The impact of national technological capabilities on foreign direct investment inflows: an empirical analysis Open Access

While academic literature reveals complex and sometimes contradictory findings on the impact of inward FDI on host countries, the conventional wisdom among most policymakers is that FDI accelerates economic growth. This optimism is reinforced by tangible success stories – such as China (Tseng, 2003), Ireland (Alfaro et al., 2005), Singapore, and Malaysia (UNCTAD, 2011) – where strategic FDI attraction played a crucial role in economic development through job creation, productivity gains, skills development, and transfer of knowledge and technologies. Moreover, international organizations like the World Bank, IMF, and OECD have historically promoted FDI as a key development strategy. Consequently, governments in both developed and developing countries actively compete to attract FDI by offering tax and non-tax incentives, establishing special economic zones, creating investment promotion agencies, and implementing favorable policy frameworks.

Global FDI flows have expanded substantially over recent decades, growing from approximately $204.9 billion in 1990 to a peak of $2.1 trillion in 2015, despite periodic fluctuations due to economic crises (UNCTAD, 2023). However, this expansion has been uneven across countries, with some economies emerging as dominant FDI recipients while others continue to struggle in attracting investment. This disparity has motivated extensive theoretical and empirical research into the determinants of FDI flows across countries (Faeth, 2009; Gupta and Tyagi, 2024; Islam and Beloucif, 2024)

Among the various factors that influence FDI attraction, a country's technological capabilities have gained increasing attention in recent literature. Technological capabilities, including R&D capacity, innovation potential, patents, and advanced manufacturing infrastructure, are more and more being viewed as critical economic determinants that shape a host country's attractiveness to multinational corporations (MNCs. These capabilities matter for two interconnected reasons. First, they enable absorptive capacity, allowing host countries to identify, assimilate, and build upon foreign knowledge transferred through FDI (Dang and Merino, 2024; Tan et al., 2023; Viglioni and Calegario, 2023). Second, they create locational advantages that attract technology-seeking and asset-seeking MNCs by signaling productivity potential and competitive strength (Anand and Kogut, 1997; Castelli et al., 2025; Kim and Choi, 2020; Tu, 2024).

Despite growing interest, empirical evidence on the relationship between technological capabilities and FDI remains mixed and often contradictory. Some studies report positive associations between host-country technology indicators and FDI inflows (Palit and Nawani, 2007; Singhania and Gupta, 2011; Tu, 2024), while others find negative effects (Anand and Kogut, 1997; Kumari and Sharma, 2017) or non-linear relationships (Kim and Choi, 2020). These inconsistencies create confusion for policymakers who invest heavily in R&D infrastructure and innovation systems with the expectation of attracting FDI, yet lack clear guidance on which technological investments yield the strongest returns.

In the current study, we argue that one of the sources of these contradictory findings lies in the failure to distinguish between innovation inputs and innovation outputs. Innovation inputs such as R&D expenditures represent early-stage investments in research and knowledge creation. While they signal potential for future innovation, they also reflect high operating costs, skilled labor wages, and uncertain commercial outcomes. On the other hand, innovation outputs like patents and high-tech exports represent tangible and commercialized results of innovation efforts. They signal proven technological capability, market integration, absorptive capacity, and economies of scale. However, prior research has treated technological capability as a monolithic construct, using R&D, patents, or exports interchangeably. No study has yet systematically decomposed technological capability into inputs and outputs within a unified empirical framework, tested their differential effects on FDI, or explained why these effects might systematically diverge, particularly in relatively homogeneous, high-income contexts such as OECD countries.

This study, therefore, addresses this gap by adopting a multi-dimensional approach to analyze how technological capabilities influence FDI inflows. Using panel data covering 38 OECD countries from 2005 to 2022, we separately examine the effects of innovation inputs (R&D expenditure as a percentage of GDP) and innovation outputs (patent applications and high-tech export share) on FDI inflows. Furthermore, by incorporating more recent data and disaggregating technological capabilities, it provides a more comprehensive and policy-relevant analysis of the role of technology in FDI inflows.

Several theories have been developed to explain why MNCs engage in FDI and why they choose to operate in certain countries over others. While no single theory fully accounts for the location decisions of MNCs (Denisia, 2010), these theories have some implications for the role of technological capabilities in attracting investment. We draw on three complementary theoretical perspectives to explain why innovation inputs and outputs should affect FDI differently: the Eclectic Paradigm (OLI), Internalization Theory, and the International Product Life Cycle (IPLC).

