EN中文

Journal of Geographical Sciences >

2025 , Vol. 35 >Issue 2: 409 - 431

DOI: https://doi.org/10.1007/s11442-025-2328-1

Research Articles

Influence mechanism of the flow of high-skilled talents on technological evolution in emerging countries

  • JIN Wenwan , 1 ,
  • ZHU Shengjun , 2, * ,
  • LIN Xiongbin 1
Expand
  • 1. Department of Geography and Spatial Information Technology, Zhejiang Collaborative Innovation Center for Land and Marine Spatial Utilization and Governance Research, Ningbo University, Ningbo 315211, Zhejiang, China
  • 2. College of Urban and Environmental Sciences, Peking University, Beijing 100871, China
*Zhu Shengjun (1984-), PhD, specialized in industrial upgrading, globalization, and regional development. E-mail:

This paper is initially published in Acta Geographica Sinica (Chinese edition), 2024, 79(10): 2621‒2637.

Jin Wenwan (1996-), PhD and Lecturer, specialized in innovative geography and talent mobility. E-mail:

Received date: 2024-08-05

  Accepted date: 2024-11-05

  Online published: 2025-03-14

Supported by

National Natural Science Foundation of China(42122006)

National Natural Science Foundation of China(41971154)

National Natural Science Foundation of China(42271201)

Key Project of National Natural Science Foundation of China(41731278)

Fold

Abstract

Globalization has resulted in a notable rise in the flow of high-skilled talent from emerging countries to developed nations. Current research on transnational talent flow mainly focuses on the destination countries, with less attention given to the perspective of the sending countries, particularly lacking a dynamic discussion on its impact on technological evolution in the origin countries. Based on the OECD REGPAT database, this paper aims to explore how talent groups migrating to developed countries facilitate the return of knowledge and technology to emerging countries and achieve breakthroughs in their technological evolution paths, while further discussing the potential mechanisms involved. The findings of this paper are as follows: (1) The technological development of emerging countries is a path-dependent process, where countries often branch into new technologies related to their preexisting knowledge base. Consequently, knowledge feedback from high-skilled talents increases the likelihood of sending countries developing unrelated technologies. (2) The mobility of talents across borders fosters more international collaborations and citations for patents that are unrelated to the local knowledge base, thus enriching the technological paths of sending countries. (3) The mobility of high-skilled talents primarily affects complex technologies, which have significant economic effects that encourage imitation by other countries. However, the effect on novel technologies is less significant due to their strong geographical stickiness. In general, this paper addresses the gaps in existing research on talent outflow and the technological evolution of origin countries, providing empirical evidence for the positive role of transnational talent mobility in the technological catch-up of emerging nations. Besides, it offers recommendations for talent export, import, and innovation policy formulation in these countries.

Key words： talent flow; technological evolution; path-breaking; mechanism research; emerging countries

Cite this article

JIN Wenwan , ZHU Shengjun , LIN Xiongbin . Influence mechanism of the flow of high-skilled talents on technological evolution in emerging countries[J]. Journal of Geographical Sciences, 2025 , 35(2) : 409 -431 . DOI: 10.1007/s11442-025-2328-1

