0 cumulative citations
View corpus contextFirms that intensively deploy AI report markedly stronger cross-border e-commerce performance—AI adoption intensity predicts outcomes more strongly than firm size, with automated translation boosting conversions and customs automation lowering logistics costs. Yet high implementation costs, talent gaps and data-privacy rules are significant barriers, especially for SMEs.
Artificial Intelligence And Its Influence On Cross-Border E-Commerce Growth
openalex
correlational
medium evidence
7/10 relevance
Full text
usable extracted full text
DOI
Source
PDF
Survey and interview evidence indicates that greater AI adoption intensity is strongly associated with higher cross-border e-commerce performance—outperforming company size as a predictor—with automated translation improving conversions most and customs automation cutting logistics costs, though high costs, talent shortages, and regulatory complexity constrain diffusion.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
The rapid advancement of Artificial Intelligence (AI) technologies has fundamentally reshaped global commerce, with cross-border e-commerce (CBEC) emerging as a domain of exceptional transformation. This study employs a convergent mixed-methods design—integrating a structured questionnaire administered to 150 e-commerce professionals with 15 semi-structured managerial interviews—to provide an empirically grounded assessment of AI\\\'s multidimensional influence on CBEC performance. Multiple regression analysis (R² = 0.612; F = 44.23; p < 0.001) identifies AI adoption intensity as the single strongest predictor of cross-border performance, surpassing even company size (β = 0.487 vs. β = 0.213). Application-level findings reveal that automated translation/localization delivers the highest conversion rate improvement (64.3%), while customs automation yields the most substantial logistics cost reduction (31.4%). Despite these gains, high implementation costs (67.3%), talent deficits (62.0%), and data-privacy regulatory complexity (58.7%) constitute the principal barriers inhibiting broader diffusion—particularly among SMEs. Theoretically, the study extends the Resource-Based View and Technology Acceptance Model to the cross-border digital-trade context, and it offers a strategic AI adoption framework for practitioners and actionable recommendations for policymakers. The findings collectively affirm that AI is not merely an incremental operational enhancer but a foundational competitive capability reshaping the structure of international digital trade.
Summary
Main Finding
AI adoption intensity is the single strongest predictor of cross-border e‑commerce (CBEC) performance in the authors' sample—outperforming firm size—and specific AI applications (notably automated translation/localization and customs automation) deliver large, measurable operational gains. However, high implementation costs, talent shortages, and regulatory/data‑privacy complexity materially constrain diffusion, especially among SMEs.
Key Points
- Core quantitative result: multiple regression: R² = 0.612; F = 44.23; p < 0.001. AI adoption intensity β = 0.487 versus company size β = 0.213.
- Application-level impacts reported:
- Automated translation/localization → highest conversion-rate improvement: 64.3%.
- Customs automation → largest logistics cost reduction: 31.4%.
- Adoption prevalence (n = 150):
- Product recommendation engines 82.0% (satisfaction mean 4.2/5)
- Chatbots/virtual assistants 74.7% (3.9)
- Automated translation/localization 68.0% (4.1)
- Fraud detection systems 66.7% (4.4)
- Predictive demand forecasting 54.7% (3.8)
- Dynamic pricing algorithms 48.0% (3.7)
- AI logistics optimization 44.0% (4.0)
- Customs automation 32.0% (3.6)
- Computer vision for imagery 28.7% (3.5)
- Principal barriers to wider AI diffusion:
- High implementation costs: 67.3%
- Talent deficits: 62.0%
- Data‑privacy/regulatory complexity: 58.7%
- Distributional pattern: strong size-adoption gradient — large firms have much higher rates of high AI adoption; SMEs (majority of sample) show substantial under-adoption.
- Theoretical contribution: extends Resource‑Based View (RBV) and Technology Acceptance Model (TAM), and links AI to the Uppsala model of internationalization—treating AI as a strategic internationalization resource that reduces liability of outsidership.
Data & Methods
- Design: convergent parallel mixed‑methods (quantitative survey + qualitative interviews), cross‑sectional (data collected Jan–Mar 2025).
- Sample:
- Survey: 150 CBEC professionals (68% India, 22% other Asia‑Pacific, 10% other markets); industry mix and firm‑size spread with SMEs = 52% of sample.
- Interviews: 15 semi‑structured managerial interviews (avg. 52 minutes).
- Instrument validity & reliability:
- Questionnaire: 45 items, 5‑point Likert scales; content-validated by experts and pilot-tested.
- Cronbach’s α: AI adoption scale = 0.84; perceived impact scale = 0.79.
- CFA: CFI = 0.92; RMSEA = 0.06.
- Inter‑rater coding agreement for qualitative themes: Cohen’s κ = 0.81.
- Quantitative analysis:
- Dependent variable: composite cross‑border performance index (standardized sales growth, conversion improvement, CAC reduction, logistics cost reduction).
- Key independent variable: AI adoption intensity (number of tools, integration depth, duration).
- Analytical tools: SPSS v26; checks for multicollinearity (VIF < 3), normality, homoscedasticity, outliers.
- Qualitative analysis: thematic analysis (Braun & Clarke) using NVivo; member checking performed.
Implications for AI Economics
- Market structure and competition:
- AI functions as a general‑purpose technology in CBEC that can materially lower entry frictions (language, compliance, logistics), enabling SME internationalization but also amplifying advantages for firms that can invest early and deeply in AI (potentially increasing concentration where adoption is asymmetric).
- Productivity and trade effects:
- Large reported gains (conversion +64.3% for localization; ~31.4% logistics cost savings for customs automation) imply substantive micro‑level productivity improvements that can scale into larger cross‑border trade flows if adoption widens.