1. The Eclectic Paradigm (OLI)

Dunning’s Eclectic Paradigm (1977, 1988, 2000) remains the most influential framework for analysing FDI determinants. The OLI framework posits that for a firm to engage in FDI, three concurrent conditions must be met. First, the firm must possess unique ownership advantages (firm-specific assets such as technology, brand, or managerial expertise) that enable it to compete effectively in foreign markets. Second, the host country must offer location advantages (country-specific factors such as market size, resources, infrastructure, or policy environment) that make direct investment preferable to exporting. Finally, internalization advantages must exist, meaning it must be more profitable for the firm to exploit its ownership advantages through its own internal networks rather than licensing them to the local firms.

Originally, location advantages were primarily associated with natural assets that are tangible, such as low labor costs, resource endowments, and market size (Andy et al., 2024). However, the emergence of the knowledge-based global economy has fundamentally reshaped the “L” component of OLI. Recent scholarly works have expanded location advantages to include digital infrastructure, data governance, innovation system quality, intellectual property regimes, and human capital (Meyer et al., 2023; Narula and Verbeke, 2015; OECD, 2022). Kafouros et al. (2012) argue that MNEs strategically choose locations based on the potential for technological advancement and benefit most from global knowledge reservoirs when they operate in industries with high technological opportunities. Iammarino and McCann (2013) reinforce this view by showing that MNCs’ location choices are driven not merely by cost considerations, but by the strength of local technological capabilities, research infrastructures, skilled human capital, and innovation ecosystems. Meyer et al. (2023) show how differences in innovation capacity, digital human capital, and entrepreneurial ecosystems create either opportunities or barriers for expansion, making technologically advanced environments more attractive for digital and digitally transforming firms. Similarly, based on a systematic review of location-attractiveness research, Andy et al. (2024) argue that innovation capability and technological advancement have become pivotal determinants of host-country attractiveness in the post-COVID “new norm,” as MNEs increasingly prioritize locations that enable knowledge creation, absorptive capacity, and digital transformation.

1. Internalization theory

Internalization theory (Casson et al., 2016; Rugman and Verbeke, 2008) argues that firms choose FDI over alternative entry modes such as licensing or joint ventures when the benefits of internalizing proprietary assets outweigh the costs of market transactions. Internalization minimizes transaction costs, protects valuable intangible assets (such as technology or brand reputation), and reduces risks associated with knowledge spillovers and opportunistic behavior by partners.

Recent extensions of Internalization Theory emphasize knowledge and data governance as central concerns, particularly in the digital economy where firm-specific advantages increasingly take the form of intangible assets and data (Banalieva and Dhanaraj, 2019; Narula et al., 2019). In technologically advanced host countries, the density of capable firms, research institutions, and skilled labor increases opportunities for collaboration and learning, thereby strengthening the location’s attractiveness for technology-seeking and strategic asset-seeking FDI. At the same time, such environments may heighten knowledge leakage, imitation, reverse engineering, and appropriation by local competitors, especially when knowledge is tacit and difficult to contractually safeguard. This creates a dual effect; technologically advanced locations can simultaneously attract knowledge-seeking FDI by offering high-quality innovation ecosystems, while also inducing foreign firms to internalize more tightly, through wholly owned subsidiaries, stronger control over R&D functions, and stricter governance of data and knowledge flows, in order to protect and appropriate returns from their proprietary assets.

From the input-output perspective, the internalization theory implies that high R&D intensity and patenting signal both opportunity and risk. On the one hand, they indicate a vibrant innovation ecosystem. On the other hand, they reflect competitive environments where knowledge protection is critical, and entry barriers are high. High-tech exports, by contrast, signal market-tested capability and established production ecosystems where integration risks are lower and collaboration opportunities are clearer. Thus, innovation outputs reduce uncertainty and facilitate entry, while innovation inputs may heighten caution and push firms toward tighter control or alternative entry strategies.

1. International Product Life Cycle theory (IPLC)

Vernon’s (1966) IPLC theory extends the traditional trade theory by emphasizing how firms’ internalization patterns are shaped by the changing nature of the products they produce. The theory highlights four distinct life cycle stages of products that influence location decisions for production: introduction, growth, maturity, and decline. In the new-product stage, innovation occurs in technologically advanced countries with strong R&D capacity, skilled labor, and proximity to lead markets. Production remains localized, costs are high, and the product is typically exported to foreign markets. As the product matures, processes become standardized, and production shifts toward locations offering cost advantages while maintaining adequate capability. In the standardization stage, production moves to low-cost locations, and the product becomes a commodity.