1 Introduction

In the context of globalization, the transnational movement of high-skilled talent has become increasingly common (Kerr et al., 2016). By 2020, the total number of international migrants rose from 153 million in 1990 to 281 million, representing 3.6% of the global population (McAuliffe and Triandafyllidou, 2021). Among these migrants, the influx of talent from emerging countries such as India, the Philippines, and China into developed countries like the United States, the United Kingdom, and France has surged, becoming a primary direction for the transnational flow of high-skilled talent. Currently, the spatial distribution of global innovation activities reveals a distinct core-periphery structure, highlighting an invisible gap between developed and developing countries. Some studies focus on the knowledge spillover from developed countries to emerging nations (Fink and Miguelez, 2017; Qiu et al., 2017; Ozgen, 2021). The mobility of talent from these emerging countries facilitates the transnational transfer of knowledge, practices, and relational networks, creating effective channels for diverse forms of global and multidirectional communication (Ma et al., 2022; He et al., 2023). This knowledge reflow benefits emerging countries by granting them access to cutting-edge insights from developed countries, thus facilitating development transformation, optimizing economic structures, and generating development momentum, which accelerates their technological catch-up with developed countries (Hou et al., 2020; Jin et al., 2022a). For instance, cities like Hsinchu in Taiwan, Bangalore and Hyderabad in India have successfully developed export-oriented information and communication technology industries locally by learning from the experiences of Silicon Valley in the United States through the transnational movement of high-tech engineers and entrepreneurs. Therefore, the influence of high-skilled individuals moving across borders on developing nations, particularly regarding technology and innovation, is not only a topic receiving increasing attention in global development studies (World Bank, 2010), but also a critical issue that emerging countries need to explore in depth.
Research on migration, particularly the transnational movement of high-skilled talents, has become a focal point in geography, economics, and sociology (Liu et al., 2013). Prior studies have primarily emphasized the advantages that receiving nations gain from talent migration, while largely overlooking the consequences for sending nations (Jin et al., 2022a). Early research suggested that the “Brain Drain” phenomenon could result in a loss of human capital for origin countries, exacerbating inequalities between nations. For instance, Bhagwati and Hamada (1974), and McCulloch and Yellen (1977) highlight the detrimental effects of talent loss on origin countries, including the intensification of international inequalities, by introducing concepts such as labor market rigidity and information asymmetry. Subsequent research has found that origin countries may also benefit from transnational talent mobility through mechanisms such as skill dissemination, knowledge sharing, and return migration (Boeri et al., 2012; Clemens, 2015). However, current studies primarily adopt a static perspective on the impact of high-skilled talent flow on knowledge and technology diffusion, lacking an in-depth analysis of the technological evolution characteristics in emerging countries, and neglecting the complex dynamic mechanisms behind technological development (Jin et al., 2022a). Therefore, exploring the relationship between transnational talent mobility and the capacity of origin countries to expand their existing knowledge base and achieve breakthroughs in technological evolution remains a significant area for theoretical inquiry (Fagerberg and Verspagen, 2020).
Based on the discussion above, this paper aims to address the following questions: First, does the outflow of high-skilled talent affect the technological evolution process in emerging countries? Second, are the potential mechanisms through which talent mobility impacts technological evolution in emerging countries related to international patent cooperation or citations? Third, in which specific technological fields is the impact of transnational talent mobility on the technological evolution of origin countries likely to be concentrated? This study not only provides systematic cross-national empirical evidence to discuss the reverse effects of talent mobility on origin countries, but also seeks to fill the gaps between talent mobility theory and evolutionary economic geography. The conclusions drawn from this research can also provide a theoretical foundation for emerging countries, such as China, to develop effective talent mobility and technological innovation policies.

2 Theoretical mechanisms and research hypotheses

The process of national technological development is dynamic and non-random (Boschma and Frenken, 2010), influenced by both internal and external factors. Evolutionary economic geography examines how existing local technological foundations shape future technological dynamics. Regional technological dynamics can be viewed as a “branching” process (Regional Branching) that derives related technologies from existing ones, exhibiting path-dependent characteristics. Technological development requires relevant knowledge and information, but limitations in the local knowledge base can restrict the direction of technological evolution (Li et al., 2013). Research conducted by Kogler et al. (2013), Colombelli and Temgoua (2014), Whittle (2020), and He and Zhu (2019) in the U.S., EU, Ireland, and China, respectively, has shown that the degree of relatedness between technology and local knowledge bases, known as technological relatedness density, has a positive impact on technological evolution. This finding reflects the path-dependent nature of technological evolution at both national and regional levels. Nevertheless, the knowledge base does not solely determine the direction of technological evolution for a country or region. Some areas manage to establish international connections through channels such as cross-border trade and investment (Zhu et al., 2019; Qian and Yang, 2022; Cao and Xie, 2023), which can expand their local knowledge base and enable them to break free from existing technological paths.
Talent mobility serves as a crucial cross-regional channel for knowledge exchange. This paper constructs a research framework from the perspective of talent mobility, as illustrated in Figure 1. Unidirectional talent flow often facilitates bidirectional knowledge flow (Miguelez and Temgoua, 2020), enhancing benefits for both sending and receiving countries simultaneously (Ma, 2017). The knowledge spillover resulting from high-skilled talents includes direct returns through their permanent or temporary return to their home country (Sun et al., 2005), as well as indirect returns through the maintenance of personal, professional, or social ties with their origin countries (Breschi et al., 2017). For example, Saxenian (2008) found that 82% of immigrant scientists and engineers in Silicon Valley have shared technological information with their home countries through social networks. As a result, national knowledge acquisition is no longer limited to local sources; it can now expand globally by exchanging non-local knowledge elements alongside talent mobility (Trippl et al., 2009). However, due to social, economic, and institutional differences, the knowledge bases between developed and developing countries are becoming increasingly specialized and differentiated (Fassio et al., 2019; Zhou and He, 2019; Ma, 2020). High-skilled talent flowing from emerging countries to developed countries can bring back diverse cutting-edge knowledge, information, and technology to their home countries. This enables technological evolution to leverage both local and external knowledge bases at a lower cost, reducing the influence of technological relatedness on the evolution process, altering existing trajectories, and facilitating breakthroughs (Esposito and Rigby, 2019). Therefore, this study proposes the following hypotheses:
Figure 1 Research framework
H1: The influx of high-skilled talents from emerging countries to developed countries may reduce the origin country’s dependence on its domestic knowledge base for technological evolution, thus entering and maintaining less related technological fields, ultimately leading to breakthroughs in the pathways of technological evolution.
The role of talent mobility is closely linked to transnational knowledge flow. Patent collaboration and citations are direct means of facilitating knowledge flow and serve as essential channels for technological learning between countries (Hu and Jaffe, 2003). High-skilled talent typically establishes knowledge collaboration or citation relationships in their home country before relocating, and these networks only partially diminish with their move (Constant and Zimmermann, 2016). This continuity allows high-skilled individuals to maintain interactions with their home countries after settling in developed nations, thereby directly increasing the probability of transnational patent collaboration and citations. In addition, based on the demonstration effect, talent mobility may promote knowledge exchange channels among countries by reducing communication barriers. Caviggioli et al. (2020) argue that domestic talents can also reduce the distance to other countries through the mobility of surrounding talents, which enhances the potential for transnational interaction and cooperation. In other words, the positive impact of high-skilled talent mobility is not limited to individuals; it can achieve widespread spillover effects through relational networks, improving communication efficiency between the two countries (Arrow, 1969). To summarize, high-skilled talents can create mechanisms for transnational knowledge exchange and sharing through both direct and demonstration effects (D’Ambrosio et al., 2019). When combined with the advanced technologies of developed countries, such collaborations or cited patents tend to exhibit lower relatedness to the knowledge bases of emerging countries (Montobbio and Sterzi, 2013; Gui et al., 2023). Consequently, emerging countries may enhance knowledge collaboration and citation with other countries during periods of talent outflow, thereby diminishing the influence of their local knowledge base on technological evolution through strengthened knowledge return (Bahar et al., 2020). This paper proposes the following hypothesis:
H2: The influx of high-skilled talents from emerging countries to developed countries may promote the formation of transnational collaborative or cited patents, thereby reducing emerging countries’ dependence on their local knowledge base for technological evolution, and ultimately achieving breakthroughs in the pathways of technological evolution.

