributes that characterize the internationalization process of Thai SMEs. The comparison shows sharp dichotomies in these two groups, as evident in the patterns of essential financial ratios and entrepreneurial performance measures.

4.1. Results

• Financial and Operational Efficiency

The analysis of the financial statement revealed that internationalized SMEs were significantly favored over their local counterparts in terms of financial health and operational performance. As shown in Figure 1, internationalized companies had significantly higher ROA, ROE, and GPM values. Operational efficiency metrics, such as INT and ART, also exhibited better performance in internationalized SMEs, although some operational metrics reflected less pronounced differences. Furthermore, as indicated by Figure 1, the internationalized SMEs cope with their financial risks better than the non-internationalized ones, as evidenced by relatively lower levels of DTE and CRT. On the other side, domestic SMEs would have more debt based on the higher values of DTE and DTA ratios, and are less cost-efficient.

The internationalized firms are thus shown to be more profitable and strategically consistent in managing their assets and risks, with more local firms adopting a conservative though for less efficient financial posture. Additional statistical analyses support these findings. As presented in Table 2, fifteen financial standard ratios were used for the enforcement of the significance of financial indicators’ differences of internationalized and localized SMEs, where the pairs of groups were compared by the Welch’s (t )-test and Mann-Whitney (U )-test. Nine ratios differed significantly or nearly significantly among the financial indicators. Differences across the groups were particularly marked for the CRR, where locally-owned SMEs held significantly greater liquidity (t-test p = 0.0007; Mann-Whitney p < 0.000001), an indication of cautious cash management, but possibly ineffective use of working capital.

Both the EM and the DTE were significantly different between the internationalized and the non-internationalized for slightly over p < 0.001, showing a stronger intention and capability of using strategic leverage for growth. ROA, ROE, GPM, OPM, and NPM were also in favor of internationalized firms, and they also showed many strong and statistically significant (p < 0.05) indicators. ART, INT, and APT were not statistically significant; however, TAT was significantly higher for internationalized SMEs (p < 0.0005), indicating better usage of resources. All in all, the financial and operational characteristics presented in Figure 1 and in Table 2 emphasize that internationalized SMEs are structurally better prepared for international competition. They shown better returns, higher asset utilization, and a more aggressive -But efficient- financial risk management against those of their local peers.

Figure 1. Financial Performance Metrics Comparison Between Internationalized and Localized Thai SMEs.

Figure 1. Financial Performance Metrics Comparison Between Internationalized and Localized Thai SMEs.

• Feature Importance in Entrepreneurial and Financial Classification

The feature importance analysis of the XGBoost model illuminated the variables driving the classification of SMEs by internationalization status. As indicated in Table 3, entrepreneurial skills became the dominant predictors where International Business Experience (NF25) was assessed the most important variable. This discovery further supports the importance of accumulated internationalization experience in supporting SME success overseas.

Other non-financial predictors such as team preparedness (NF14) and Business Association Membership (institutional connectivity) (NF26) were also strong predictors, indicating the importance of internal human resource capacity and external institutional networking for engaging with international markets. Other key entrepreneurial traits, such as risk attitude (NF3) and global strategic orientation (NF11), emphasize the importance of an active management orientation in mitigating the obstacles of internationalization.

The financial factors saw a similar effect. Leverage strategies utilizing either DTA or the EM, as well as total asset turnover and profitability ratios, such as ROA and ROE, were main financial predictors. Effective management of liquidity along with the efficiency of receivables also available in the predictive model but in a slightly less than other.

The combined analysis in Table 3 indicates that SME internationalization performance does not stem from isolated capabilities, but rather, a configuration of financial flexibility, entrepreneurial preparedness, and institutional embeddedness. Developed international SMEs not only perform better in financial terms, but they have non-financial capabilities that are essential to mitigate or adapt to the associated risks and challenges of entering and competing in overseas markets.

In contrast, the regional SMEs seem to be trapped with lesser entrepreneurial skills and with more conservative financial behaviors that do not allow them to enter on or sustain internationalization. These results offer strong empirical evidence of the multidimensional nature of the process of SME internationalization and the necessity for simultaneous development of financial, strategic and relational capabilities to achieve success in a global context.

