The role of digital intermediation platforms in opening up international markets

Unpacking the relationship between restrictions to digitally enabled services trade and DIPs’ diffusion

This section presents the findings of a cross-country panel analysis examining the association between regulatory measures to digitally enabled services trade – captured through the five policy areas of the DSTRI – and the diffusion of DIPs across three key sectors.1 Covering OECD Member countries from 2010 to 2024, the analysis assesses both the overall presence of platforms and the footprint of foreign firms by analysing total traffic, foreign firm traffic,2 and the foreign share of traffic.

Regulatory measures affecting digitally enabled services trade may influence DIPs diffusion by shaping the costs and complexity of operating across borders. Requirements related to cross-border data flows, payment systems, licensing, or local presence can increase compliance burdens and affect market entry decisions, particularly for foreign providers (Ferracane and van der Marel, 2020[21]). Conversely, more open and interoperable regulatory frameworks can facilitate platform expansion by lowering operational frictions and enabling platforms to scale across markets.

To facilitate interpretation, the results are illustrated with bar plots showing marginal associations derived from regression estimates. The values in these charts represent estimated coefficients from a log-linear regression model, which are not directly interpretable as percentage changes. To provide a meaningful benchmark, a 0.01-point change in the policy score – reflecting a small domestic regulatory reform – is used to translate the coefficients into percentage changes.3 Further methodological details, the analytical framework, and full results can be found in Annex A and Annex C respectively.

Before detailing the key findings, it is important to highlight that the results should be interpreted as correlations, rather than causal relationships. The link between regulation and platform expansion is inherently complex. Many platforms operate across multiple sectors and evolve their business models to navigate or adapt to regulatory frameworks originally designed for traditional services (Calvino, Criscuolo and Menon, 2016[22]). In some cases, regulation itself may be reactive – introduced in response to platform growth – raising the risk of reverse causality (Costa et al., 2021[5]).4 While the following discussion focuses exclusively on observed statistical associations, further explanation of sector-specific dynamics, platform heterogeneity, and other mechanisms influencing digital platform diffusion is provided in endnote 9.

Furthermore, while the DSTRI provides valuable cross-country insights into regulatory measures of digitally enabled services trade, it does not capture sector-specific regulatory nuances relevant to the platforms studied.5 Similarly, while web traffic is a useful proxy for platform diffusion, it does not account for all dimensions of platform activity. In addition, the country-level indicators constructed for this analysis also cannot fully reflect the diversity of platform characteristics and business models, which influence how firms respond to different regulatory environments.

The results indicate that a 0.01-point decrease in restrictions on IPR protection and enforcement is associated with an 8% increase in overall DIP diffusion in the travel and accommodation booking sector (Figure 6). The effect is even more pronounced for foreign-headquartered firms, where such a moderate regulatory reform corresponds to nearly a 25% increase in platform traffic, all else being equal.

Figure 6. Greater restrictions on IPR protection and enforcement are associated with lower platform diffusion in the travel and accommodation sector

Coefficients of association between DSTRI policy areas and DIP diffusion indicators, 2014-2024

Note: Please refer to endnote.6

Sources: Author’s elaboration based on data from Semrush and Crunchbase.

These findings suggest that strong, non-discriminatory IPR enforcement can strongly support DIPs diffusion. A predictable IPR regime not only protects intangible assets but also reduces legal uncertainty, facilitating the expansion of foreign digital platforms. This is particularly relevant in the travel and tourism sector, where competitive advantage depends on brand recognition, user-generated content, and proprietary booking algorithms.

Similarly, the diffusion of DIPs is especially sensitive to restrictions on IPR protection and enforcement within the ride-hailing and carpooling sector (Figure 7). The impact of IPR-related regulatory measures on foreign platforms is even more striking in this sector where a moderate policy reform would be associated with a 90% increase in traffic for foreign-headquartered ride-hailing and carpooling platforms. Given their reliance on proprietary software, real-time algorithms, and consumer data, this finding underscores the importance of a predictable IPR protection and enforcement environment for enabling cross-border platform activity and growth.

Figure 7. Restrictions on IPR protection and enforcement are associated with lower platform diffusion in the ride-hailing and carpooling sector

Coefficients of association between DSTRI policy areas and DIP diffusion indicators, 2014-2024

Notes: Please refer to endnote.¹¹

Sources: Author’s elaboration based on data from Semrush and Crunchbase.