3 Data and methodology

3.1 Data collection and processing

This study measures transnational high-skilled talent mobility using inventor mobility data from the database of Fink and Miguelez (2017). This database includes nationality information for inventors under the Patent Cooperation Treaty (PCT), and accurately reflects inventor migration from 1980 to 2010. It is currently the only dataset available that provides both nationality and residency information for inventors, and has been widely utilized in empirical research (Miguelez and Temgoua, 2020). The database contains data from approximately 241 countries/regions and tracks the mobility of 460,997 inventors. Additionally, this paper utilizes the OECD REGPAT database (REGPAT2021.07) for global patent data to assess technological innovation at the national level. There are two primary reasons for using patent data. First, patents are public documents that are easily accessible and contain detailed information about inventors’ backgrounds and activities, making them suitable for cross-country comparisons (Fink and Miguelez, 2017; Miguelez and Temgoua, 2020). Second, patents serve as indicators of innovation output that link R&D activities with productivity, enabling the construction of quantitative indicators of innovative activity. The database includes global patent applications to the European Patent Office (EPO) and the Patent Cooperation Treaty (PCT) from 1977 to 2021, comprising 3,881,997 patent entries across eight International Patent Classification (IPC) categories: A: Necessities of life; B: Performing operations, transporting; C: Chemistry, metallurgy; D: Textiles, paper; E: Fixed constructions; F: Mechanical engineering, lighting; G: Physics; H: Electricity. Generally, applicants from different countries in this database exhibit similar application tendencies, which helps mitigate potential “home bias” issues. Moreover, higher application costs are more likely to reflect substantive innovation. Previous studies by Bahar et al. (2020), Caviggioli et al. (2020), and Wigger (2022) have also used this database to measure innovative activity. Furthermore, this paper incorporates other relevant control variables from databases such as the United Nations Conference on Trade and Development (UNCTAD) and the International Bilateral Trade Database.
The data preprocessing in this paper involves the following steps: (1) Following Miguelez and Temgoua (2020), this paper excludes the years 2019, 2020, and 2021 due to missing data, as well as the pre-1990 period where usage restrictions exist. This results in four defined phases: 1996-2000, 2001-2005, 2006-2010, and 2011-2015, where the application date used as the patent year to accurately reflect the occurrence of innovation activities. (2) Patents are assigned to countries based on inventors’ addresses to minimize potential sampling bias from large companies holding numerous patents through outsourced research activities. (3) The research focuses on both emerging and developed countries. Based on classifications from the IFC (International Finance Corporation, 1999) and Hoskisson et al. (2002), we identify rapidly growing developing countries in Asia, Latin America, Africa, and the Middle East, along with transition economies classified by the European Bank for Reconstruction and Development (EBRD,1998). This ultimately narrows the focus to 50 emerging economies, including China, India, and South Korea. The list of developed countries is derived from UN classification, comprising a total of 22 nations. A detailed list of these countries can be found in the footnote 1 . (4) Unmatched countries, regions (non-sovereign entities), and certain islands, such as the Faroe Islands and the Turks and Caicos Islands are excluded. To avoid interference from unstable innovative countries, we remove those with fewer than 50 patent applications during the study period or whose patent records appear in only a limited number of periods. (5) For patents developed through international collaboration, each is treated as a completely independent patent in every participating country, with all relevant technological classification information noted.