The combined financial and non-financial predictors investigation with the XGBoost classification provides important information on the distinctive features of internationalized versus localized Thai. The model maintained good predictive power at key entrepreneurial and financial variables. NF25 was the most significant factor among non-financial predictors. This variable (FFI) as a sign of the firm’s previous exposure to international business confirms the other findings of the internationalization literature that international exposure would enable CCN to entangle inside foreign markets. Similarly, global indicators like NF14 (team readiness) and NF26 (association membership) highlight the role of internal human capital and external institutional connections in enabling the internationalization process. Other entrepreneurial dimensions, such as NF3 (risk orientation) and NF11 (strategic orientation), also indicate that a management orientation and capacity for development are important in influencing international behavior.

From the financial ratios, two leverage measures (DTA and EM) appeared to be the strongest predictor. This finding suggests that internationalized firms are more prone to strategic financial risk-taking, increasing the use of debt to finance their growth. Furthermore, the operating capability through the variable TAT and profitability (ROA and ROE), at large have shown courteous results in favor of internationalized SMEs, ensuring that in term of performance and resource utilization, SMEs depend directly and closely in the global competition.

Altogether these findings offer strong empirical validation for a multidimensional model of SME internationalization. The complementary relationship between entrepreneurial readiness and financial strategy seems to be that firms which show both readiness and capability are more likely to succeed in international operations. This highlights the need for policy that offers more than just funding support, and develops international experience, managerial capability and long-term strategy for locally based SMEs.

To summarize, the composite financial and entrepreneurial analysis demonstrates that internationalized SMEs exhibit a distinct configuration of capabilities combining higher levels of financial flexibility, a strategic management of operations, entrepreneurial preparedness and institutional embeddedness. These firms exploit profitability, efficiency in operations, risk-taking propensity, and external networks to maintain competitive advantages in overseas markets. On the other hand, local SMEs, although having an higher liquidity cushion and a more conservative financial posture, seem to have a weaker profitability, less strategic readiness and less connection with the external world. This differentiation in ability highlights the fact that what distinguishes successful internationalization among the SMEs is not their financial capacity or entrepreneurial qualities (all things being equal); it is how they combine these in a coherent whole. Identifying and plugging these multi-faceted capability gaps is key to helping the international expansion goals of local firms.

Consequently, the findings show that internationalized SMEs exhibit a distinctive bundle of financial agility, disciplined leverage, asset-utilization efficiency, international experience, team readiness, and institutional embeddedness. We interpret this not as evidence of having 'more' resources, but as evidence of deeper systemic coupling with cross-border information, relational, and capital networks. Localized SMEs, by contrast, exhibit liquidity-heavy and more conservative profiles consistent with a buffered.

5. Discussion

This study provides new empirical evidence that SME internationalization is a multidimensional capability phenomenon, shaped by the interplay of financial agility, entrepreneurial readiness, and institutional embeddedness. By combining inductive machine learning techniques with theoretical frameworks, the findings extend prior international business and finance scholarships in several directions.

• Financial Capabilities and Strategic Agility

Our results demonstrate that internationalized SMEs consistently outperform localized firms in terms of profitability (ROA, ROE, GPM), leverage management, and resource efficiency (TAT). This confirms prior evidence that profitability and financial strength facilitate internationalization by lowering financing constraints and absorbing risk (Lu & Beamish, 2006; Contractor et al., 2005). From the RBV, these financial indicators reflect effective resource deployment, yet our findings go further by showing that financial agility, strategic use of leverage, balanced liquidity, and efficient cost control, operates as a dynamic capability enabling adaptation to uncertain international markets (Teece, 2023; Sapienza et al., 2006).

The findings also reconcile tensions in POT and AGT. While localized SMEs exhibit liquidity hoarding consistent with pecking order reasoning, internationalized firms deploy leverage strategically, consistent with the disciplining and growth-enabling effects of debt (Campello et al., 2010). This suggests that financial constraints hinder international expansion (Fauceglia, 2015; Manova, 2012), yet financial flexibility, conceptualized as a proactive capability, enables SMEs to overcome liabilities of foreignness and exploit global opportunities (Cumming et al., 2014).

• Entrepreneurial Orientation and Organizational Preparedness

The feature importance analysis revealed that international experience, team readiness, and risk-taking are among the most influential drivers of SME internationalization. This resonates with the entrepreneurial orientation framework, which identifies innovativeness, proactiveness, and risk-taking as central to cross-border expansion. The critical role of managerial readiness and absorptive capacity reinforces the OLT, indicating that SMEs with teams capable of integrating external knowledge are more likely to internationalize successfully.

