National Contexts and R&D Transference: A Cross-Country Analysis of Entrepreneurial Perceptions and Policy Impact

Abstract

This study aims to assess whether country-level conditions traditionally associated with fostering entrepreneurship (such as digitization, economic climate, and government policies) explain the extent to which R&D leads to new commercial opportunities for small and medium-sized enterprises. Using data from the Global Entrepreneurship Monitor’s National Expert Survey, this study applies factorial analysis of variance models to evaluate the single and interaction effects of multiple categorical variables on R&D transference. Three models are constructed to test the significance of country: entrepreneurship-supportive factors, entrepreneurs’ recommendations, and pandemic-related government actions. The results reveal that country location consistently and significantly explains variance in R&D transference. In contrast, traditional entrepreneurship-fostering factors are not found to be statistically significant. Pandemic-related government actions also fail to demonstrate measurable explanatory power on R&D transference. These findings suggest that broader national contexts, rather than specific entrepreneurship-supportive conditions, are more critical in shaping R&D outcomes. Policymakers should reconsider the assumption that general entrepreneurship support mechanisms automatically enhance R&D transference. Instead, targeted interventions may be more effective. Entrepreneurs and investors should evaluate national contexts beyond perceived support factors when pursuing R&D-based innovation. The study underscores the need for context-sensitive strategies in fostering innovation and challenges one-size-fits-all approaches to entrepreneurship policy.

1. Introduction

Research and development (R&D) transference is a critical driver of innovation and economic growth (Audretsch et al., 2024). However, the extent to which R&D translates into viable commercial opportunities for small and medium-sized firms is influenced by both internal capabilities and external environmental factors (Arshad et al., 2023). While prior research has extensively examined the role of firm-level characteristics in facilitating R&D transfer (Zhang et al., 2022; Davcik et al., 2020), the broader national context remains relatively less explored. This study addresses a significant gap in the literature by investigating whether country-level conditions traditionally associated with advancing entrepreneurship are meaningfully associated with the level of R&D transference. Drawing on data from the Global Entrepreneurship Monitor (GEM)’s National Expert Survey (NES), the research applies factorial analysis of variance (ANOVA) models to evaluate the significance of country, entrepreneurs’ perceptions, and pandemic-related government actions for explaining variance in R&D transference.

The objectives of this study are threefold. First, it seeks to determine whether macro-level entrepreneurship-supportive factors are related to R&D transfer. Second, it examines the extent to which entrepreneurs’ recommendations for improving the entrepreneurial ecosystem align with R&D transference. Third, it evaluates the effects of government interventions during the COVID-19 pandemic. By adopting a cross-country analytical approach, this paper contributes to a more refined understanding of the conditions under which R&D can be effectively translated into entrepreneurial opportunity. Broadly, the paper strives to add to the understanding of the relative importance of national context vs. specific entrepreneurship-supportive conditions in fostering innovation.

Previous work found that R&D transfer for small businesses is shaped by internal capabilities (Awan et al., 2021), policy support (Adomako & Tran, 2024), and organizational arrangements (K. Chen et al., 2022). The current investigation aims to enhance the understanding of the overall relationship between the confluence of factors across diverse national contexts and the R&D transference. These findings are intended to provide information to entrepreneurs seeking to recognize when and how they can benefit from R&D transfer and to policymakers looking to improve the national climate that supports R&D transference. Despite the wide array of existing research, some gaps remain, particularly regarding the long-term impacts of different types of R&D support and the interplay between internal and external factors in diverse economic contexts. The literature provides support for the fact that the impact of entrepreneurial characteristics and organizational arrangements are context-dependent (V. Kumar et al., 2024), while the roles of external cooperation and government support are more globally accepted as drivers of R&D (Guerrero et al., 2020).

The purpose of the current study is to address the gap in knowledge about the connection between R&D transfer level and the key factors that foster entrepreneurship at the macro level. The extant literature provides robust evidence of distinct positive impacts of individual factors on R&D transfer in specific situations or at particular locations (Chun & Mun, 2012; Ngo, 2020; Kou et al., 2020; Fletcher et al., 2022; Okamuro, 2007; Zhang et al., 2022; Song et al., 2017; Davcik et al., 2020; Alam et al., 2019). Yet, there are few studies aiming to test whether the factors supporting entrepreneurship, such as digitization and government policies (de Lucas Ancillo & Gavrila, 2023), make a difference on the extent of R&D transfer, when considered together, at the country level. The present paper adds to this understanding.

The main studies on the phenomenon of firm-specific and national-level factors that drive R&D transfer explored and tested it in a country-level context. Examples include a firm-level analysis of Korean small businesses (Chun & Mun, 2012), a similar investigation of FDI linkages and technology transfer based on a sample of Vietnamese small companies (Ngo, 2020), an empirical analysis of R&D financing of Chinese small enterprises (Kou et al., 2020), and an examination of organizational and contractual factors in the context of Japanese small businesses (Okamuro, 2007). The present paper advances this knowledge by taking a more holistic approach to investigating the phenomenon by studying it across countries. The methodology employed (analysis of variance based on survey results) is in line with the models employed by these studies that include a probit model based on data obtained from a firm-level survey (Chun & Mun, 2012) or regressions on data from surveys (Okamuro, 2007; Zhang et al., 2022).

In line with prior studies and other cross-country GEM-based analyses (e.g., Martinez-Gonzalez et al., 2021), the present paper provides a cross-country comparative lens. This perspective broadens the geographic scope and allows for conclusions on universal vs. context-specific factors in innovation and entrepreneurship. This paper’s emphasis on entrepreneurial perceptions and policy interventions complements institutional support and policy coherence themes (such as in Kruger & Steyn, 2020) and strategic planning and innovation frameworks (as discussed by de Lucas Ancillo & Gavrila, 2023). Generally, this study contributes to the discourse on innovation strategies in varied national contexts.

The present study evaluates the extent of R&D transference, specifically the variance in R&D transference that is explained by contextual entrepreneurship factors. The literature discusses a wide array of factors (e.g., institutional quality, digital readiness, government support). The current study focuses on perceptions of key factors that can explain differences in R&D transference levels. The structure of this paper includes a literature review that identifies existing studies on internal and external factors that are related to R&D transfer, including institutions, policies, and entrepreneurial perceptions. This section concludes with hypotheses drawing from current understanding and gaps in extant knowledge. A methodology section follows, which describes the use of the GEM NES dataset and its methodological rigor. This section also details the statistical approach (factorial ANOVA) and introduces three models analyzing different combinations of variables. Results of statistical analysis follow, highlighting key tests and findings. The ensuing section discusses the relevance of findings for policymakers, entrepreneurs, and innovation strategists. The last part summarizes key insights, discusses limitations, and proposes directions for further investigation into national mechanisms and innovation metrics.

2. Literature Review

R&D transference to small businesses is a complex phenomenon that depends on external and internal factors. The ease with which R&D provides value and new commercial opportunities is determined by the institutional environment and by the firm’s internal capabilities (Kou et al., 2020). Government interventions and the entrepreneurs’ perceptions can add to the complexity. Five main themes in the prior literature are relevant to the present study on R&D transfer in the entrepreneurial context: research on external factors; works identifying internal factors; multi-factor studies; lessons from the COVID-19 pandemic interventions; and findings of papers on entrepreneurs’ perceptions.

Focusing on the first theme of external factors, the extant literature addresses three core aspects: cooperation; national innovation systems; and specific factors that are directly related to R&D transference.

First, it was found that cooperation is among the most critical factor affecting R&D transfer. Being part of a cooperation network significantly increases the likelihood of R&D transfer, especially for the smallest firms, though their limited size can be a barrier (Chun & Mun, 2012; Okamuro, 2007). Cooperation among academia, industry, and government (the Triple Helix model) is critical for resource mobilization, credibility, and market access (Aarikka-Stenroos & Sandberg, 2012; Farinha et al., 2016). More specifically, discoveries from universities and research institutions can strongly enhance R&D capabilities in small firms (García-Vega & Vicente-Chirivella, 2020; Song et al., 2017). Recent extant studies advocate for the entrepreneurial university concept and the need for targeted R&D collaboration (Kuzior et al., 2024; Tolin & Piccaluga, 2025). National laws regarding intellectual property ownership have a major impact on university–industry technology transfer (K. Chen et al., 2022). Cooperation with supply-chain and foreign direct investment (FDI) partners is also important, especially in developing economies, as they can facilitate direct technology transfer between multinational firms and young small businesses (Ngo, 2020). Clear costs and outcomes sharing rules, and low transaction and coordination costs in cooperative R&D arrangements, increase the chances of successful R&D transfer and commercialization (Okamuro, 2007; Song et al., 2017).

Secondly, extensive research examined national innovation systems (NIS) to evaluate how countries organize, coordinate, and leverage institutions, policies, and resources to foster innovation and drive economic growth (Narayanan et al., 2022; Lopez-Rubio et al., 2021; Vetsikas & Stamboulis, 2023; De Freitas, 2023). Comparative studies revealed significant diversity in NIS structures, strategies, and outcomes across countries, shaped by historical, institutional, and policy contexts (such as in Balzat & Hanusch, 2004; Anouze et al., 2024; W. Chen & Song, 2024). In developing countries, NIS research focused on adapting models to local contexts by addressing resource constraints and fostering knowledge absorption (Weerasinghe et al., 2023). This research avenue concluded on the need to enhance institutional coordination (W. Chen & Song, 2024) and to include elements of digital economy and sustainability into NIS frameworks (W. Chen & Song, 2024).