3.2 Key variables

3.2.1 Entry (Entryc,i,t) and maintenance (Maintainc,i,t) in technological fields

The core dependent variables in our model are binary variables: country technology entry (Entryc,i,t) and maintain (Maintainc,i,t). Technological entry (Entryc,i,t) is defined as 1 when a “country-technology” combination does not have a comparative advantage at the initial period t1 but has a technological advantage at the final period t2. If the “country-technology” combination lacks comparative advantage at both the initial stage t1 and the final stage t2, indicating that the country has not successfully entered the technological field, and technology entry (Entryc,i,t) is 0. Similarly, technological maintenance (Maintainc,i,t) is defined as 1 when a “country-technology” combination has a technological comparative advantage at both the initial period t1 and the final stage t2. If the combination has a comparative advantage at the initial period t1 but loses it by the final stage t2, technological maintain (Maintainc,i,t) is 0, signaling that the country has not successfully maintained its technological advantage in the corresponding field. The calculation of a country’s technological comparative advantage follows Eq. (1): where pat represents the number of patent applications at different scales, c represents the country, i represents the technology category, and t indicates the period.
R T A c , i , t = p a t c , i , t / ∑ i p a t c , t ∑ c p a t i , t / ∑ c ∑ i p a t t

3.2.2 Technological relatedness density (Densityc,i,t)

Drawing from the study by Hidalgo et al. (2007), this paper employs the concept of technological relatedness density (Densityc,i,t) to measure the extent to which a country’s technological evolution is influenced by its existing knowledge base. This variable ranges from 0 to 1, with higher values indicating a stronger dependency of a country’s technological development on its knowledge base. Additionally, this paper performs a dichotomization of the RTAc,j,t variable: when it is greater than 1, it is considered that the country has a comparative advantage in that technology for the year t; otherwise, it is considered as lacking a comparative advantage, as specified in Eq. (2).
D e n s i t y c , i , t = ∑ i ∅ i , j , t × R T A c , j , t ∑ i ∅ i , j , t
where ∅ i , j , trepresents the technological relatedness between different patent categories. Using co-occurrence analysis methods, this paper constructs a technological relatedness indicator based on the probability of two technology types appearing simultaneously in the same patent. The higher the minimum conditional probability of technology categories i and j co-appearing in the same patent, the stronger the technological relatedness, resulting in a greater value for ∅ i , j , t. Compared to the co-occurrence relationship at the geographical scale, this method can avoid errors caused by an overly large geographical spatial range, which is listed below:
∅ i , j , t = min P x i , t ∣ x j , t , P x j , t ∣ x i , t
where the variables i and j represent different technology types, while xi,t represents the probability of technology category i appearing in patents during period t. Similarly, xj,t indicates the probability of technology type j being present in patents during the same period t. This study averages the innovation output over five-year intervals and calculates the correlation between each technology type at each stage.

3.2.3 Technological diffusion effect of talent migration (Migrationc,i,t)

Building on the work of Valentina and Miguelez (2022), the core independent variable Migrationc,i,t combines the outflow of inventors with a country’s technological comparative advantage to measure the technology diffusion resulting from inventor mobility. The outflow of inventors refers to the scale of immigration across various technology types between countries, which is categorized into five groups: Electrical Engineering, Instruments, Chemistry, Process Engineering, Mechanical Engineering, and Others. Following the methodology of Schmoch (2008), this paper matches the four-digit IPC technology classification with these five categories of migrants, ultimately yielding the “origin country-technology-time” level variable Migrationc,i,t.
M i g r a t i o n c , i , t = ∑ c ′ I n v e n t o r c , c ′ , i , t × R c ′ , i ∈ t a , t
where I n v e n t o r c , c ′ , i , tdenotes the number of high-skilled inventor flows between origin country c and the destination country c' for technology type i during period t. R c ′ , i ∈ t a , tis a binary variable indicating whether technology type i has a comparative advantage in country c' during period t. It is 1 if there is a comparative advantage, and 0 otherwise.