The IPLC logic suggests that FDI is increasingly driven by cost considerations as products mature and become standardized (Vernon, 1966, 1992). Therefore, host technological capability is less central than wage advantages and production cost efficiency. Regarding the input-output distinction, innovation inputs (R&D expenditure) correspond to the new-product stage: high research intensity, localized production, uncertain commercial outcomes, and limited FDI inflows as the innovating country acts primarily as an exporter of products rather than a recipient of investment. Innovation outputs (patents, high-tech exports), however, correspond to the maturing-product stage: proven technologies, standardized processes, established markets, and economies of scale. At this stage, FDI flows toward countries that offer proven manufacturing capability, value-chain integration, and market access rather than frontier research. High-tech exports are a particularly strong signal of this transition: they demonstrate that a country has successfully moved from R&D to large-scale commercialization and international competitiveness (Ackrill and Çetin, 2025).

There are considerable, though not overwhelming, empirical studies that examined the role of the technological capabilities of host countries in attracting FDI inflows. Anand and Kogut (1997) provided one of the earliest studies, analyzing FDI entries into the United States. Contrary to expectations, they found little evidence that foreign firms invested in the US primarily to access superior American technology as measured by R&D expenditure. Instead, FDI appeared driven by technological rivalry among firms and industry-specific conditions rather than the technological attractiveness of the host country. A decade later, Palit and Nawani (2007) examined Asian economies and reached the opposite conclusion: countries with strong R&D-based technological capabilities and well-developed IT infrastructure attracted significantly more FDI. Singhania and Gupta (2011) similarly reported a positive effect of a positive effect of patent applications on FDI inflows to India, while Kumari and Sharma (2017) found a negative impact of R&D expenditure on FDI in developing countries.

More recent work continues to produce mixed findings. Kim and Choi (2020) use panel data from 35 OECD countries between 2000 and 2015 and uncover a U-shaped relationship between a host country’s technological capability and FDI, suggesting non-linearity. Tu (2024) documents a positive influence of innovation (proxied by high-tech exports and patent intensity) in enhancing FDI attraction in developing countries from 2013 to 2021.

Considering studies that use composite indices, Iqbal et al. (2016) construct their technological capability indicator as a weighted average of R&D manpower, patents, and scientific publications, to examine the impact of domestic technological capabilities on FDI in the Indian sub-continent. They found that the host country’s technological capability is a primary determinant for attracting FDI.

To sum up, prior research has used various proxies, R&D expenditure, patent counts, patent citations, high-tech exports, and researcher density, often interchangeably and sometimes as composite innovation indices. No study has systematically decomposed technological capability into inputs and outputs within a unified empirical framework to test their differential effects on FDI. However, as we discussed in detail above, the input-output distinction remains empirically relevant even in relatively homogeneous, advanced economies.

Based on the above-discussed theoretical and empirical synthesis, we propose two key testable hypotheses:

To investigate the impact of technological capabilities on FDI inflows, this study employs panel data spanning 2005–2022 for 38 OECD countries ^[1]. The analysis is confined to this period as the FDI inflow series utilized in this analysis is only available from 2005 onwards.

FDI inflow data for the sample countries is obtained from the OECD's FDI Statistics database, which is compiled according to the Benchmark Definition 4th Edition (BMD4) standards (OECD, 2008). A key advantage of this dataset is its identification and separation of Special Purpose Entities (SPEs) – entities that primarily engage in facilitating internal financing of multinational enterprises with little or no physical presence in the host country – from FDI statistics (Wacker et al., 2025). This exclusion enables more reliable capture of substantive cross-border investment by filtering out transactions that would otherwise misrepresent actual investment flows. In addition, the BMD4 approach has introduced the ultimate investing country (UIC) framework, which tracks capital flows to their original source rather than stopping at immediate investors, thereby identifying the true origin of international investments (Borga and Caliandro, 2023).