Third, the prior literature explored specific external factors like regulation and infrastructure (Kruachottikul et al., 2023; Son et al., 2020). The institutional quality at the location of R&D transfer, including rule of law and corruption, strongly influences R&D investment and transfer (Alam et al., 2019). Sympathetic cultural and social norms can positively influence entrepreneurs’ motivation to engage with R&D and to pursue innovation (Lobo, 2021). Adequate physical infrastructure supports the R&D transfer into business growth opportunities (Putri et al., 2023). Related and in addition to such studies, specific research avenues have focused more specifically on two standpoints: R&D support factors and global perspectives.

On the standpoint of R&D support, government programs and subsidies can significantly boost private R&D investment and transference (Reshid et al., 2024). Some of these findings may be country-specific. For instance, Özçelik and Taymaz (2008) discovered that public R&D support programs affect private R&D investment in the Turkish manufacturing industry, with smaller firms benefiting more. Reshid et al. (2024) used firm-level data on Swedish companies and found that R&D subsidy had a positive and significant effect on turnover of scientific and technology workers. Selviaridis (2020) found that a UK-based small business research program had positive impact on capability transfer but a less clear effect on the transfer of innovation.

Also, organizational support such as gap-bridging instruments (e.g., funding, managerial support) and tech offices are key to successful technology commercialization, with effectiveness varying by size and resources (Tolin & Piccaluga, 2025; Turdalina & Kozhakhmet, 2025). Technology transfer offices can play a central role in managing IP, facilitating industry partnerships, and supporting spin-offs (Van Norman & Eisenkot, 2017; Markman et al., 2005; Z. Chen et al., 2024; Farjoo, 2024). Human capital, especially the competencies of technology transfer specialists, represents a bottleneck, particularly in emerging economies. Targeted training and professional development are recommended to bridge research and market application (Turdalina & Kozhakhmet, 2025; Huian et al., 2024; Sá & Pinho, 2019; Lobo, 2021). Entrepreneurs in digitally ready economies report greater success in channeling national R&D into innovative ventures, especially when combined with government entrepreneurship programs (Mahdi et al., 2021; Lobo, 2021). Government support can be especially important for green technology transfer (Adomako & Tran, 2024).

Using global perspectives, the literature concluded that developed countries emphasize patenting, licensing, spin-offs, and state-level policies to stimulate commercialization (Sitenko & Holienka, 2022), while emerging markets face challenges in infrastructure, funding, and policy but benefit from targeted government programs (Kenzhaliyev et al., 2021; Jjagwe et al., 2024). Advanced economies often have denser, more integrated policies for R&D transfer, while emerging economies focus on adapting policies to local needs and business absorptive capacities (Kergroach et al., 2018; Munari et al., 2016). Institutional quality is more important in emerging markets (Le et al., 2023). Institutional “distance” between countries can either facilitate or hinder international technology transfer, with effects varying by sector and firm type (Ellimaki et al., 2022). Across all countries, R&D commercialization failure significantly increases business closure risk, especially for smaller firms lacking resources (Yoo & Jung, 2024).

On the second theme—internal factors—several considerations have been addressed in the extant literature. Technological and marketing capabilities within the firm are crucial for leveraging R&D, especially in international markets (Davcik et al., 2020). The absorptive capacity of firms and the selection of external partners are crucial (Jha & Basu, 2024; V. Kumar et al., 2024). The entrepreneur’s characteristics (such as education, age, and prior experience) influence the likelihood of R&D investment and transfer, with younger, more educated, and experienced entrepreneurs more likely to pursue such initiatives (Zhang et al., 2022).

While many extant studies focused on examining certain factors affecting R&D transference, some comparative research has engaged with multi-factor conditions. This third theme in the literature advocated for improving the institutional context (such as establishing innovation spaces like incubators and accelerators), while at the same time striving for policy coherence and creating industry linkages (Kruger & Steyn, 2020). In the context of the COVID-19 pandemic, multi-factor studies suggested that the pandemic accelerated initiatives related to innovation, digitization, and digital transformation, often bypassing traditional R&D processes (de Lucas Ancillo & Gavrila, 2023). This work underscored the importance of redefining innovation frameworks to foster long-term sustainability and economic value in a post-pandemic world.

The COVID-19 pandemic has produced significant disruption to entrepreneurship (H. Kumar & Ramesha, 2024). The literature on this fourth theme has focused on three avenues of research: the general impact of government actions; the effectiveness of fiscal and monetary policies; and the effect of interventions on R&D transfer.

To begin with, entrepreneurs generally perceived government actions during the pandemic as both crucial and, at times, insufficient for supporting entrepreneurship. Entrepreneurs widely recognized the importance of government interventions (such as financial aid, tax waivers, and policy adjustments) in helping businesses survive the pandemic. However, many also expressed frustration over bureaucratic hurdles, slow implementation, and the inadequacy of support (Hussain et al., 2022; Nasar et al., 2021; Madeira et al., 2020; Giones et al., 2020; Galindo-Martín et al., 2021; Victor & Elangovan, 2022).

Moreover, government expansionary fiscal and monetary policies, in particular, had a positive (although sometimes limited) effect on entrepreneurship by providing liquidity and stimulating demand (Galindo-Martín et al., 2021). However, the effectiveness of these measures varied by country and sector, with some entrepreneurs feeling left behind, especially for companies in early development stages (Giones et al., 2020; Victor & Elangovan, 2022). The research identified the need for agile, innovative responses alongside policy measures (Nasar et al., 2021; Madeira et al., 2020; Cueto et al., 2022).

Lastly, the COVID-19 pandemic also influenced specifically the transfer of R&D. While immediate relief measures (e.g., loans, grants) helped address cash flow issues, these actions were often not tailored to the unique needs of innovative startups and R&D-intensive ventures, risking long-term innovation potential and lowering R&D transference (Kuckertz et al., 2020; Madeira et al., 2020; Belitski et al., 2021).

A collection of studies focused on the fifth theme of entrepreneurs’ perceptions. The prior literature in this area is meaningful to the current studies in two ways: primarily, it identifies perceptions on home country’s dynamics that affect entrepreneurship and R&D transfer; secondarily, it confirms the usefulness of GEM data for cross-country analyses.

Primarily, entrepreneurs perceive a range of home country conditions as either fostering or hindering entrepreneurship and R&D transfer. In general, entrepreneurs who perceive their home country as having reliable institutions and policies are more likely to start ventures (Gruenhagen, 2020; Yang et al., 2023; Muralidharan & Pathak, 2020). Business infrastructure and accessible markets are viewed positively, while excessive bureaucracy, unstable policies, and regulatory burdens are interpreted as major obstacles (Guerrero et al., 2020; Chowdhury & Audretsch, 2020; Nkongolo-Bakenda & Chrysostome, 2020). Social and cultural factors are also important. When entrepreneurship has high social legitimacy, it provides access to resources, networks, and community support (Capelleras et al., 2023; Guerrero et al., 2020). On the contrary, restrictive norms can inhibit entrepreneurship, especially for underrepresented groups (Pidduck et al., 2024). Strong local networks can facilitate resource acquisition. Entrepreneurs with enduring ties perceive greater opportunities (Lin et al., 2019; Duan et al., 2021; Nkongolo-Bakenda & Chrysostome, 2020). Some entrepreneurs may leverage networks to navigate challenges (Chowdhury & Audretsch, 2020; Nkongolo-Bakenda & Chrysostome, 2020). Specifically for R&D transfer, entrepreneurs perceive government support programs, favorable entry regulations, and targeted education or training as strong enablers (Sá & Pinho, 2019; Amorós et al., 2019; Lobo, 2021; Putri et al., 2023). However, if government support is seen as insufficient or untrustworthy, entrepreneurs may view it as a barrier (Gomes et al., 2022; Kuckertz et al., 2020).

Secondarily, several studies have leveraged GEM data to conduct cross-country analyses of entrepreneurial activity, attitudes, and determinants. These studies confirmed the theoretical soundness and practical utility of GEM variables for comparative entrepreneurship analysis (Martinez-Gonzalez et al., 2021, 2022; Bosma, 2013). With the use of GEM data, the present study connects the less-researched global phenomenon of how country factors overall are related to R&D transfer with the better-understood array of entrepreneurs’ perceptions of the specific drivers of R&D transfer at their locations.

Overall, this literature review highlights the importance of national context and cross-country analysis. The literature is broad in how it considers R&D transference, evaluating conditions, and drivers of R&D transference. This study is more focused on the extent of R&D transference, aiming to evaluate how it varies across countries. This research strives to understand the interplay of factors and to fill gaps in knowledge about how the overall macro level can explain varying levels of R&D transfer. The investigation is empirical, testing hypotheses about which factors explain variance in R&D transference. Each hypothesis is set in the context of small and medium-sized enterprises. Except for the country, all other aspects included in the hypotheses draw from entrepreneurs’ perceptions. Hypotheses are set as alternative hypotheses rather than null hypotheses.