4 Transnational talent mobility and national technological breakthroughs

Figure 2 illustrates the global mobility of high-skilled talent from 2006 to 2010 across different scales. Figure 2a shows that Europe and North America are the primary destinations for talent inflow at the Oceania level, while Asia is the main origin region. Among the numerous migration cases from emerging countries to developed nations, the most significant flow of talent is from Asia to North America, followed by Asia to Europe. At the national scale, this analysis retains the main talent flow combinations from emerging countries to developed countries, specifically the top 25%. The United States is the most significant destination for talent inflows, while China, India, and South Korea are the primary talent-exporting countries (Figure 2b). As a global superpower, the United States stands at the forefront of innovation development, continually attracting talents from other countries and facilitating the dissemination of knowledge and technology (World Bank, 2010). In addition to the United States, Germany, Japan, the United Kingdom, and Canada are also significant destinations for talent inflow. The scale of high-skilled talent inflow in these countries is often comparable and closely related to their political, geographical, and cultural ties with the origin countries; for instance, Japan primarily receives international talent from South Korea and China. This frequent and large-scale talent mobility between emerging and developed countries may be linked to the expanding capabilities in emerging countries (Qiu et al., 2017).
Figure 2 International mobility of inventors from 2006 to 2010
This study also highlights the extent of talent migration from emerging nations to developed countries, as illustrated by the heatmap in Figure 3. As overall economic development progresses and human capital accumulates, the scale of talent mobility between most countries increases. Notably, the increase in talent flow between the United States and various countries is particularly significant, especially with China, India, South Korea, and Russia. Additionally, internal talent mobility within some Asian countries, such as China and Japan, has also risen substantially. Conversely, there are a few cases of decreased talent flow between specific countries, such as Tunisia to Belgium, Slovakia to Australia, and Poland to Finland. These trends indicate that the channels for knowledge exchange through high-skilled talent mobility between most countries are stabilizing and strengthening, which is expected to have an increasingly significant impact.
Figure 3 Changes in the international mobility of inventors (1990 and 2010)
In addition, Table 1 presents the percentage of breakthrough patents for different countries during various periods. From 2011 to 2015, emerging countries showed strong performance in achieving breakthroughs in technological evolution paths, with countries like China, South Korea, and India achieving proportions exceeding 1%. Meanwhile, developed countries such as the United States and Japan also reported relatively high proportions of breakthrough patents, at 27.18% and 19.13% respectively. Nevertheless, some developed European countries have seen a decline in their patent proportions, dropping to less than 1%. This trend may be attributed to these countries already holding advantages in most technological fields, making it increasingly challenging to achieve further innovation breakthroughs, and this difficulty is expected to increase over time. For example, the United Kingdom classified nearly 2215 of the 25,524 patents initially applied for as breakthrough patents. However, among the 35,901 patents applied for later, only 1625 were breakthroughs. This indicates that while the total number of new patent applications has increased, the number of breakthrough patents has decreased. Norway and Denmark have also experienced a notable reduction in their proportions. Conversely, emerging nations like China and South Korea are increasingly demonstrating their capacity for innovation breakthroughs. Between 1996 and 2000, China applied for only 206 breakthrough patents, but by the period of 2006-2010, this number had risen to 2002, and the proportion continues to grow. This suggests that the global innovation landscape is not completely stagnant; rather, emerging nations are showing tremendous growth potential (Ma, 2020). The accumulation of innovation potential in these countries coincides with the flow of high-skilled talent to developed countries, and the possible causal relationship between the two will be explored further in subsequent discussions.
Table 1 Proportion of patents with technological novelty in different periods
Period Proportion of breakthrough patents
0-0.1% 0.1%-1.0% 1.0%-10.0% > 10.0%
Number Main countries Number Main countries Number Main countries Number Main countries
1996-2000 83 Malaysia, Thailand, Vietnam, Chile, Colombia 24 India, Singapore, Mexico, Russia, Portugal 15 China, South Korea, Japan, Norway, Denmark 2 USA,
Germany
2001-2005 85 Thailand, Vietnam, Chile, Egypt, Colombia 23 India, Malaysia, Denmark, Norway, Russia 14 China, South Korea, UK, Germany, France 2 USA, Japan
2006-2010 86 Thailand, Vietnam, Egypt, Colombia,
Argentina		 24 Malaysia, Denmark, Norway, Chile, Russia 12 China, South Korea, India, UK, Germany 2 USA, Japan
2011-2015 85 Egypt, Vietnam, Argentina,
Portugal,
Philippines		 30 Malaysia, Denmark, Norway, Thailand, Colombia 6 South Korea, France, Germany, India, UK 3 USA,
Japan, China