Importantly, our findings show that entrepreneurial orientation is not sufficient in isolation but interacts synergistically with financial agility. This supports configurational perspectives in strategic management (Fiss, 2011), emphasizing that international success arises from complementary bundles of capabilities rather than isolated traits. SMEs that combine entrepreneurial proactiveness with financial resourcefulness are more likely to overcome internationalization barriers and exploit foreign opportunities (Autio, 2017; Prashantham & Floyd, 2019).

• Institutional Connectivity and Relational Capital

Institutional memberships, international networks, and trade-fair participation emerged as strong non-financial predictors, reinforcing the importance of network theory of internationalization Johanson & Vahlne, 1977). Relational capital mitigates information asymmetries, provides legitimacy in foreign markets, and enables SMEs to mobilize external resources. Evidence from emerging markets suggests that institutional connectivity is particularly critical where formal market-supporting institutions are weaker (Jackson, 2012). This finding also aligns with research on born globals and accelerated internationalizers, which shows that networks enable firms to bypass gradual stages of internationalization. For Thai SMEs, participation in associations and international partnerships not only compensates for resource limitations but also amplifies learning effects, enhancing export intensity and resilience.

6. Conclusions

6.1. Key Findings

This study examined the capability configurations that differentiate internationalized from localized SMEs in Thailand using an inductive, machine learning–based approach. By integrating financial indicators with entrepreneurial and institutional variables, the analysis provides a multidimensional understanding of SME internationalization. Three findings stand out.

First, financial agility is a defining capability of internationalized SMEs. These firms combine higher profitability, disciplined leverage, and efficient resource utilization with the ability to reconfigure financial structures to support expansion. Localized SMEs, in contrast, maintain liquidity-heavy, risk-averse profiles that restrict growth beyond domestic markets.

Second, entrepreneurial readiness and organizational preparedness are critical. International experience, risk tolerance, and globally oriented leadership significantly predict cross-border expansion. These capabilities help SMEs leverage knowledge-based assets, identify opportunities, and mitigate liabilities of foreignness.

Third, institutional connectivity and relational capital enable SMEs to overcome structural barriers common in emerging markets. Internationalized firms benefit from association membership, export-support programs, and cross-border networks that provide knowledge, legitimacy, and resource access.

In conclusion, internationalized Thai SMEs are best understood not as a different type of firm, but as the same population operating with a more tightly integrated bundle of capabilities. Financial agility, entrepreneurial readiness, and institutional connectivity are not three independent traits but mutually reinforcing: relational capital opens information channels that lower the perceived risk of expansion; entrepreneurial orientation translates that information into commitments; and disciplined financial structures supply the slack that turns commitments into action. Localized SMEs, by contrast, do not lack any single capability in isolation — they exhibit a configuration in which liquidity-preserving financial postures, limited international exposure of decision-makers, and thinner cross-border ties co-occur and reinforce one another, producing a stable but internally bounded equilibrium. This configurational reading reframes "internationalization" in the Thai SME context as a question of capability coupling rather than capability possession, with direct implications for policy: interventions that target only one dimension — financing alone, training alone, or matchmaking alone — are unlikely to shift firms across the boundary the data reveal.

6.2. Theoretical Contributions

This study contributes to international business and finance research in four ways.

First, it advances the resource-based view (RBV) and the dynamic capabilities view (DCV) by reframing financial agility not as a static endowment of valuable, rare, and inimitable assets, but as a dynamic capability in its own right. Internationalized SMEs do not merely hold more profitable, less leveraged, or more efficiently deployed financial positions; they are observably more capable of reconfiguring those positions to absorb shocks and fund expansion under volatility. This repositions financial management — typically treated as a back-office function — as a source of international competitiveness, and complements the largely operations- and innovation-centred treatment of dynamic capabilities in the existing literature.

Second, the study extends entrepreneurial orientation (EO) research. The dominant EO literature treats innovativeness, proactiveness, and risk-taking as quasi-independent drivers of international expansion. Our findings point to a more configurational reading: entrepreneurial readiness predicts internationalization most strongly in combination with financial agility, rather than on its own. This is consistent with configurational and equifinality arguments in strategic management, which hold that competitive outcomes arise from coherent bundles of practices rather than from any single dimension.

Third, the study reinforces and qualifies network theory in emerging markets. Internationalized SMEs in our sample depend heavily on alliances, association memberships, and relational capital — consistent with the network-internationalization argument that ties supply the knowledge, legitimacy, and resource access required to overcome the liability of outsidership (Johanson & Vahlne, 2009). In weak institutional environments, such ties also act as functional substitutes for missing formal mechanisms (Jackson, 2012; Meyer & Peng, 2015). Taken together, these three contributions converge on an integrated framework in which financial, entrepreneurial, and relational capabilities operate as complementary bundles whose joint configuration — not their individual levels — shapes SME international expansion.