The first set of hypotheses identify country and perceived entrepreneurship-supportive factors as potential variables related to R&D transference:

H1a: 

Country differences account for variance in the extent of R&D transference.

H1b: 

Perceived entrepreneurship-supportive factors explain R&D transference variation.

H1c: 

There is a statistically significant interaction between country and perceived entrepreneurship-supportive factors in explaining variance in R&D transference.

The second set of hypotheses refers to factors that entrepreneurs perceive as critical improvements:

H2a: 

Necessary entrepreneurial environment improvements perceived by entrepreneurs significantly explain variance in R&D transference.

H2b: 

There is significant interaction between country and perceived necessary entrepreneurial environment improvements in shaping R&D transference.

H2c: 

Perceived entrepreneurship-supportive factors and necessary improvements interact significantly to explain variance in R&D transference.

H2d: 

A three-way interaction among country, entrepreneurship-supportive factors, and necessary improvements is significantly associated with R&D transference.

The third set of hypotheses considers pandemic-related government actions:

H3a: 

Government actions perceived as negatively affecting entrepreneurship during the COVID-19 pandemic significantly explain variance in R&D transference.

H3b: 

Government actions perceived as positively affecting entrepreneurship during the COVID-19 pandemic significantly explain variance in R&D transference.

H3c: 

There is significant interaction between country and negatively perceived government actions in explaining variance in R&D transference.

H3d: 

There is significant interaction between country and positively perceived government actions in explaining variance in R&D transference.

H3e: 

Negatively and positively perceived government actions during the COVID-19 pandemic interact significantly when associated with R&D transference.

H3f: 

A three-way interaction among country, negatively perceived government actions, and positively perceived government actions significantly explains variance in R&D transference.

3. Methodology

This research paper uses a GEM dataset. GEM gathers and creates the NES Global Individual Level Nations dataset through a standardized, multi-stage process involving expert surveys and harmonized methodologies across participating countries. The selection of countries included in the analysis is based on data availability. No countries were eliminated for the purpose of this study. The dataset includes countries that have established a national GEM team willing to conduct the survey according to GEM’s standardized methodology. There is no fixed set of eligibility criteria; instead, participation is determined by a country’s interest, ability to form a national team, and capacity to follow GEM protocols for data collection and expert recruitment (Levie, 2015). The actual set of countries in any given year depends on which national teams are active and able to meet GEM’s requirements (Levie, 2015).

The NES is designed to assess the entrepreneurial framework conditions in each country. It involves recruiting a panel of national experts across diverse fields such as finance, government, education, and business. These experts complete a detailed questionnaire evaluating country factors like access to finance, government policy, education, R&D transfer, and cultural norms. The NES uses a standardized questionnaire to ensure comparability across countries and years. The survey includes both quantitative (Likert-scale) and qualitative (open-ended) questions, focusing on the perceived strengths and weaknesses of the national entrepreneurial ecosystem. Experts are selected to represent a broad cross-section of relevant sectors. GEM applies strict protocols for expert selection, data validation, and harmonization to maintain data reliability and cross-national comparability (Reynolds et al., 2005; Reynolds, 2017; Rietveld & Patel, 2022).

Individual expert responses are anonymized and aggregated at the country level, but the dataset retains individual-level responses for micro-level analysis (Reynolds et al., 2005; Reynolds, 2017). The harmonization process means that GEM employs uniform definitions, coding, and data-cleaning procedures to ensure that data from different countries are directly comparable (Reynolds et al., 2005; Reynolds, 2017; Acs et al., 2008). After internal validation, the dataset is made publicly available for research, typically with a lag to ensure data quality and confidentiality (Reynolds, 2017). The present research uses the GEM 2021 NES Global Individual Level Nations dataset. This was the most recent data available from GEM at the time of the analysis. While the data from the 2021 GEM NES predates current global conditions, it remains highly relevant for analysis. Data from 2021 provides a critical snapshot of entrepreneurial ecosystems during a pivotal moment in global history, the COVID-19 pandemic. This allows the researcher to analyze how entrepreneurial dynamics respond to crisis conditions, offering valuable insights into perceptions of resilience, adaptability, and policy effectiveness. Such historical benchmarks are essential for understanding the trajectory of entrepreneurial ecosystems. The timing of the data is also crucial for testing the significance of government interventions. These findings support the inclusion of government interventions perceptions and make the determination that they enhance the differentiation between countries. Thus, 2021 data provide an opportunity to draw important conclusions on systemic interventions, suggesting that such interventions may amplify existing disparities rather than directly influence R&D outcomes.

The GEM survey’s standardized approach enables cross-country comparisons that remain valid even as conditions evolve. The robustness of the methodology ensures that findings are not merely time-bound but contribute to broader theoretical and empirical discussions. The 2021 data reflect foundational conditions and early responses after the COVID-19 pandemic that may shape current outcomes. Understanding these roots is crucial for evaluating the effectiveness of newer initiatives. The study based on pandemic-era data can be seen as complementary to more recent studies. It provides a baseline against which newer data can be compared, helping to identify shifts, continuities, and emerging patterns in entrepreneurship and policy environments.

It should be noted that the NES component of GEM also has some constraints, as it may suffer from low interrater reliability. Experts’ subjective evaluations vary widely, making cross-country and even within-country (longitudinal) comparisons potentially imprecise. This may undermine the validity of country data and broad policy recommendations based on these data (Rietveld & Patel, 2022). Previous studies noted that the set of indicators presented in the NES data may not fully capture all relevant demographic, economic, and institutional determinants, limiting the generalizability of findings across diverse national contexts (Teixeira et al., 2018). In addition, differences in how entrepreneurship is defined, measured, and perceived across countries can lead to inconsistent or even contradictory findings, complicating efforts to draw global conclusions (Acs et al., 2008). Despite its limitations, NES provides practical and unique data on entrepreneurs’ perceptions, thus making it a meaningful source for this study.

The data used for this investigation are at the individual level, across all variables, with no nested or aggregated data. Data represent responses from individual experts across participating countries. Each expert provides assessments of the national conditions that influence entrepreneurship, based on their professional experience and knowledge. Each individual response is associated with a country identifier. Frequency analysis is used to examine the distribution of responses across categories for each variable. The average frequency across all categories can be used to evaluate the spread of responses. The mode is a key descriptive statistic of the data. The most frequent categories indicate dominant perceptions among entrepreneurs. Reliability and internal consistency of the analysis are supported by the fact that GEM data undergo harmonization to ensure comparability across countries and years. This process includes standardizing items and response formats, which supports the reliability of cross-national measures (Reynolds et al., 2005). Harmonization helps reduce measurement bias and enhances the validity and reliability of the dataset (Tristan-Lopez, 2025; Reynolds et al., 2005). The GEM NES data explicitly reports internal consistency using Cronbach’s alpha, supporting strong internal consistency across constructs (Rietveld & Patel, 2022).

For the purpose of this study, the following are interpreted as country conditions: the first factor fostering entrepreneurship (the GEM NES variable NES_F1 as the foremost entrepreneur’s choice of what supports entrepreneurship, across categories varying from policies to economy and digitization) and first recommendation category that the entrepreneur made regarding needed improvements in the entrepreneurial environment (the GEM NES variable NES_R1, varying on the same categories). The country or country location represents the country itself in light of its characteristics, such as political, social, and entrepreneurial (the GEM NES variable NES_COUNTRY as a country identifier). Thus, country conditions refer to primary factors and primary recommendations identified by entrepreneurs in relation to their respective country’s entrepreneurship environment. In contrast, the term country location simply denotes the nation-level entity as the designated country identifier in the dataset.

Factorial ANOVA is deployed to analyze the effects of these independent variables on R&D transference simultaneously. By deploying factorial ANOVA, it is possible to examine not only the individual (main) effect of each factor but also the interaction effect. IBM’s SPSS version 30 statistical software suite was used for analysis. A factorial ANOVA in SPSS is implemented as a special case of the univariate General Linear Model (GLM). The software computes parameter estimates, F-tests, and post hoc comparisons (Davis, 2008; Popović, 2015; Leech et al., 2014). The output includes tests for main effects, interactions, and error, all derived from the GLM.

Three factorial ANOVA models were used to examine how various factors explain the variance in R&D transference, defined as the extent to which R&D leads to new commercial opportunities accessible to small and medium-sized enterprises. The data on all variables draws from entrepreneurs’ perceptions. The definition and operationalization of each variable was provided by the GEM’s original methodology. The dependent variable is the same for every model. The data represents standardized average scores for the R&D level of transference, evaluated by the source of data as the extent to which R&D will lead to new commercial opportunities and is available to small and medium-sized enterprises. According to GEM methodology, the original data on entrepreneurship was collected on a 1 to 9 scale, from which a principal components procedure was applied to identify key aspects related to entrepreneurship. R&D transference was identified as such an aspect. Individual scores are reported on the aspects reflecting R&D transference. Corresponding standardized scores are also calculated.

Standardized scores are used as statistical measures that express the relative position of an individual observation within a distribution. Specifically, a standardized score represents the number of standard deviations by which a data point deviates from the mean of the distribution. GEM converts raw scores into this dimensionless metric to provide information on the relative entrepreneurial framework conditions across countries. The bound nature of standardized scores enhances interpretability. Moreover, the statistical properties of standardized scores enable efficient cross-country comparisons and facilitate the identification of patterns and outliers within the data. The R&D transference data includes a wide range of scores, with no evident outliers. No pattern is observed within or across countries, aside from the overall distribution.