Data on technological capabilities are sourced from the July 2024 edition of the OECD Main Science and Technology Indicators (MSTI) database. This database provides internationally comparable statistics on science, technology, and innovation activities across OECD member and partner countries. It includes key indicators such as research and development expenditure, personnel engaged in R&D activities, patent statistics, and other metrics that capture various dimensions of technological capabilities, making it a widely recognized source for cross-country comparative analyses of innovation systems and technological development. All the additional control variables, including GDP, labor cost and productivity, education levels, and corporate tax, are also obtained from OECD databases.

Dependent variable: Inward FDI is defined as investment made by non-resident direct investors in enterprises resident in the host (reporting) economy. The data source for this variable (expressed in millions of euros) is OECD’s “FDI main aggregates, BMD4”, which is based on balance of payments statistics published by Central Banks and Statistical Offices following the recommendations of the IMF’s BPM6 and the OECD’s BMD4. And it comprises investments by foreign direct investors in resident enterprises, adjusted for reverse investments and cross-border transactions between fellow enterprises under common foreign control (OECD, 2008).

Independent variables: To capture the effect of the different stages of host countries’ technological capabilities, this study adopts the third dimension of Lall’s (1992) country-level capability framework, which relates to technological effort. This framework allows us to distinguish between the inputs and outputs of the innovation process (Lall, 1992).

For input-level technological capability, we use Gross Domestic Expenditure on R&D as a percentage of GDP. This indicator captures all spending on R&D carried out within each economy each year, comprising sectoral breakdowns by performing and funding sectors. R&D expenditure is used in the innovation and technological capability literature as a standard proxy for innovation input (Dziallas and Blind, 2019).

For innovation output, we employ two distinct measures. First, we use the number of patent applications to the Patent Co-operation Treaty (PCT), representing the total number of applications filed across all sectors. The priority date – the date of the first international filing of a patent – is chosen as the reference date. Patents are classified as innovation output indicators in major international frameworks (Eurostat IOI; WIPO Global Innovation Index; OECD Oslo Manual, 2018), as they represent completed inventions that have undergone examination for novelty and non-obviousness, distinguishing them from earlier-stage research investments. Second, we use high-tech export market share, which we calculate as a single composite index using Principal Component Analysis (PCA), incorporating export market shares for three selected groups of products: “pharmaceutical products”, “computer, electronic, and optical products”, and “aerospace products”. This again is consistent with the institutional definitions and measurement frameworks like the European Commission’s Innovation Output Indicator (IOI) and WIPO’s Global Innovation Index (GII).

Control variables: Following prior literature, we control for other factors that have the potential to influence FDI inflows. GDP (million current PPP $) is included to control for the effect of the size of the host country’s economy, a traditional determinant of market-seeking FDI. Labor market characteristics such as labour force (thousand persons), unit labour costs by hours worked (index, 2015 = 100), labour productivity annual growth rate (%), and population with tertiary education among 25–34-year-olds (% of age group) is also included to take into account the quality, availability, and cost competitiveness of the human capital available in the host country, which are key determinants of resource and efficiency-seeking FDI. We control for the tax effect using Tax on Corporate Profits (% of GDP), as relatively low corporate taxes can be a significant incentive for direct investment. Summary statistics for all variables used in the analysis are reported in Table 1.

Given the longitudinal nature of our dataset, we employ a panel estimation method, specifically Feasible Generalized Least Squares (FGLS), which is well-suited to address common econometric challenges in panel data analysis. Panel data structures typically exhibit heteroskedasticity, where the variance of errors differs across cross-sectional units, and autocorrelation, where residuals are correlated over time within the same unit, both of which can lead to inefficient estimates and biased standard errors if left unaddressed (Parks, 1967). These econometric issues commonly arise in longitudinal FDI data due to country-specific variance patterns and serial correlation in investment flows over time. The FGLS estimator corrects for panel-specific heteroskedasticity and within-panel autocorrelation, thereby enhancing estimator efficiency and the robustness of statistical inferences (Bai et al., 2021).

To determine the appropriate model specification, we conducted the Breusch-Pagan Lagrange Multiplier test and the Hausman test. The Chi-squared statistics indicate that the random-effects estimator is preferred over fixed effects alternative. We, therefore, estimate the baseline models under the random-effects assumption. Nevertheless, given the importance of this assumption in the FDI–technology literature, we additionally report fixed-effects estimates as a robustness check in  Appendix Table A1; the sign and overall pattern of results remain unchanged.