5 Empirical analysis

5.1 The empirical model

To examine the relationship between the cross-border mobility of high-skilled talent and the technological evolution in origin countries, this paper develops the following baseline model at the “country-time-technology” level:
Y c , i , t + 1 = α + β 1 M i g r a t i o n c , i , t + β 2 D e n s i t y c , i , t + β 3 M i g r a t i o n c , i , t × D e n s i t y c , i , t + γ C o n t r o l s c , i , t + θ c , t + μ i , t + ε c , i , t
where c represents the origin country of the high-skilled talent, i denotes the four-digit IPC technology type, and t indicates different periods. The binary dependent variable Yc,i,t+1 encompasses both the entry and maintenance of technology. The core independent variables of the model include Densityc,i,t, Migrationc,i,t and their interaction term Migrationc,i,t× Densityc,i,t. Densityc,i,t reflects the proximity of the knowledge base for technology i to the origin country c during period t, measuring the extent to which technological evolution relies on the local knowledge base. The coefficient β2 reflects the impact of the local knowledge base on the technological development of the origin country. Migrationc,i,t combines the outflow of talent with the comparative technological advantage of the destination country’s technology, reflecting the knowledge implications for the origin country stemming from the mobility of high-skilled talent. The coefficient β1 represents the role of talent mobility in the technological development of the origin country. The term Migrationc,i,t×Densityc,i,t evaluates the impact of talent mobility on the path dependence of national technological evolution. The coefficient β3>0 indicates that the outflow of high-skilled talent strengthens the path dependence of the origin country’s technological evolution, while β3<0 suggests that it may facilitate breakthroughs in the technological evolution of the origin country. The variable α represents the constant term in the model.
In addition to talent mobility, the main global channels of knowledge flow also include trade and investment (Isaksen and Trippl, 2017). To simultaneously control for foreign direct investment and export trade, this study adopts the calculation method of the core dependent variable Migrationc,i,t as outlined by Bahar et al. (2014), combining trade and capital outflows with the comparative advantage of the destination country’s technology. Additionally, the total number of patents from the previous stage is included to control for the national innovation base. The coefficient γ represents the impact of various control variables on the technological development of the origin country. To avoid omitted variable bias in the control variables, this study incorporates “country-year” fixed effects θc,t to control for time-varying characteristics at the national level related to the origin country of mobile inventors, such as national income, size, and institutional factors. It also includes “technology-year” fixed effects μi,t to address time-varying characteristics at the technology type level, such as the scale of patent applications for different technology types across various years. Finally, all variables undergo a hyperbolic sine transformation. This linear monotonic transformation is similar to a logarithmic transformation and can be defined as 0 (Bahar et al., 2014). Table 2 presents the statistical descriptions of the variables above. Moreover, the model in this study passes the VIF test, indicating no severe multicollinearity issues.
Table 2 Descriptive statistics of variables
Variables Interpretation Observations Mean value Standard deviation Minimum Maximum
Density Technological relatedness density 36,404 0.28 0.12 0.028 110.59
Migration Perform an inverse hyperbolic sine transformation on the product of the number of migrants and the technological comparative advantage of the destination country 36,404 3.27 2.03 0.00 6325.20
FDI Perform an inverse hyperbolic sine transformation on the product of the foreign direct investment value and the technological comparative advantage of the destination country 36,404 12.64 0.99 0.00 9625.40
Trade Perform an inverse hyperbolic sine transformation on the product of trade value and the technological comparative advantage of the destination country 36,404 21.26 1.36 0.00 43.00
Total_pat Perform an inverse hyperbolic sine transformation on the cumulative number of patents from the previous period in the country 36,404 0.53 1.10 0.00 8.44

5.2 Empirical regression results

5.2.1 Baseline regression results

Tables 3 and 4 present the baseline regression results for technology entry and maintenance when emerging countries are considered as origin countries. Model (1) in Table 3 examines the relationship between talent flow and technological entry from the origin country. The significant positive coefficient of the core independent variable Migration indicates that talent flow helps trigger the return of technological knowledge to the origin country, enabling emerging nations to learn advanced technologies from developed countries. This finding is consistent with the conclusions of Miguelez and Temgoua (2020) and Fackler et al. (2020), both of which emphasize the crucial role of mobile talent in international knowledge dissemination. Model (2) assesses the impact of the technology density variable Density on technological entry in emerging countries. The significant positive coefficient for the Density variable indicates that the technological evolution in these countries exhibits path dependency characteristics. Models (3) and (4) further incorporate the interaction term Migration ×Density, along with controls for trade and investment. The results indicate that the interaction term’s coefficient is consistently significant and negative, suggesting that high-skilled talent mobility allows emerging countries to rely less on domestic knowledge bases and to explore fields with lower technological relatedness (He and Zhu, 2020).
Table 3 Regression of talent outflow and technological entry of sending country
Entry
(1) (2) (3) (4)
Migration 0.0226*** 0.0173*** 0.0269***
(0.0029) (0.0037) (0.0037)
Density 12.2057*** 12.5667*** 12.9290***
(0.0679) (0.0792) (0.0794)
Migration×Density -0.0943*** -0.1270***
(0.0112) (0.0112)
Trade -0.0236***
(0.0037)
FDI 0.0346***
(0.0118)
Total_pat 0.0829***
(0.0031)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
N 36383 36383 36383 36383
R2 0.169 0.575 0.576 0.576