Model 1 tests if country and first factor fostering entrepreneurship explain variance in the R&D level of transference. The country independent variable (NES_COUNTRY) is a nominal variable, categorizing 50 countries across the Americas, Europe, Middle East and Asia. The first factor fostering entrepreneurship (NES_F1) represents the foremost entrepreneur’s choice of what supports entrepreneurship and is a nominal variable with categories of financial support; government policies; government programs; education and training; transmission of knowledge; commercial and professional infrastructure; market openness; access to physical infrastructure; cultural and social norms; capacity for entrepreneurship; economic climate; work force features; perceived population composition; political, institutional and social context; economic crisis; corruption; differences in performance of small, medium and large companies; internationalization; labor costs; information access; ICT/digitization; and COVID-19 pandemic protocols.

Model 1 is structured as follows:

$${ZNES}_{ijk}=\mu +{\alpha }_i+{\beta }_j+{\left(\alpha \beta \right)}_{ij}+{\epsilon }_{ijk}$$

where

${ZNES}_{ijk}$ is the standardized average score of the R&D level of transference for the _k observation in the _i-th level of the NES_COUNTRY factor and the _j-th level of the NES_F1 factor,

$\mu$ is the overall mean,

${\alpha }_i$ is the effect of the _i-th level of the NES_COUNTRY factor,

${\beta }_j$ is the effect of the _j-th level of the NES_F1 factor,

${\left(\alpha \beta \right)}_{ij}$ is the interaction effect between the NES_COUNTRY factor and the NES_F1 factor,

${\epsilon }_{ijk}$ is the random error.

Model 2 evaluates the effect of country, the first factor fostering entrepreneurship, and the first recommendation category that entrepreneurs made regarding improving the entrepreneurial environment. The additional factor (NES_R1) included in this model denotes a nominal variable with categories from coding entrepreneurs’ open responses as best matches to the same classifications as the first factor variable (NES_F1).

Model 2 is defined as follows:

$${ZNES}_{ijkl}=\mu +{\alpha }_i+{\beta }_j+{\gamma }_k+{\left(\alpha \beta \right)}_{ij}+{\left(\alpha \gamma \right)}_{ik}+{\left(\beta \gamma \right)}_{jk}+{\left(\alpha \beta \gamma \right)}_{ijk}+{\epsilon }_{ijkl}$$

where

${ZNES}_{ijk}$ is the standardized average score of the R&D level of transference for the _k observation in the _i-th level of the NES_COUNTRY factor, the _j-th level of the NES_F1 factor, and the _k-level of the NES_R1 factor,

$\mu$ is the overall mean,

${\alpha }_i$ is the effect of the _i-th level of the NES_COUNTRY factor,

${\beta }_j$ is the effect of the _j-th level of the NES_F1 factor,

${\gamma }_k$ is the effect of the _k-th level of the NES_R1 factor,

${\left(\alpha \beta \right)}_{ij}$ is the interaction effect between the NES_COUNTRY factor and the NES_F1 factor,

${\left(\alpha \gamma \right)}_{ik}$ is the interaction effect between the NES_COUNTRY factor and the NES_R1 factor,

${\left(\beta \gamma \right)}_{jk}$ is the interaction effect between the NES_F1 factor and the NES_R1 factor,

${\left(\alpha \beta \gamma \right)}_{ijk}$ is the interaction effect between the NES_COUNTRY factor, NES_F1 factor and the NES_R1 factor,

${\epsilon }_{ijk}$ is the random error.

Model 3 tests the effect of the country and two other factors on the R&D level of transference. The two new dependent variables for this model are the first government action that has negatively affected entrepreneurship in the country since the start of the pandemic (NES_GN1) and the first government action that has positively affected entrepreneurship in the country since the start of the pandemic (NES_GP1). The government actions factors represent two nominal variables with categories from coding entrepreneurs’ open responses as best matches to the same classifications as the first factor variable (NES_F1), with additional categories denoting government subsidies for employment preservation and wages, credit moratorium, deferment of tax liabilities, loan extensions; lockdown, COVID-19 protocols, restrictions on public gatherings, travels, borders, and lack thereof; unpredictable restrictions, shifting policies with short notice, communication; support measures in general, government COVID-19 programs, support to digitalization; ineffective measures, time to do administrative work, public institutions closed, unequal treatment, customs, tariffs, taxes, permits increases; no access to the COVID-19 vaccines; and Brexit (for UK). This model evaluates if pandemic policies explain variance in R&D transference. The data used for this study corresponds to the time when the COVID-19 pandemic affected business. Testing the consequences of pandemic-related policies contributes to the understanding of R&D transfer during a disruptive period.

Model 3 is expressed as follows:

$${ZNES}_{ijkl}=\mu +{\alpha }_i+{\delta }_j+{\phi }_k+{\left(\alpha \delta \right)}_{ij}+{\left(\alpha \phi \right)}_{ik}+{\left(\delta \phi \right)}_{jk}+{\left(\alpha \delta \phi \right)}_{ijk}+{\epsilon }_{ijkl}$$

where

${ZNES}_{ijk}$ is the standardized average score of the R&D level of transference for the _k observation in the _i-th level of the NES_COUNTRY factor, the _j-th level of the NES_F1 factor, and the _k-level of the NES_R1 factor,

$\mu$ is the overall mean,

${\alpha }_i$ is the effect of the _i-th level of the NES_COUNTRY factor,

${\delta }_j$ is the effect of the _j-th level of the NES_GN1 factor,

${\phi }_k$ is the effect of the _k-th level of the NES_GP1 factor,

${\left(\alpha \delta \right)}_{ij}$ is the interaction effect between the NES_COUNTRY factor and the NES_GN1 factor,

${\left(\alpha \phi \right)}_{ik}$ is the interaction effect between the NES_COUNTRY factor and the NES_GP1 factor,

${\left(\delta \phi \right)}_{jk}$ is the interaction effect between the NES_GN1 factor and the NES_GP1 factor,

${\left(\alpha \delta \phi \right)}_{ijk}$ is the interaction effect between the NES_COUNTRY factor, NES_GN1 factor and the NES_GP1 factor,

${\epsilon }_{ijk}$ is the random error.

The application of factorial ANOVA provides a robust framework for analyzing multifaceted influences on R&D transference from the perspective of small and medium-sized enterprises. By leveraging the General Linear Model structure, the analysis captures both main and interaction effects across a diverse set of categorical variables derived from entrepreneurs’ perceptions, as operationalized by the GEM methodology. The three models progressively incorporate additional contextual factors (from entrepreneurial support mechanisms to pandemic-related government actions), allowing for a nuanced understanding of how national environments and policy responses shape R&D transference.

This methodological approach validates the relevance of country-specific and thematic factors for R&D transference and highlights the complexity of their interdependence. The inclusion of pandemic-era variables in Model 3 underscores the importance of adaptive policymaking in sustaining innovation during periods of disruption. Overall, the factorial ANOVA models offer a comprehensive lens through which to evaluate the conditions under which R&D can be effectively translated into entrepreneurial opportunity, providing valuable insights for both researchers and policymakers aiming to strengthen innovation ecosystems. The total number of valid observations across all variables was 1159, accounting for some missing data. The missing data is absent from the original dataset. Since the data cannot be replicated by the researcher, the missing observations could not be included in the analysis.

4. Results

Analysis of data reveals several important characteristics. Data representation across countries varies somewhat but is widespread between least and most frequent countries. The variability is moderate. Most countries cluster in representation around the mid-point range, with around 20–30 observations each. Although Morocco has the strongest representation, other countries are close to this representation. Several countries are close to each other in terms of more limited representation. The dataset demonstrates a sufficiently broad and balanced representation across countries, which supports its suitability for statistical analysis. While there is some variability in the number of observations per country, this variability is moderate and does not compromise analytical integrity. Several countries have relatively high levels of data presence; however, the difference is not extreme. Additionally, countries with lower representation tend to be grouped closely together, minimizing the risk of outlier-driven distortion. This distribution pattern provides a reasonably equitable basis for cross-national statistical exploration.

Table 1. Observed frequencies of entrepreneurship-related variables.

	NES_COUNTRY	NES_F1	NES_R1	NES_GN1	NES_GP1
Most frequent category (mode)	Morocco	Government policies	Government policies	Government policies	Financial support
Average frequency across all categories 1	2.0%	4.0%	4.3%	3.4%	3.7%
Category/ies with lowest frequency	Norway (1.3%)	Corruption; COVID-19 protocols (0.0%)	Corruption; ICT/Digitization (0.2%)	Perceived population composition (0.0%)	Perceived population composition (0.0%)
Category/ies with highest frequency	Morocco (3.7%)	Government policies (8.7%)	Government policies (22.4%)	Government policies (16.5%)	Financial support (13.5%)

¹ Average percentage of representation across all categories of the variable.

provides a synthesis of the data across the variables included in the statistical models. Average frequency across all categories shows that the data is relatively balanced across the different categories for each of the variables. Frequency analysis identifies government policies and financial support as important aspects perceived by entrepreneurs to affect entrepreneurship.