Note that the relationship between technological capabilities and FDI may be bidirectional, as foreign investment can generate spillovers that enhance host-country innovation capacity (Ali et al., 2023), while simultaneously, a country's technological environment influences its attractiveness to foreign investors (Kim and Choi, 2020). To mitigate potential simultaneity bias and ensure our estimates are not driven by contemporaneous reverse causality, we conduct robustness checks using lagged technological variables. Specifically, we re-estimate our baseline model by replacing current-period technological indicators (R&D expenditure, patent applications, and high-technology exports) with their one-year (t−1) and two-year (t−2) lagged values. This approach ensures that the technological capability measures are predetermined relative to current FDI inflows, thereby reducing the likelihood that our results reflect FDI affecting technology rather than the reverse. Furthermore, the use of lagged explanatory variables is a well-established method in the FDI literature for addressing endogeneity concerns (e.g. Alfaro et al., 2004; Saeed et al., 2024).

Table 2 presents our regression results across three model specifications designed to assess both the baseline relationships and their robustness to endogeneity concerns. Column 1 reports the baseline FGLS random effects estimates, which account for panel-specific heteroskedasticity and autocorrelation in our cross-country longitudinal data. Columns 2 and 3 present robustness checks using one-year and two-year lagged technological variables, respectively, thereby ensuring that innovation indicators are predetermined relative to current FDI flows and reducing potential simultaneity bias.

The results reveal a clear and consistent differentiation among the three technological capability measures. In the baseline specification, R&D expenditure and patents exhibit negative and statistically significant coefficients, while high-technology exports show a strong and positive effect on FDI inflows. The coefficient for R&D expenditure (% of GDP) is −5,360.32 (p < 0.05), and that for patent applications is −2.15 (p < 0.01). Furthermore, the one-year lagged specification (Column 2) closely replicates the baseline pattern, with R&D expenditure remaining negative and significant (−5,882.73, p < 0.05), and patents remaining negative and significant (−1.48, p < 0.01). Although the magnitude and statistical significance of R&D expenditure weaken at the two-year lag (Column 3), the negative direction persists. These negative relationships indicate that, on average, countries with higher domestic innovation intensity tend to attract less inward FDI.

Two complementary explanations can account for this finding. First, market competition effects arise when strong domestic R&D activity and patenting signal a well-established local innovation base. In such environments, local firms are already investing heavily in technology, reducing the need for external sources of innovation. As a result, foreign firms face strong domestic competitors and fewer opportunities for technological spillovers or market capture. FDI is therefore less motivated to enter economies where indigenous innovation systems are already mature and self-sufficient.

Second, high investment costs often accompany strong R&D intensity. Economies characterized by advanced innovation systems typically display high wages, skilled labor, and strict regulatory environments, including robust intellectual-property regimes. While these factors strengthen long-term competitiveness, they also raise operational costs and barriers to market entry. Cost-sensitive or efficiency-oriented investors may thus prefer alternative locations offering lower expenses and greater flexibility.

In contrast, high-technology exports exert a strong positive and highly significant impact on FDI inflows (c = 14,185.12, p < 0.01). This positive relationship proves highly robust across all model specifications: the one-year lagged estimate (11,215.57, p < 0.01) and two-year lagged estimate (8,345.67, p < 0.01) both maintain strong statistical significance, confirming that countries with larger high-tech export shares are demonstrably more successful in attracting foreign investors.

This relationship can be interpreted through two interconnected effects. First, market-opportunity effects: a robust high-tech export base indicates that a country is integrated into global value chains (GVCs) and maintains a competitive presence in international markets. This integration signals stability, openness, and the potential for collaboration, thereby attracting investors seeking access to established networks and technologically sophisticated partners. Second, economies-of-scale effects: a strong export base reflects large-scale manufacturing capacity and efficient production infrastructure, enabling lower unit costs and productivity advantages for incoming investors. These features reduce uncertainty and facilitate the establishment of joint ventures, supplier relationships, and R&D partnerships.

Consequently, export performance serves as a tangible signal of technological maturity and absorptive capacity, converting innovation outcomes into credible investment opportunities.

The opposing signs of the coefficients for R&D/patents and high-tech exports can be reconciled by considering the different stages of technological development represented by these variables. R&D expenditure and patents capture the input side of innovation – the early phases of research and experimentation. At this stage, technologies are still being developed, and commercial viability is uncertain, limiting immediate investment opportunities for foreign firms. By contrast, high-tech exports represent the output stage of innovation, when technologies have been successfully commercialized and embedded within production systems. These mature technological capabilities provide MNEs with ready-to-integrate opportunities for production, sourcing, or technology collaboration.