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

Similar to the results presented in Table 3, Table 4 demonstrates that the coefficients for the variables Migration and Density in Models (1) and (2) are both significantly positive. This indicates that talent mobility also helps emerging countries maintain their advantages in relevant technology fields, a process characterized by path dependence. Models (3) and (4) also further reveal that the interaction term Migration ×Density has significantly negative coefficients. This suggests that the outflow of high-skilled talent enables emerging countries to reduce their dependence on domestic knowledge bases and maintain comparative advantages across various and differentiated technologies. These results align with studies on transportation (Jin et al., 2022b), trade (Mao and He, 2019), and investment (Zhou and He, 2019), which indicate that high-skilled talent flowing from emerging countries to developed countries can bring back non-local knowledge and technology to the origin country. This process reduces emerging countries’ reliance on local knowledge bases and encourages them to enter and maintain new technology fields with lower technological association density, thereby achieving breakthroughs in their technological evolution paths. This supports the validity of Hypothesis 1 2 . These findings provide new evidence regarding the impact of talent outflow on knowledge diffusion and indicate that talent mobility may serve as a long-term transnational channel for emerging countries, accelerating their innovation catch-up with developed nations (Docquier and Rapoport, 2012).
Table 4 Regression of talent outflow and technological maintenance of sending country
Maintain
(1) (2) (3) (4)
Migration 0.0343*** 0.0119* 0.0215***
(0.0052) (0.0066) (0.0066)
Density 14.0775*** 14.5398*** 14.5946***
(0.1071) (0.129) (0.1277)
Migration×Density -0.0981*** -0.1224***
(0.0176) (0.0176)
Trade -0.0346***
(0.0063)
FDI -0.0034
(0.0166)
Total_pat 0.0830***
(0.0047)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
N 16445 16445 16445 16445
R2 0.247 0.660 0.662 0.670

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

Except for the core variables, the coefficients of other control variables mostly remain consistent across different models and align with theoretical expectations. The results indicate that the coefficient for the multinational trade variable (Trade) is significantly negative, while the coefficient for the multinational investment variable (FDI) is significantly positive (Table 4). This suggests that the flow of investment between countries facilitates the origin country’s acquisition of advantageous technologies from the destination country, whereas the multinational trade channel struggles to demonstrate a corresponding effect. In other words, transnational learning in the technology field is more closely related to the multinational investment variable. Wan and Long (2023) highlight a strong connection between investment and technological development. Moreover, the knowledge spillover effects through trade channels are more closely linked to industrial relationships and tend to have lesser impact on technological innovation activities (Duan and Du, 2020). International trade networks may encourage countries to establish comparative advantages and deepen specialization, potentially hindering breakthroughs in existing technological pathways (Ding and Li, 2022). Furthermore, the initial level of innovation in a country (Total_pat) positively influences the entry and maintenance of new technologies, indicating that countries with stronger innovation capabilities can more easily achieve technological expansion.