- Distributional concerns and policy levers:
- Barriers (cost, skills, regulatory complexity) suggest market failures where SMEs underinvest despite positive returns. Policy responses include subsidized access to AI-as-a-service, targeted training programs, public‑private shared infrastructure (e.g., customs automation platforms), and streamlined cross‑border data‑governance frameworks to lower compliance costs.
- Regulation and governance:
- High effectiveness of fraud detection and dynamic pricing raise regulatory tradeoffs—consumer protection and anti‑collusion oversight will be important as AI reconfigures pricing, fraud risk, and platform behavior in international markets.
- Research and evaluation needs:
- The cross‑sectional design limits causal claims; longitudinal and experimental work is needed to quantify causal impacts of AI on trade volumes, prices, and firm survival. Broader multi‑country samples would clarify heterogeneity by regulatory regime and platform ecosystem.
- Practical takeaway for economists studying digital trade:
- Incorporate AI adoption intensity and application‑level heterogeneity into models of firm export participation, trade costs, and platform competition; treat AI capabilities as firm‑level intangible capital that can shift comparative advantage in digital goods and goods sold via digital channels.
Limitations noted by the authors: cross‑sectional sample skewed to India/Asia‑Pacific, purposive sampling, and self‑reported firm outcomes. Future work should expand geographic coverage, use administrative outcomes, and apply longitudinal designs to strengthen causal inference.
Assessment
Paper Typecorrelational
Evidence Strengthmedium — Findings are supported by quantitative associations with a relatively high R² and significant coefficients and are triangulated by managerial interviews, which increases confidence in the patterns observed; however, reliance on cross-sectional, self-reported survey data, modest sample size, and absence of credible exogenous identification limit causal claims and raise concerns about endogeneity and reporting bias.
Methods Rigormedium — Use of a convergent mixed-methods design and multivariate regression indicates reasonable methodological care, but key rigor shortcomings include likely convenience sampling or limited representativeness, potential common-method and measurement biases, lack of longitudinal or experimental design, and no reported robustness checks or strategies to address endogeneity.
SampleStructured questionnaire responses from 150 e-commerce professionals (firm representatives) plus 15 semi-structured interviews with managers; sample details such as country coverage, industry breakdown, sampling frame, and exact firm-size distribution are not reported in the summary, though SMEs are explicitly discussed as constrained by barriers.
Themesadoption productivity skills_training
IdentificationCross-sectional survey of firms (n=150) combined with 15 semi-structured managerial interviews; causal claims are based on OLS multiple regression (AI adoption intensity and covariates predicting self-reported cross-border performance) and qualitative triangulation—no exogenous variation, instrumental variables, panel / pre-post design, or randomized assignment to support causal inference.
GeneralizabilityModest sample size (n=150) limits statistical power and precision for subgroup analysis, Unknown sampling frame and likely non-probability sampling reduce representativeness across countries and sectors, Self-reported performance and improvement metrics (e.g., conversion gains, cost reductions) may be biased or measured inconsistently, Cross-sectional design prevents inference about dynamics or long-run causal effects, Findings may not generalize beyond the specific CBEC contexts, firm types, or regions sampled (geographic coverage unspecified)
Claims (9)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| The study used a convergent mixed-methods design integrating a structured questionnaire administered to 150 e-commerce professionals with 15 semi-structured managerial interviews. Other | positive | study_design / data_collection |
Reading fidelity
high
Study strength
high
|
n=165
|
| A multiple regression model predicting cross-border performance has R² = 0.612, F = 44.23, p < 0.001. Firm Productivity | positive | cross-border performance (variance explained by model) |
Reading fidelity
high
Study strength
medium
|
n=150
R² = 0.612; F = 44.23; p < 0.001
|
| AI adoption intensity is the single strongest predictor of cross-border performance, surpassing company size (β = 0.487 vs. β = 0.213). Firm Productivity | positive | cross-border performance (predictor strength) |
Reading fidelity
high
Study strength
medium
|
n=150
β = 0.487 vs. β = 0.213
|
| Automated translation/localization delivers the highest conversion rate improvement (64.3%). Task Completion Time | positive | conversion rate |
Reading fidelity
high
Study strength
medium
|
n=150
64.3%
|
| Customs automation yields the most substantial logistics cost reduction (31.4%). Firm Productivity | positive | logistics costs |
Reading fidelity
high
Study strength
medium
|
n=150
31.4% cost reduction
|
| High implementation costs (67.3%), talent deficits (62.0%), and data-privacy regulatory complexity (58.7%) are the principal barriers inhibiting broader AI diffusion—particularly among SMEs. Adoption Rate | negative | perceived barriers to AI adoption |
Reading fidelity
high
Study strength
medium
|
n=150
67.3%; 62.0%; 58.7%
|
| SMEs are disproportionately inhibited from broader AI diffusion due to the listed barriers. Adoption Rate | negative | AI adoption rate / diffusion among SMEs |
Reading fidelity
medium
Study strength
medium
|
n=150
|
| The study extends the Resource-Based View (RBV) and Technology Acceptance Model (TAM) to the cross-border digital-trade context. Governance And Regulation | positive | theoretical framework extension |
Reading fidelity
high
Study strength
speculative
|
not reported
|
| AI is not merely an incremental operational enhancer but a foundational competitive capability reshaping the structure of international digital trade. Market Structure | positive | strategic competitive capability / structure of international digital trade |
Reading fidelity
high
Study strength
medium
|
n=165
|