The predominant fostering factor identified across countries is government policies. This reflects widespread perception that the regulatory and policy environment plays a central role in shaping entrepreneurial activity. In alignment with the fostering factor, the most frequent recommendation for improving the entrepreneurial ecosystem was also government policies. This suggests a global consensus that reforms or enhancements in policy frameworks are essential for fostering a more supportive environment for entrepreneurs. The data on government actions during the COVID-19 pandemic highlights restrictive government policies, likely tied to lockdowns and unpredictable regulations, as having negatively impacted entrepreneurship. Conversely, financial support measures were recognized as beneficial. The data underscores the pivotal role of government policy in both fostering and sustaining entrepreneurship. Countries aiming to strengthen their entrepreneurial ecosystems may benefit from reviewing and refining their policy approaches, especially in light of lessons learned during the pandemic.

Before applying statistical modeling, the Kolmogorov–Smirnov (K-S) test method was used to check the normality of the dependent variable within each group combination. This technique is commonly used to test for normality, especially for relatively large datasets (Steinskog et al., 2007; Mendes & Pala, 2003). As a caution, for large samples, the K-S test may be overly sensitive (Steinskog et al., 2007; Mendes & Pala, 2003). However, the test seemed to have worked well for the data in the current investigation. Conforming with the literature (Habibzadeh, 2024; Hanusz & Tarasinska, 2015; Mendes & Pala, 2003), the normality hypotheses were set below. Failure to reject the null hypothesis (i.e., high significance levels for the K-S statistic) means that data are considered normally distributed. The results are presented in

Table 2. Test of normality of the standardized scores of R&D level of transference within each group combination: summary results.

	Model 1		Model 2		Model 3
	Kolmogorov–Smirnov Statistic across Groups	Group Combination with Low Significance Level *	Kolmogorov–Smirnov Statistic across Groups	Group Combination with Low Significance Level	Kolmogorov–Smirnov Statistic across Groups	Group Combination with Low Significance Level
NES_ COUNTRY	Ranging from 0.081 to 0.203	None	Ranging from 0.081 to 0.254	None	Ranging from 0.080 to 0.269	Germany
NES_F1	Ranging from 0.064 to 0.230	None	Ranging from 0.067 to 0.244	None
NES_R1			Ranging from 0.044 to 0.410	Government policies
NES_GN1					Ranging from 0.062 to 0.333	None
NES_GP1					Ranging from 0.076 to 0.288	None

* p < 0.001.

. The hypotheses were as follows:

Null Hypothesis (H₀): The dependent variable within each group combination is normally distributed.

Alternative Hypothesis (H₁): The dependent variable within each group combination is not normally distributed.

The Kolmogorov–Smirnov test assessed whether the dependent variable follows a normal distribution within each combination of independent variable groups. For most group combinations, the K-S test failed to reject the null hypothesis. There is no strong evidence against normality in those groups. In two specific group combinations, the test rejected the null hypothesis, suggesting that the dependent variable may not be normally distributed in those cases.

For the groups where normality was confirmed, the dependent variable behaves in a way that is consistent with a normal distribution. This supports the use of the factorial ANOVA statistical method within these groups. The deviations from normality in two specific groups suggest that the assumption of normality may not hold universally across all subgroups. However, the problem seems to be localized. Also, the severity of the problem may be mitigated by the sample size. Parametric tests are generally robust to moderate non-normality due to the Central Limit Theorem, so the impact may be limited (Knief & Forstmeier, 2018; Sainani, 2012; Bartlett, 1935).

Table 3. Model-based significance testing: ANOVA summary results.

	Model 1		Model 2		Model 3
Test of equality of error variances: Levene Statistic 1	1.107		1.616 *		1.054
White test for heteroskedasticity: Chi-Square	556.415		1029.479		458.056
Tests of Between-Subjects Effects
	F Statistic	Partial Eta Squared	F Statistic	Partial Eta Squared	F Statistic	Partial Eta Squared
NES_COUNTRY	7.721 *	0.307	5.317 *	0.472	3.573 *	0.299
NES_F1	1.192	0.030	1.316	0.087
(NES_COUNTRY) (NES_F1)	1.085	0.364	1.984	0.569
NES_R1			0.703	0.046
(NES_COUNTRY) (NES_R1)			1.056	0.465
(NES_F1) (NES_R1)			1.019	0.296
(NES_COUNTRY) (NES_F1) (NES_R1)			0.981	0.208
NES_GN1					1.092	0.062
NES_GP1					0.701	0.039
(NES_COUNTRY) (NES_GN1)					1.058	0.258
(NES_COUNTRY) (NES_GP1)					0.837	0.221
(NES_GN1) (NES_GP1)					0.669	0.109
(NES_COUNTRY) (NES_GN1) (NES_GP1)					0.501	0.014

* p < 0.001. ¹ Based on median.

shows statistical results of the factorial ANOVA models. The reliability of Model 2 is uncertain due to the violation of error variance assumptions, suggesting caution in interpreting its results. A key result of the statistical tests is the country-level significance. The country variable consistently explains a significant portion of variance in R&D transference (approximately 30%), underscoring the importance of national context in shaping entrepreneurial outcomes.

Levene’s test of equality of error variances based on the Levene statistic indicates we can proceed with the analysis with some confidence for Models 1 and 3; however, for Model 2, the Levene statistic is statistically significant, indicating that this model is less reliable as the assumption of equality of error variances may not be met.

The White test for heteroskedasticity across the three models shows that the Chi-Squared test is not statistically significant, indicating that heteroskedasticity is not present, so the assumption of equal variances across the groups is not violated.

The test of between-subjects effects based on the F statistic finds that for all three models, the country factor (NES_COUNTRY) is the only one that affects the R&D transfer dependent variable. The estimate of effect size is measured by the Partial Eta Squared.

The purpose of Model 1 is to assess if country and the primary factor fostering entrepreneurship explain variance in R&D transference. The results indicate that country-level differences are significant, accounting for approximately 30% of the variance. The fostering factor and its interaction with country are not statistically significant.

The purpose of Model 2 is to evaluate the individual and combined effects of country, fostering factor, and entrepreneurs’ recommendations for improving the entrepreneurial environment. Country remains the dominant factor, explaining nearly 47% of the variance. Other variables and interactions do not significantly explain variance in the dependent variable. However, the model’s reliability is compromised due to unequal variances. The violation of the homogeneity of variance assumption means that the p-values and confidence intervals from the standard ANOVA may not be trustworthy for Model 2. The results from Model 2 should therefore be interpreted with caution, as the unequal variances may bias the test statistics. Effect sizes (i.e., Partial Eta Squared) can provide additional context. However, overall interpretations and post hoc comparisons, like Tukey’s HSD, are likely affected (Sauder & DeMars, 2019).

Model 3 aims to analyze pandemic-related government actions. Country-level differences again emerge as the only significant factor, explaining nearly 30% of the variance. Pandemic-related government actions, while relevant in context, do not explain significant variance in R&D transference. Similarly to results of the first and second models, no interaction effects are found to be significant. No independent variable depends on the level of another, either synergistically or antagonistically. There is no indication of moderation.

When taking into consideration the original data, the policy perception cannot be ignored. Entrepreneurs widely perceive government policies as both a key fostering factor and a top recommendation for improvement, suggesting a global consensus on the need for policy reform to support entrepreneurship. However, the tests of the various factors (including policies and infrastructure) traditionally associated with R&D transference do not conclude on their significance. This may suggest that, to support innovation, the intervention of governments should be made directly on the processes and opportunities for R&D transference. For entrepreneurs, these findings suggest that what they perceive as a positive entrepreneurial environment may not be sufficient for innovation.

Similarly, the original data and its descriptives find that restrictive and unpredictable government actions during COVID-19 hindered entrepreneurship. Financial support measures were beneficial, highlighting the importance of responsive and supportive government interventions during crises. However, as shown by causal models’ results, none of these pandemic-related intercessions made a difference specifically on R&D transfer.

An additional examination was conducted to assess both the direction and magnitude of the observed effects. Since the only independent variable with a significant effect on the dependent variable is country, the effect is only detailed for this variable.

Table 4. Parameter estimates for the NES_COUNTRY variable: selected results.

Model 1		Model 2		Model 3
Country	Parameter Estimate	Country	Parameter Estimate	Country	Parameter Estimate
Netherlands	3.087 **	France	3.064 **	USA	17.023 **
Latvia	1.797 **			Russia	10.134 **
Croatia	2.694 **			Greece	9.829 **
United Arab Emirates	3.591 **			Netherlands	10.340 **
Qatar	3.282 *			Spain	6.066 **
				Switzerland	9.746 **
				Germany	4.515 **
				Colombia	18.429 *
				Morocco	4.830 **
				Sudan	19.908 **
				Luxembourg	9.306 **
				Latvia	5.058 **
				Slovenia	5.371 **
				Guatemala	6.596 **
				Uruguay	8.579 **
				Saudi Arabia	5.634 **
				Oman	15.467 **
				Israel	10.266 **
				Qatar	7.530 **

* p < 0.001. ** p < 0.005.

shows parameter estimates. Only estimates with high confidence levels are shown with the corresponding country. In Model 2, Levene’s test was statistically significant, indicating a violation of the homogeneity of variance assumption required for classical ANOVA. This suggests that the risk of Type I error inflation or reduced power is higher for this model. Model 2 findings are therefore not fully substantiated.