This interpretation mirrors the logic of the IPLC theory. In the new-product stage, countries invest heavily in R&D and innovation, but they tend to act as innovation exporters rather than FDI recipients. In the maturing-product stage, production processes become standardized, and FDI shifts toward economies with proven manufacturing capacity and established technological networks. Thus, the empirical pattern observed here – negative R&D and patent effects but positive export effects – reflects the structural transition from innovation creation to commercialization.

Among the control variables in Table 2, several results conform to theoretical expectations. GDP shows a positive and highly significant effect (c = 0.02, p < 0.001), reaffirming the importance of market size as a core determinant of FDI attraction. Tertiary education is also positive and significant (c = 820.45, p < 0.001), underscoring the relevance of skilled human capital for technology-oriented investment. Conversely, unit labor costs display a negative and significant coefficient (c = −532.31, p < 0.001), suggesting that cost competitiveness remains a key location factor, especially for efficiency-seeking investors. Other controls – labor productivity growth, labor-force size, and corporate-tax burden – are statistically insignificant, implying that their influence is largely captured by the broader economic and structural variables already included in the model.

This study examined whether innovation inputs and outputs exert differential effects on FDI inflows and why such differences emerge. Drawing on panel data from 38 OECD countries (2005–2022), we tested two hypotheses derived from the OLI paradigm, Internalization theory, and IPLC framework. Our findings reveal a clear but nuanced pattern: innovation input (proxied by R&D expenditure) negatively affects FDI inflows, while innovation output in terms of high-tech exports strongly and positively attracts foreign investment. Unexpectedly, our other output measure, patents, exhibits a negative relationship with FDI.

Our finding that high R&D expenditure discourages inward FDI supports our first hypothesis and contradicts simplistic assumptions that research intensity universally attracts investment. Two complementary mechanisms explain this negative relationship. One explanation is the competition and crowding effect: countries with strong R&D capacity already host advanced domestic innovation systems, reducing the marginal advantage that foreign firms can contribute and intensifying local rivalry. This interpretation aligns with Anand and Kogut's (1997) early finding that US R&D intensity did not attract technology-seeking FDI, as foreign firms faced technological rivalry rather than complementary opportunities.

A second mechanism is the cost-structure effect. Advanced research-intensive economies typically feature high wages for skilled labor, strict regulatory environments, and robust intellectual property protection – all of which raise barriers to entry. This finding resonates with Kumari and Sharma (2017) evidence that R&D expenditure negatively impacts FDI in developing economies where cost considerations dominate location decisions.

Viewed through Internalization Theory, high R&D environments present heightened appropriation risk. They signal the presence of capable local firms and research institutions that could potentially imitate, reverse-engineer, or appropriate foreign technologies. MNCs may therefore avoid high-R&D locations to protect proprietary knowledge, or alternatively, enter only through tightly controlled wholly owned subsidiaries rather than collaborative ventures. Firm-level research supports this selectivity: sophisticated investors scrutinize the structure and governance of local innovation ecosystems before committing capital (Micocci et al., 2025).

In contrast, the strong positive coefficient on high-technology exports indicates that innovation outputs matter more than inputs when it comes to attracting FDI. High-tech exports serve as credible signals of successful integration into global value chains. Countries that export advanced products demonstrate proven market access, established supply networks, competitive production costs, and stable trade relationships, factors that reduce uncertainty for foreign investors seeking to integrate into international production systems., competitive production systems, and outward orientation. This interpretation aligns with Tu’s (2024) recent evidence that innovation positively influences FDI attraction of host countries. Slavik et al. (2025) similarly frames “market acceptance” as demonstrated product-market fit with growing sales, which strengthens the investment case behind commercialization.

Within the OLI framework, high-tech exports represent tangible location advantages that MNCs can immediately exploit. Unlike R&D expenditure – which signals potential but uncertain future capability – exports demonstrate that a country has successfully converted innovation into competitive, market-tested products embedded in functioning production ecosystems. This aligns with recent extensions of OLI emphasizing that in knowledge-based economies, location advantages derive from proven innovation outputs integrated into digital infrastructure, skilled workforces, and established value chains (Meyer et al., 2023; Narula and Verbeke, 2015).