Figure 8 shows that in the X2C online marketplace sector restrictions to payment systems emerge as a key barrier for foreign pure DIPs (e.g., eBay). A moderate policy reform that reduces payment restrictions by 0.01 is associated with a 200% increase in traffic to these platforms, all else being equal. The foreign share of traffic also increases by 4%, implying that such restrictions hinder both market presence and competition.

These findings highlight the key role of secure, interoperable payment systems in facilitating cross-border X2C online marketplace activity. Restrictions such as inconsistent security requirements or regulatory measures to non-resident payment processing impede the scalability of foreign platforms and their ability to provide seamless payment solutions.

Broader restrictions to trade in digitally enabled services – such as local presence requirements, limits on digital advertising, and anti-competitive practices – are also associated with reduced traffic and market share for foreign platforms. A reduction of 0.01 points would lead to a 37% increase in traffic from foreign firms and a rise in their market share ranging from 1% for pure and hybrid DIPs to 2% for strictly pure DIPs. Notably, these restrictions seem to have a much stronger effect on pure DIPs than on hybrid ones.

This divergence may reflect structural differences in business models discussed earlier. Pure DIPs typically scale through digital-only channels and avoid heavy physical infrastructure, making them more exposed to local presence rules. Hybrid platforms, in contrast, often invest in local warehouses or logistics networks, enabling greater compliance with domestic regulations. Consequently, stringent local presence requirements can represent a higher hurdle for foreign pure DIPs, potentially forcing localisation or market exit.

Figure 8. Regulatory measures to payment systems and broader requirements are linked with lower penetration of foreign X2C online marketplaces

Coefficients of association between DSTRI policy areas and DIP diffusion indicators, 2014-2024

Notes: Please refer to endnote.¹¹

Sources: Author’s elaboration based on data from Semrush and Crunchbase.

Overall, while the type and level of regulatory measures to trade in digitally enabled services vary, the analysis reveals consistent and statistically significant effects on platform participation. Taken together, these restrictions can raise entry costs, dampen competition, and limit the availability and diversity of digital services for consumers. While the analysis identifies statistical associations between DSTRI components and the diffusion of DIPs, unpacking the precise causal mechanisms remains beyond the scope of this paper. Future research could explore these dynamics using quasi-experimental or instrumental-variable approaches to better identify how specific regulatory shocks affect cross-border platform penetration and market contestability.

Notes

← 1.  While the DSTRI provides an aggregate measure of regulatory measures to trade in digitally enabled services, this analysis focuses on its individual policy subdimensions to better identify which specific regulatory areas are most strongly associated with the DIPs diffusion across the three sectors studied. Given that the DSTRI is a composite index, analysing its policy area separately allows for more granular insights into the relative impact of restrictions related to IPR, payment systems, infrastructure, and other restrictions. For reference, regression results using the overall DSTRI score are also provided in Annex C.

← 2.  Two distinct model specifications were estimated for the foreign firm traffic share: one including a control for domestic market competition and one without. The proxy for domestic competition is platform traffic from domestically headquartered firms. Regressions on structural characteristics suggest that this variable is significantly and negatively associated with foreign platform traffic, indicating its relevance as a control.

However, the inclusion of domestic traffic may introduce a bias. On one hand, domestic platform traffic is likely influenced by the same structural characteristics that drive online platform diffusion. Moreover, restrictions to digital trade in services may affect both foreign and domestic platforms. As such, controlling for domestic traffic may inadvertently capture part of the policy effect, leading to a potential downward bias in the estimated association between digital trade restrictions and foreign platform traffic. That is, the true effect of restrictive policies could be underestimated when the domestic competition proxy is included. Conversely, excluding this control could result in an upward omitted variable bias, since the model would not account for the market dynamics between foreign and domestic platforms. In this case, the estimated coefficients might overstate the impact of restrictions if domestic competition serves as a partial substitute for foreign entry. As a result, the estimates from these two specifications can be understood as providing lower and upper bounds for the true association, respectively.

To account for this trade-off, both specifications were estimated. Results are broadly consistent across models, though some differences emerge. Table A.B.2 shows that the model including domestic competition reveals stronger and statistically significant negative associations between restrictions to payment systems and other regulatory measures and traffic from foreign-headquartered pure X2C marketplaces, as well as between restrictions on IPR protection and enforcement and foreign traffic in travel and accommodation platforms. Notably, the negative association between restrictions on IPR protection and enforcement and foreign presence in ride-hailing and carpooling platforms is weaker and less significant in the model that omits the domestic traffic control. This result is intuitive, given the near-complete dominance of foreign platforms in this sector – like Uber, Bolt and Lyft – across most countries, reducing the relevance of the domestic competition proxy.