Model 1 has a smaller set of countries with significant estimated effect. The parameter estimates show moderate effects. Model 2 has only one country with significant effect. Model 3 includes a broader range of countries with larger estimates and a number of countries with moderate effects. Model 3 might allow for greater differentiation between countries. This implies that the inclusion of government interventions strengthens the explanatory power of the country in association to R&D transference.

Tukey’s HSD was used to explore pairwise comparisons between group means. Tukey’s HSD (Honestly Significant Difference) test provides a post hoc analysis for a significant factorial ANOVA result (country as a main effect on R&D transference) to determine which specific group means differ from each other. The test controls for Type I error rate across all comparisons. This helps identify precisely which groups are different from each other, a detail that the omnibus ANOVA test does not provide. A low significance level indicates that the difference between those two specific group means is considered statistically significant.

Table 5. Post hoc analysis using Tukey’s HSD test: selected results.

Model 1			Model 2			Model 3
Group 1	Group 2	Mean Difference	Group 1	Group 2	Mean Difference	Group 1	Group 2	Mean Difference
Morocco	USA	−0.980 **	Morocco	USA	−0.979 **	USA	Russia	0.572 **
Dominica	USA	−1.135 **	Dominica	USA	−1.152 **	Russia	USA	−0.572 **
USA	Russia	0.765 **	USA	Russia	0.798 **	Greece	USA	0.611 **
Russia	USA	−0.765 **	Russia	USA	−0.798 **	Netherlands	USA	0.908 **
Egypt	USA	−0.590 **	Egypt	USA	−0.569 **	Spain	USA	0.781 **
Netherlands	USA	0.666 **	Netherlands	USA	0.727 **	Romania	USA	−0.813 **
Spain	USA	0.665 **	Spain	USA	0.665 **	Switzerland	USA	1.049 **
Romania	USA	−0.832 **	Romania	USA	−0.832 **	Norway	USA	0.802 **
Switzerland	USA	0.862 **	Switzerland	USA	0.837 **	Brazil	USA	−0.719 **
Norway	USA	0.635 **	Norway	USA	0.609 **	Iran	USA	−0.743 **
Mexico	USA	−0.567 **	Mexico	USA	−0.567 **	Morocco	USA	−0.933 *
Brazil	USA	−0.856 *	Brazil	USA	−0.853 *	Sudan	USA	−0.688 **
Iran	USA	−0.925 *	Chile	USA	−0.545 **	Finland	USA	0.820 **
Morocco	USA	−0.980 *	Iran	USA	−0.885 **	Lithuania	USA	0.863 **
Sudan	USA	−0.960 *	Morocco	USA	−0.979 *	Belarus	USA	−0.834 **
Finland	USA	0.758 **	Sudan	USA	−0.941 *	Guatemala	USA	−0.842 **
Lithuania	USA	0.859 **	Finland	USA	0.707 **	Panama	USA	−0.517 **
Belarus	USA	−0.971 *	Lithuania	USA	0.818 **	Kazakhstan	USA	−0.612 **
Guatemala	USA	−0.896 *	Belarus	USA	−0.971 *	Dominica	USA	−1.014 *
Panama	USA	−0.701	Guatemala	USA	−0.896 *	Jamaica	USA	−0.594 **
Kazakhstan	USA	−0.702 **	Panama	USA	−0.702 **
Saudi Arabia	USA	0.575 **	Kazakhstan	USA	−0.716 **
Oman	USA	−0.589 **	Saudi Arabia	USA	0.536 **
United Arab Emirates	USA	0.695 **	Oman	USA	−0.574 **
Dominica	USA	−1.135 *	Dominica	USA	−1.152 *
Jamaica	USA	−0.707 **	Jamaica	USA	−0.681 **
Morocco	USA	−0.980 **	Morocco	USA	−0.979 **
Dominica	USA	−1.135 **	Dominica	USA	−1.152 **

* p < 0.001. ** p < 0.005.

shows Tukey’s HSD results that show some consistent differences across all models. USA vs. Russia, Morocco, Dominica, Netherlands, Spain, Romania, Switzerland, and Norway are country pairs that consistently show significant differences in R&D transference scores across all three models. This suggests a robust and persistent disparity in how R&D is transferred when comparing the USA and these countries. Some country pairs are only significant in the third model (USA vs. Iran and Greece vs. USA), indicating that this model may be more sensitive to certain country-level differences or includes variables that better capture these disparities. This result highlights the importance of government interventions. Pair significance not present in the third model (USA vs. Egypt) but significant in the first two models suggests that the third model may account for factors that reduce the perceived difference in R&D transference between these countries. A pair significance (USA vs. Brazil) in the first and third model but not the second model indicates possible model-specific sensitivity or data variation.

For Model 1, the mean differences identify significant differences between the USA and some developing nations. Countries like Dominica, Morocco, Egypt, and Romania show higher R&D transference scores than the USA. This may suggest that these countries are more actively engaging in knowledge transfer or innovation diffusion. A conjecture could be that this is possibly due to targeted policies or international collaborations. When comparing the USA to European nations, the USA shows higher scores than Russia, Netherlands, Spain, Switzerland, and Norway, indicating stronger R&D transference in comparison to these developed economies. The USA vs. Russia comparison is significant in both directions, reinforcing the robustness of the observed difference. For Model 2, the significant comparisons closely mirror those in Model 1, suggesting that the underlying patterns are stable across models. However, a significant Levene’s test signaled that Model 2’s results may be unreliable due to unequal variances. For Model 3, Dominica, Morocco, Romania, Brazil, and Iran all show higher R&D transference scores than the USA. The USA has higher scores than Switzerland, Netherlands, Spain, and Norway, indicating a stronger level of R&D transference compared to these developed economies. Greece vs. USA is significant only in this model, suggesting that Model 3 may capture nuances in Greece’s R&D transference that earlier models missed. This model reveals a more nuanced landscape, with both developing and developed countries showing significant differences with the USA, with some outperforming it and others lagging behind.

The analysis provides conclusions on the hypotheses. In the first set of hypotheses, only the first one (H1a) is supported. Model 1 shows that country differences account for approximately 30% of the variance in R&D transference. This effect is consistent across models, with Tukey’s HSD test confirming significant pairwise differences (e.g., USA vs. Russia, Morocco, Dominica, etc.). The entrepreneurship fostering factor and the interaction between country and the fostering factor was not statistically significant in Model 1. None of the hypotheses in the second and third set are supported by the statistical findings. Additionally, Model 2’s reliability is compromised due to violations of homogeneity of variance, which undermines confidence in its results.

Hypotheses testing concludes that country differences are the only consistently significant factor across all models. Entrepreneurial perceptions (supportive factors, necessary improvements, and government actions) do not significantly explain variance in R&D transference. Model 3 provides greater differentiation between countries, suggesting that including government interventions may enhance the explanatory power of country-level effects. Tukey’s HSD reveals consistent country pair differences, especially between the USA and several other nations, reinforcing the robustness of country-level effects.

Based on the statistical findings, certain variables failed to show significance in the models employed for analysis. The perceived entrepreneurship fostering factors may not have a direct or strong enough association with R&D transference to be statistically detectable. These factors may be more distal or only indirect contributors, mediated through other variables like policy or infrastructure. The interaction between country and fostering factors also failed to reach significance, suggesting that the fostering factors may not vary meaningfully across countries, or that the sample size was insufficient to detect such nuanced effects. Entrepreneurs’ recommendations are subjective and may not translate directly into measurable outcomes like level of R&D transference. Their lack of significance could reflect a disconnect between perception and actual systemic circumstances. Notably, pandemic-related government actions are not found statistically significant for R&D transference. While intuitively relevant, these actions may not have a uniform or measurable explanatory effect. The actions may be contextually important but not statistically significant.

Overall, several strategic recommendations emergence from the analysis. They revolve around policy frameworks, financial support, and lessons from the pandemic-era policy successes and failures. Governments aiming to enhance R&D transference in their countries should act directly on the mechanisms of innovation and the ease of transfer, rather than creating broad frameworks to support entrepreneurial ecosystems. Global crisis such as the COVID-19 pandemic disruption affects the entrepreneurs’ perspectives on the ease of doing business. However, these aspects may not impede their ability to commercialize innovations. While policies and financial support affect entrepreneurial activity, governments need to also focus on innovation explicitly and unambiguously to support this path of growth for small businesses.

Entrepreneurs view government policy and financial support as important fundamentals for their businesses. However, if innovation is a key strategy, they should seek out specific R&D transference opportunities and observe the locations where such opportunities are abundant. Partnering or working with other companies in such locations may be key for their success.

5. Discussion

The findings of this study emphasize the pivotal role of the country-level context in shaping R&D transference for small and medium-sized businesses. This result aligns with a substantial body of literature emphasizing the importance of national circumstances, including institutional quality, government support, and digital infrastructure, for entrepreneurial outcomes and innovation diffusion (Alam et al., 2019; Mahdi et al., 2021; Guerrero et al., 2020).