Fourth, and most relevant to the present special issue, the firm-level findings provide a candidate micro-foundation for the asymmetric, time-varying integration patterns documented in the financial globalization literature, in which national markets exhibit selective rather than uniform co-movement with global systems (Gupta & Guidi, 2012; Haddad, 2023; Gupta, Haddad, & Selvanathan, 2024). One source of this macro asymmetry may lie below the country level: within a single emerging market, firms differ systematically in the capability bundles through which cross-border information, capital, and relational flows are channelled. Internationalized SMEs can thus be read as firm-level carriers of integration, while localized SMEs occupy partially buffered positions arising from capability gaps, structural exclusion, or rational equilibria under uncertainty. The argument is interpretive rather than econometric: SMEs are not claimed to co-move in the technical sense, but aggregate measures of integration (e.g., Paramati, Mo, & Gupta, 2017) likely mask substantial firm-level heterogeneity in how that integration is realized. Within this framing, the XGBoost model functions as an inductive instrument for surfacing non-linear capability bundles, and the conditional-globalization lens supplies their interpretive context. The paper's contribution is therefore the identification of firm-level conditions associated with selective participation in globalized systems, rather than a binary classification of firms.

6.3. Practical and Policy Implications

For SME managers, the evidence emphasizes the need to balance financial discipline with entrepreneurial risk-taking. Solely prioritizing liquidity may ensure short-term security but hinders long-term competitiveness. Managers should cultivate financial agility, invest in innovation, and build globally competent teams capable of absorbing international knowledge.

For policymakers, the findings highlight that financial support alone is insufficient. Effective interventions must integrate financial assistance with capability development—improving managerial readiness, enhancing institutional connectivity, and facilitating network participation. Export-promotion agencies and industry associations should adopt holistic support models that strengthen financial literacy, global orientation, and relational capital simultaneously.

For investors and financial institutions, the results suggest that international potential is not reflected solely in financial metrics. Firms with strong entrepreneurial orientation and institutional embeddedness may offer more resilient and growth-oriented prospects than those judged solely on balance-sheet indicators.

6.4. Limitations and Avenues for Future Research

This study has several limitations that suggest opportunities for further work. First, although the dataset is extensive, the findings reflect the Thai context and may not fully generalize to other emerging economies. Replicating this configurational approach in Southeast Asia, Latin America, or Africa would help test boundary conditions.

Second, the analysis relies on financial ratios and survey-based indicators, which do not capture broader intangible capabilities such as innovation strength, digital readiness, or ESG practices. Future research should incorporate these variables to provide a more comprehensive view.

Third, while machine learning reveals patterns and interactions, it does not establish causality. A mixed-method or hybrid approach—combining machine learning with structural equation modeling or qualitative comparative analysis—would deepen theoretical interpretation.

Fourth, cross-sectional data limit the study’s ability to assess capability evolution. Longitudinal research could trace how SMEs reconfigure capabilities in response to market shocks, technological change, or international learning.

Finally, future research could explore how specific policy instruments interact with firm-level capabilities to influence internationalization outcomes, offering more actionable insights for emerging-market governments.

6.5. Concluding Remarks

This study shows that SME internationalization in emerging markets is best understood through a capability-based, configurational lens. Financial agility, entrepreneurial readiness, and institutional connectivity function not as isolated drivers but as mutually reinforcing elements enabling firms to overcome constraints and pursue opportunities abroad. By applying advanced machine learning methods, the study highlights the complex, non-linear interactions underlying these capabilities and contributes to methodological innovation in international business research.

Positioned within RBV, Dynamic Capabilities, Entrepreneurial Orientation, and Network Theory, the findings offer a more integrated understanding of how SMEs navigate globalization. For practitioners and policymakers, the results underscore the need for strategies and interventions that strengthen the complementarities among financial, entrepreneurial, and relational resources. Ultimately, the study demonstrates that successful SME internationalization arises not from financial strength alone or entrepreneurial ambition alone, but from the orchestrated interplay of resources, adaptive capabilities, and institutional ties. This multidimensional configuration provides a pathway for localized firms to overcome barriers, compete internationally, and contribute to economic development in emerging markets.