Given the potential for upward bias when omitting the domestic competition control, the results presented in the figures (e.g., bar plots in the report) correspond to the specification that includes this control. This approach aims to provide more conservative estimates of the association between domestic policies restricting trade in digitally enabled services and foreign DIPs diffusion.

← 3.  The coefficients displayed in the charts are the direct outputs from log-linear regressions, where the dependent variable – such as total traffic or foreign traffic share – is expressed in natural logarithms. These coefficients indicate the expected change in the log of the outcome variable associated with a one-unit increase in the relevant policy score. However, they do not directly correspond to percentage changes in the original scale of the diffusion indicator.

To translate the estimated coefficients into more intuitive percentage changes, we apply the following exact formula:

$Percentage change =\left(e^{\beta \times \Delta x} -1\right) \times 100$

where $\beta$ is the estimated coefficient and $\Delta x$ is the size of the change in the policy variable.

For example, consider the strong negative association observed between restrictions on IPR protection and enforcement and total traffic in the travel and accommodation booking sector. The chart shows an estimated coefficient of -8, which reflects the effect of a full one-unit decrease in the score related to restriction on IPR protection and enforcement – a very large and likely unrealistic policy shift. To better reflect a more plausible reform, we consider a smaller change: a 0.01-point decrease in the policy score. Applying the formula yields:

$Percentage change =\left(e^{-8.1\times (-0.01)} -1\right) \times 100 = \left(e^{0.081} -1\right) \times 100 \approx 8.4\%$

This means that a 0.01-point reduction in digital trade restrictions on IPR protection and enforcement is associated with an 8.4% increase in total traffic in the accommodation booking sector.

All percentage effects discussed in the main text are derived systematically using this exact method, allowing for consistent and policy-relevant interpretation of the results.

← 4.  The observed associations between digitally enabled services trade restrictions and platform diffusion are broadly consistent with expectations: lower regulatory measures tend to support greater traffic and higher foreign participation, particularly for pure DIPs. Sector-specific competition dynamics discussed in Box 3 provide additional context. In ride-hailing, for example, increasing competition over the past decade has likely amplified the impact of moderate regulatory reforms, while in more concentrated markets such as travel and accommodation booking, established platforms’ structural advantages may moderate these effects. This perspective helps explain instances where some restrictions appear positively associated with platform activity, albeit often insignificantly, potentially reflecting the interplay of market structure and incumbent platform strategies.

Other mechanisms beyond regulatory policy likely contribute to platform diffusion patterns. Network effects, platform heterogeneity, and differences in business models shape responsiveness to restrictions. Pure DIPs, which rely on digital-only channels, are more sensitive to local presence requirements than hybrid platforms with physical infrastructure. Similarly, platform features such as capital intensity, review systems, digital versus physical service delivery, internationalisation, and sectoral diversification influence uptake and competitive dynamics (Costa et al., 2021[5]). While these factors are not fully captured in the current macro-level model, the competition dynamics lens offers a plausible framework for interpreting their likely influence.

Future research could extend the analysis by integrating alternative measures of platform activity, exploring joint policy effects, and examining heterogeneity across platform characteristics and sectors. Such work would provide a more granular understanding of the channels through which restrictions to digitally enabled services shape competition, diffusion, and foreign participation.

← 5.  Individual STRI scores for specific sectors, like the STRI for Computer Services, may be better adapted to the sectoral realities of DIPs, rather than the more cross-cutting nature of the DSTRI.

← 6.  The charts display estimated coefficients from equation (1) in Annex A. Each DSTRI policy area is tested individually, controlling systematically for structural factors influencing platform diffusion. For significant results, the dark blue bars represent the marginal effect on the respective diffusion indicator (Total traffic, Traffic from foreign firms, or Share of traffic from foreign firms) for pure DIPs in the ride-hailing and carpooling sector, associated with a one-unit increase in the DSTRI policy score. Black error bars denote clustered standard errors at the country level. Significance levels for rejecting the null hypothesis (no significant association between restriction and diffusion indicator) are indicated as follows: *** p < 0.01 (1%), ** p < 0.05 (5%), * p < 0.10 (10%). For a detailed explanation of the full methodology and empirical specifications, refer to Annex A. For full regression results, see Annex C.