Two variables have limited significance: entrepreneurship fostering factors and entrepreneurs’ recommendations for improving the entrepreneurial environment. This implies that macro-level influences may overshadow micro-level determinants in cross-national analyses. While prior studies have highlighted the significance of internal capabilities and entrepreneur situations (Kou et al., 2020; Zhang et al., 2022), the current findings imply that these factors may not exert a uniform influence across diverse national settings. This discrepancy points to the need for more nuanced investigations into how internal and external factors interact within specific country contexts.

Model 3, which integrates government interventions related to the pandemic, does not demonstrate statistically significant explanatory power for these variables. However, their inclusion enhanced the differentiation between countries, suggesting that such interventions may amplify existing disparities rather than directly influence R&D outcomes. This observation resonates with the literature, indicating that while pandemic-related policies were essential for business survival, they often lacked specificity for R&D-intensive ventures and failed to support long-term innovation (Kuckertz et al., 2020; Belitski et al., 2021).

The comparative analysis of country pairs further illustrates the complex landscape of R&D transference. Notably, countries such as Dominica, Morocco, and Romania consistently outperformed the USA in R&D transference, while developed economies like Switzerland, Netherlands, and Norway trailed. These findings suggest that targeted policies and international collaborations may be particularly effective in enhancing R&D transfer in developing nations (Ngo, 2020; Song et al., 2017). Conversely, structural or policy-related barriers may hinder R&D performance in some advanced economies despite their technological capabilities. More research is needed to confirm such phenomena.

Overall, this work adds to the academic literature by demonstrating a strong empirical link between the national macro level and the R&D transference. It highlights the need for future research to delve deeper into the specific economic, cultural, and policy-related characteristics that drive these national differences. Such investigations could inform more effective strategies for fostering R&D transfer, tailored to the unique contexts of individual countries. This approach aligns with calls in the literature to expand the understanding of institutional mechanisms and innovation ecosystems that ultimately support more resilient and inclusive entrepreneurial development (Kruger & Steyn, 2020; de Lucas Ancillo & Gavrila, 2023).

While previous work emphasized the acceleration of innovation and digitization during the COVID-19 pandemic, this new study takes a structural and comparative approach. It finds that pandemic-related policies or entrepreneurship-supportive conditions are not consistently associated with R&D transference. This reframing suggests that macro-level conditions (e.g., regulatory environments, cultural norms, infrastructure) may outweigh micro-level interventions in shaping innovation outcomes. These findings contradict the assumption (such as in Kruger & Steyn, 2020) that institutional mechanisms like incubators and accelerators inherently enhance R&D transfer. Instead, the paper argues that general entrepreneurship support mechanisms may not be sufficient or effective across all contexts.

The findings of this study carry significant implications for policymakers aiming to enhance R&D transference in entrepreneurial ecosystems. First, the substantial variance in R&D transference explained by country differences underscores the need for tailored national strategies. Additional investigation into what these national strategies should entail must be carefully examined. Governments should assess and address specific barriers within their own context (such as regulatory inefficiencies, lack of trust in institutions, or inadequate support mechanisms) that may hinder R&D diffusion. Policies that promote overall transparency, reduce bureaucratic hurdles, and ensure consistent support for innovation can create a more conducive environment for entrepreneurial growth.

Second, the limited impact of individual-level perceived factors and entrepreneurs’ recommendations in the study suggest that top-down policy measures may be more effective than bottom-up initiatives in driving R&D transfer. While firm-level capabilities remain important, their influence may be constrained without robust national support systems. Thus, investment in capacity-building initiatives should be complemented by systemic reforms.

Third, the nuanced findings from the model, including pandemic intercessions, highlight the importance of responsive and targeted government interventions during crises. Although pandemic-related actions were not directly significant for R&D transference, their ability to amplify country-level differences indicates that crisis policies must be designed with an eye toward long-term innovation outcomes. Further analysis is needed to identify what these might entail. Governments should consider implementing resilience-focused R&D support, such as flexible funding mechanisms, innovation grants, and digital transformation incentives.

Furthermore, the comparative performance of developing countries in R&D transference suggests that international collaboration and knowledge-sharing can be powerful tools for enhancing innovation capacity. Policymakers in both developed and developing nations should explore partnerships that facilitate technology transfer, joint research initiatives, and access to global innovation networks.

The results suggest that a rethinking of entrepreneurship support policies may be timely. Since aspects that typically foster entrepreneurship (like digitization, economic environments, and government policies, as assessed in the first factor and first recommendation) show limited explanatory power for R&D transference, governments may need to reassess the effectiveness of current entrepreneurship support mechanisms. Instead of focusing solely on general entrepreneurship-friendly environments, policies should be developed to directly support R&D transference, such as through tech transfer offices, innovation hubs, or targeted grants. The recommendation to pivot from broad entrepreneurship support mechanisms toward more targeted commercialization instruments is, at present, speculative. Future research incorporating objective performance metrics and program-level evaluations is necessary to inform evidence-based policy design.

The results in the present study underscore the importance of national context, albeit not the type of specific areas of context that entrepreneurs perceive would support them. This study finds that country itself significantly affects R&D transference. This suggests that national culture, regulatory frameworks, and institutional structures may play a bigger role than previously thought. Entrepreneurs and investors should therefore consider country-specific overall factors, not just those precisely designed to drive entrepreneurship, when evaluating the potential for R&D-based innovation.

Since the analysis was applied in a global context, these findings have implications related to strategic international expansion. Specifically, small businesses aiming to leverage R&D for growth might benefit from international partnerships or relocation to countries with higher R&D transference rates. From the perspective of policymakers, governments and incubators could facilitate cross-border innovation programs or R&D mobility schemes to help entrepreneurs tap into more favorable systems.

The results indicate that future analysis of entrepreneurship could use more refined metrics for evaluating innovation potential. Traditional indicators of entrepreneurship (like ease of starting a business or digital readiness) may not be sufficient to predict R&D transference success. Innovation agencies and researchers should develop new metrics that better capture real-world R&D potential, possibly including measures of tech transfer efficiency or success rates. Alternative constructs and analytics can be drawn from the literature.

Achuthan et al. (2025) show how emerging frameworks can operationalize innovation ecosystems through data-driven analyses of entrepreneurial education (immersive learning to enhance entrepreneurial capabilities), virtual technology adoption (as drivers of new business models and customer engagement), and digital transformation (allowing the creation of hybrid business models). Their framework leads to concrete and measurable potential indicators for each domain: immersive learning engagement, entrepreneurial intention index, skill acquisition rate; virtual technology adoption velocity, technology cost-effectiveness ratio, interoperability compliance score; and, finally, digital integration index, innovation density, and cybersecurity robustness.

Raman et al. (2025) proposed analytics and frameworks, such as immersive self-regulation and ethical design benchmarks, that can be used not only to evaluate innovation ecosystems but also to guide entrepreneurial education and learning. Chandran et al. (2025) proposed that training can be a critical enabler of positive outcomes. The findings of the present study may have implications for building entrepreneurs’ education and awareness. Entrepreneurs may not be fully aware of how their national context affects their ability to commercialize R&D. Educational programs and advisory services should include country-specific guidance on navigating R&D circumstances, including legal, financial, and institutional support structures. Entrepreneurs’ training can address two challenges discussed in the current study. On one hand, it can address the perception–reality disconnect of what affects R&D outcomes. On the other hand, it may provide entrepreneurs with critical information on how to navigate the institutional environment. The literature has shown that training strengthens the link between entrepreneurial intention and actual business creation by providing actionable knowledge and skills (Sancho et al., 2022). Programs are most effective when they integrate institutional factors (such as access to incubators, funding, and policy support) into their design (Serpente et al., 2025). In less entrepreneur-friendly environments, training can compensate for institutional barriers by fostering resilience and adaptability (Aparicio et al., 2021).

The current study contributes to a deeper understanding of the cross-national undercurrents influencing the transfer of R&D. The investigation challenges conventional assumptions in entrepreneurship literature and policy design, which often prioritize support mechanisms as key drivers of innovation outcomes. The results underscore the importance of national context, suggesting that broader institutional frameworks, innovation policies, and cultural attitudes toward technology adoption may be more critical to R&D transference than previously recognized. The limited significance of entrepreneurs’ recommendations and pandemic-related policies further highlights the complexity of the innovation ecosystem and the resilience of R&D transference to short-term disruptions. For policymakers, these insights call for a shift toward targeted interventions that directly support R&D transference, rather than relying solely on entrepreneurship initiatives. For researchers, these findings open new avenues for exploring specific national mechanisms that may better facilitate R&D outcomes.

6. Conclusions

This study advances the understanding of cross-national variation in the transfer of R&D. The analysis reveals that the level of R&D transference is significantly associated with country location, while home-country conditions traditionally believed to foster entrepreneurship do not explain a large portion of the variance in R&D transference. These findings suggest that the mechanisms through which R&D translates into commercial opportunities are not uniformly shaped by entrepreneurship-supportive environments but rather by broader national factors that may include institutional frameworks, innovation policy development, and cultural attitudes toward technology adoption. This idea challenges prevailing assumptions in entrepreneurship literature and policy design, which often emphasize distinct entrepreneurial conditions as key drivers of innovation outcomes. The implications are twofold. First, policymakers should consider targeted interventions that directly support R&D transference, rather than relying solely on entrepreneurship support. Second, future research should explore which specific national-level factors, such as intellectual property regimes, bureaucracy, or availability of skilled workforce, most effectively facilitate R&D transference for small businesses.