The negative coefficient on patent applications, despite patents being an innovation output, warrants explanation. One possible explanation can be that in the advanced, homogeneous OECD context of our sample, patents signal competitive intensity rather than collaborative opportunity. In technologically saturated markets, extensive patenting may reflect defensive strategies by domestic firms to protect market positions, creating barriers for foreign entrants. Moreover, it may indicate technological maturity and market saturation rather than emerging opportunities. Countries with extensive patent portfolios have already occupied technological niches, leaving less room for foreign firms to establish competitive positions. This contrasts with high-tech exports, which signal ongoing production activity and market demand rather than past inventive activity.

The overall pattern, negative R&D and patents, positive high-tech exports, can be interpreted through IPLC theory, which predicts that FDI patterns shift as technologies and products mature from innovation to standardization. High R&D intensity corresponds to the early “new-product” phase, in which countries lead innovation, but production remains localized and expensive. Patents, while technically outputs, may in OECD contexts represent a transitional stage, inventions have been formalized but not yet fully commercialized or integrated into large-scale production. As products mature and technologies standardize, production migrates to economies that can operate at scale – precisely those characterized by strong high-tech export performance. The positive coefficient on exports, therefore, captures the “mature-product” stage: a point at which innovation has been commercialized, processes standardized, and the environment made conducive to FDI inflows.

Taken together, these results provide empirical support for the modernization of classical FDI theories. They confirm the OLI proposition that location advantages have evolved from cost-based to capability-based, the Internalization argument that technological risk and information control shape entry modes, and the IPLC logic that FDI follows the product’s maturation from R&D to standardized production. The findings also extend Kim and Choi (2020) finding of a U-shaped relationship between technological capability and FDI in OECD countries. Their non-linear pattern likely captures the progression from input-intensive research systems (low FDI) through patent-intensive competition (continued low FDI) to export-intensive commercialization (high FDI).

Finally, the negative effects of R&D and patents should not be misconstrued as evidence against innovation. Rather, they reveal structural imbalance: innovation without commercialization capacity produces knowledge that cannot be economically absorbed. As Crescenzi and Iammarino (2017) and Narula et al. (2019) emphasize, the embedding of innovation within regional systems and global networks is essential for translating technological potential into investment. When such embedding is weak, even strong R&D performance fails to attract foreign capital. Our results, therefore, substantiate the “technology paradox”: innovation can both attract and deter FDI depending on whether it raises entry costs or enhances market opportunities. Khan et al. (2025) support those findings and add that banks have a significant effect on new business/startup with local financing infrastructure, which can materially shape new-firm formation alongside FDI. The evidence thus positions technological capability not as a static determinant but as a dynamic moderator shaped by institutional evolution and the structure of global value chains.

The policy implication is clear: governments should not treat R&D expenditure or patent counts as sufficient to attract inward FDI. Instead, they must focus on translating innovation inputs into trade-ready, export-oriented capability, integrating local firms into global value chains, and fostering digital and institutional infrastructure that makes foreign investment in technology viable. Recent reports highlight that digital readiness and investment facilitation matter greatly in shaping the composition of FDI.

Limitations of our study reflect areas for future research: First, our model does not explicitly test nonlinear effects or interactive effects. Scholars have found threshold effects in innovation–investment relationships that our linear specification may not capture fully. Second, while we establish differential effects of innovation inputs versus outputs on aggregate FDI inflows, we do not explore potentially important heterogeneities across FDI types, sectors, or institutional contexts. Future research could disaggregate, for instance, FDI by entry mode (greenfield vs merger and acquisition) to assess how technological capability interacts with the type of investment or examine sectoral variation to determine whether the input-output distinction holds uniformly across industries.

In conclusion, the evidence supports a more refined theory of technology and FDI: not all innovation is equal in the eyes of foreign investors – observable exports of high-technology goods matter more than simply inputs into innovation systems. Locational advantages thus derive from demonstrated production and trade capabilities, not merely from potential. This recognition invites both scholars and policymakers to move beyond quantity of innovation inputs toward quality, connectivity, trade-oriented technology systems, and to revisit the assumptions of classic FDI theories in a digital, globalized economy.

Portions of the text were refined with the assistance of an AI tool (ChatGPT, OpenAI, version GPT-5.1). The tool was used exclusively to improve the clarity, grammar, and overall readability of the manuscript. All ideas, interpretations, and conclusions remain the sole responsibility of the author. The final content has been thoroughly reviewed and verified for accuracy and alignment with the research.