The current study’s results highlight the importance of national context in shaping entrepreneurial outcomes. Policymakers should reconsider the assumption that general entrepreneurial support mechanisms automatically enhance R&D transference. Entrepreneurs’ recommendations for improving the ecosystem, although aligned with fostering factors, did not have a measurable effect on R&D transference. This disconnect may reflect a gap between perceived needs and actual systemic levers that influence innovation outcomes.

While entrepreneurs may have experienced pandemic policies as impactful, their influence on R&D outcomes may be diluted by other factors such as institutional resilience, digital readiness, or pre-existing innovation ecosystems. This highlights the complexity of measuring policy impact during crises and suggests that R&D transference may be more resistant to short-term shocks than anticipated.

The consistent significance of the country variable across all statistical models reinforces the idea that national context plays a dominant role in shaping R&D transference. Cross-national comparisons should be approached with caution. Policymakers and researchers must account for country-specific conditions when designing entrepreneurship and innovation strategies. One-size-fits-all approaches are unlikely to yield consistent results across diverse environments. Based on its findings, the present paper advocates for complementing perception-based recommendations with empirical evaluations of policy effectiveness.

Several limitations were discussed throughout the study. Expanding on these limitations is important for understanding fully the grounding of the research and for suggesting further investigation.

The study uses data from the GEM’s NES, which is based on expert perceptions rather than objective measures. While perceptions offer valuable insights into entrepreneurial environments, they may not accurately reflect actual conditions or outcomes. This introduces subjectivity and potential bias, especially when comparing across countries with different cultural norms and expectations. Reliance on perception-based standardized scores can introduce several types of bias that may affect the validity and generalizability of research findings. The data may suffer from cultural and contextual bias, due to differing cultural norms around entrepreneurship, innovation, or government roles and language nuances that may affect how questions are understood. This can distort cross-country comparisons and lead to misleading conclusions about the effectiveness of policies or the state of entrepreneurial ecosystems. Respondents may also unconsciously rate factors based on pre-existing beliefs or expectations, rather than objective evidence, which can reinforce policy myths and obscure actual performance gaps. Perceptions are often shaped by recent events (e.g., the COVID-19 pandemic), which may not reflect long-term trends. This can inflate or deflate scores based on short-term sentiment and undermine the reliability of analyses. It must be acknowledged that these aspects are limitations for all research employing GEM data, including this study. Robustness checks, like those accompanying the statistical models used in the methodology, can improve confidence in findings.

Despite testing factors traditionally associated with fostering entrepreneurship, their limited significance suggests that the chosen variables may not fully capture the mechanisms that drive R&D transference, or that the model specification could be improved. Model 2 shows violations of error variance assumptions and highlights the challenge of integrating qualitative data into quantitative models. Model 3 incorporates pandemic-era government actions, both positive and negative. However, these variables do not show significant effects on R&D transference. This may be due to circumstances not accounted for in this study, such as the short-term nature of pandemic policies, the complexity of appraising policy impact during crises, and the possibility that R&D transference is more resilient to temporary disruptions than expected.

The timeframe of the data allowed for the examination of the mitigating effects of entrepreneurship-related government interventions on R&D transfer. However, it may have also created limitations on the interpretation of results during this interval. As Linan and Jaen (2022) identified, demand- and supply-side shocks have disrupted entrepreneurial activity. Their study examined the multifaceted impact of the COVID-19 pandemic on entrepreneurship, with a focus on emerging economies, and drew on prior financial crises and early pandemic data. The authors found that reduced consumer demand, increased uncertainty, restricted mobility, and limited access to funding heightened barriers for new ventures. Their paper revealed important nuances in the evolution of necessity-driven vs. opportunity-driven entrepreneurship. The authors noted that, during times of disruption, while overall entrepreneurial activity may initially decline, necessity entrepreneurship is likely to surge due to rising unemployment and limited job alternatives. Opportunity entrepreneurship may also grow, albeit more slowly and selectively, contingent on institutional support and economic recovery. Differences among economies support the role of institutional context in shaping entrepreneurial dynamics. In fragile states, necessity entrepreneurship tends to flourish, often in informal sectors, while opportunity entrepreneurship is hindered by weak governance and limited economic freedom. The authors emphasized the need for robust institutions to foster high-potential ventures. The study highlighted several research gaps, including the need to explore how perceptions of risk and opportunity evolve during crises. The present paper sheds some light on such perceptions.

The current study emphasizes that the country-level context is relevant for R&D transference, but it does not identify which specific national mechanisms (e.g., IP laws, tech transfer offices) are most influential. This limits the actionability of these findings for policymakers and entrepreneurs. The results suggest that one-size-fits-all approaches are ineffective and do not offer a clear alternative framework. The context factors and national mechanisms were outside the scope of this study but should be at the center of future research. The analysis is cross-sectional, based on data from a single year (2021). This limits the ability to assess long-term trends, policy evolution, or lag effects of entrepreneurship support structures on R&D outcomes. This study does not differentiate between sectors, which may vary significantly in their R&D transfer dynamics. This omission may obscure sector-specific insights that could be critical for targeted policy design.

Future research can address the limitations of the current study in several strategic ways. To reduce reliance on perception-based data, future studies can use quantitative national indicators such as R&D expenditure, patent filings, tech transfer rates, and innovation output. Expert surveys’ data can be combined with firm-level performance data to validate perceptions against actual outcomes. Instead of broad categories like the ones defined by the NES, future studies could break down these factors into specific subcomponents (e.g., tax incentives, startup grants, broadband penetration) or use composite indices (e.g., Global Innovation Index) to capture nuanced differences.

To address statistical reliability issues, alternative statistical techniques can be applied, such as hierarchical linear modeling or structural equation modeling. Homogeneity of variance can be ensured through data transformation or stratified sampling. For capturing long-term effects, future studies can track changes in R&D transference over time in response to policy shifts or economic changes or use panel data to observe delayed effects and cumulative impacts.

Furthermore, future research can account for industry variation and compare sectoral sensitivity to national context and policy interventions. To enhance generalizability, new studies can include more countries, especially underrepresented regions, or consider entrepreneur demographics (e.g., gender, age, education) to assess differential impacts. A related research area could be to evaluate policy implementation quality by assessing how well policies are implemented, not just whether they exist. Case studies or mixed-method approaches can be used to explore real-world dynamics related to R&D transfer.

In addition to the new ways in which new research can address the limitations of the current studies, several potential future directions could build on the current study. The differentiating potential of the location on the R&D transference suggests that a comparative analysis of R&D transference mechanisms may be warranted. Future research can investigate how different national mechanisms (e.g., tech transfer offices, innovation hubs, intellectual property laws) influence R&D transference success. Since general entrepreneurship conditions are not strong predictors, a deeper dive into specific innovation transfer mechanisms could reveal more actionable insights. This type of investigation may identify the need for new innovation metrics by answering a future question about alternative indicators that can better predict R&D transference success across countries. Traditional metrics like ease of doing business or digital readiness may be insufficient. Research could explore metrics such as tech transfer efficacy or performance.

An interesting future avenue of investigation may focus on the role of national culture in innovation to further understand how cultural dimensions and institutional trust can affect R&D transference. Since the current study suggests that national context matters more than expected, frameworks like Hofstede’s cultural dimensions may deliver valuable information. The framework identifies key cultural traits, such as uncertainty avoidance (the degree to which societies tolerate ambiguity and risk), long-term orientation (the emphasis on future rewards vs. short-term gains), power distance (acceptance of hierarchical structures), and individualism vs. collectivism, that shape organizational behavior and decision-making. These dimensions are highly relevant for R&D transfer because they influence how entrepreneurs approach risk, collaboration, and knowledge sharing, which are critical for R&D transference. For example, countries with high uncertainty avoidance may be less willing to invest in unproven technologies, slowing R&D adoption, while those with strong long-term orientation may prioritize sustained investment in innovation. Similarly, institutional trust can determine whether firms engage in cross-border partnerships or technology-sharing agreements, both essential for leveraging R&D outcomes.

Related studies can focus on the level of awareness among entrepreneurs regarding their national R&D transference context and how it can be improved. Such evaluations would offer input on educational interventions that could improve strategic decision-making. In the same avenue, future research could explore the outcomes of international R&D mobility programs for small businesses, expanding the concept of R&D transference to the international sphere, to appraise the effectiveness of cross-border incubators, innovation visas, or international partnerships.

Expanding on the type of investigation pursued in this study, future research can evaluate how changes in national entrepreneurship policies over time can affect R&D transference outcomes. Pursuing this inquiry with a longitudinal approach could reveal lag effects or unintended consequences of policy reforms. Finally, sector-specific R&D transferability can be examined to appraise if certain sectors (e.g., biotech, professional services, clean energy) may be more sensitive to national context in R&D transference. Such analysis could help policymakers tailor plans and support mechanisms more precisely.

Ultimately, the present paper affirms that one-size-fits-all approaches to fostering innovation are insufficient. Tailored strategies that account for country-specific conditions and sectoral sensitivities may be essential for unlocking the full potential of R&D for small businesses. By highlighting the nuanced role of the national context, the current study contributes to a more differentiated understanding of global innovation dynamics and offers a foundation for more effective, context-sensitive policies and entrepreneurial strategies.