5.2.2 Endogeneity issues

The findings above indicate that the mobility of high-skilled talents influences the entry and maintenance of technology in emerging countries that serve as origin countries. In fact, the technological development in the origin country may also affect the transnational flow of talent, introducing an endogeneity issue in the model presented in this paper. Drawing on the research of Bahar et al. (2014, 2022), this study attempts to mitigate the endogeneity problem of the model using two types of instrumental variables. The first type uses historical migration data, measuring talent mobility through the lagged stock of immigrants from the same country over 30 years, with specific instrumental variable indicators developed as follows:
i n v I V c , c ′ , t = ∑ c ′ I n v e n t o r c , c ′ , t a , t − 30 e m × R c ′ , t ∈ t a , t w − 1
The second type of instrumental variable is based on the “push-pull” model. Following Card (2001), this approach predicts the actual stock of inventors migrating from the origin country c to destination country c' by considering both the “push” and “pull” factors. These instrumental variables are specifically constructed as follows:
p u s h c ′ , t − c = ∑ i i n v e n t o r s i , c ′ , t ∑ i ∑ c ′ i n v e n t o r s i , c ′ , t , ​           i ≠ c
p u l l c , t − c ′ = ∑ j i n v e n t o r s c , j , t ∑ c ∑ j i n v e n t o r s c , j , t ,   j ≠ c ′
i n v I V c , c ′ , t = p u s h c ′ , t − c × p u l l c , t − c ′ × ∑ c ∑ c ′ i n v e n t o r s c , c ′ , t
where c and c' represent the destination and the origin country respectively. The variables i and j are partner countries that have bilateral flows. To measure the outward push factors of the origin country, we calculate the proportion of inventors migrating from country c' to other countries in year t relative to the total outflow of inventors from all countries, denoted as pushc',t. Similarly, to measure the attractiveness of the destination country, we compute the proportion of inventors flowing into country c from other countries in year t relative to the total inflow of inventors to all countries, denoted as pullc,t. By combining the number of inventors migrating between countries c and c', along with the “push” (pushc',t) and “pull” (pullc,t) indicator, we generate a predicted value for the inventor flow between the two countries over a specific period. Ultimately, these predicted flow values are then aggregated for the specific destination country to serve as an instrumental variable for the outflow of inventors from the origin country. To further reduce endogeneity, this study excludes the existing bilateral flow scale between the original countries c and c' during the instrumental variable calculation process. Table 5 presents the results of the origin country’s entry (Entry) and maintenance (Maintain) models after incorporating the instrumental variables. Notably, the interaction term Migration×Density remains significantly negative, reinforcing the robustness of our findings. This indicates that the mobility of inventors indeed reduces the origin country’s reliance on local knowledge bases when entering and maintaining technological evolution. Moreover, the Cragg-Donald Wald F statistic for the model is much greater than 20, indicating no weak instrument problem.
Table 5 Regression of IV variable
IV Entry Maintain
(1) (2) (3) (4)
Migration -0.0394* 0.4764*** 0.0221 0.1210***
(0.0213) (0.0293) (0.0243) (0.0405)
Density 13.2815*** 15.7947*** 15.3863*** 16.7016***
(0.1313) (0.2266) (0.2335) (0.3783)
Migration×Density -0.1846*** -1.0103*** -0.2948*** -0.6241***
(0.0313) (0.0581) (0.0499) (0.0916)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
Control variables YES YES YES YES
N 36383 36383 16445 16445
R2 285.140 239.076 255.673 166.417

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

5.2.3 Discussion on potential mechanisms

To further explore the potential mechanisms linking talent mobility and national technological evolution, this study will examine the impact of talent mobility on patent collaboration and international citations in emerging countries. Additionally, it will investigate the differences in the likelihood of achieving breakthroughs in technological evolution pathways between multinational collaborations or citations of patents and general patents. In Table 6, the collaboration_rate refers to the growth rate of the ratio of multinational collaborative patents to all patents over a specific period, representing the compound average growth rate of the collaboration ratio within technological field i for country c during period t. Similarly, the citation_rate captures the growth rate of the ratio of multinational cited patents to all patents over the specific period. The specific calculation process is as follows:
G r o w t h c , i , t = r a t i o c , i , t r a t i o c , i , t − 1 1 / 5 − 1         if   r a t i o t − 1 > 0
Table 6 Regression on talent outflow and probability of patent collaboration (citation)
Collaboration_rate Citation_rate
(1) (2) (3) (4)
Migration 0.0496** 0.0609*** -0.0194 0.0650***
(0.0247) (0.0112) (0.0150) (0.0037)
Trade 0.4187*** 0.3609***
(0.0022) (0.0007)
FDI 0.0146 0.0125***
(0.0098) (0.0041)
Previous rate -0.3602*** -0.2729***
(0.0079) (0.0034)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
N 11707 11707 22695 22695
R2 0.005 0.792 0.003 0.938

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

Table 6 presents regression analyses using growth-type dependent variables, the variable Migration, and a series of fixed effects. The findings indicate that the migration variable is significantly positive in most cases, suggesting that talent outflow significantly enhances the ratio of multinational cooperation and citation of patents in relevant fields within the origin country. This implies that the outflow of talent fosters innovation collaboration and learning between emerging countries and other nations. Skilled individuals who have migrated tend to maintain communication channels with their home countries (Constant and Zimmermann, 2016), such as patent collaborations and citation relationships. This connection strengthens interactions between talents from both countries based on demonstration effects, ultimately leading to a reverse flow of knowledge from developed countries back to emerging nations (Agrawal et al., 2011). Previous studies by Fink and Miguelez (2017) and D’Ambrosio et al. (2019) have also found similar results. They examined how migration affects cross-border innovation networks and found a positive correlation between international collaboration in patent activities in the home country and the number of inventors who move across borders. Lissoni (2018) points out that patent citations play a mediating role in the knowledge dissemination process of high-skilled talent. This cross-border knowledge diffusion relationship is also evident in literature citations. For example, Bosetti et al. (2015) found that migration flows benefit the international citations of scientific papers in the host country.
A further comparison is made between multinational and general patents to highlight their distinctions. This paper distinguishes between two binary variables, collaboration and citation, to indicate whether patents achieve “cross-national collaboration” and “cross-national citation.” When there is multinational cooperation or citation, the variable is assigned a value of 1; otherwise, it receives a value of 0. The findings presented in Table 7 reveal that the coefficients of the interaction terms between technological relatedness density and the collaboration and citation variables are significantly negative. This indicates that internationally cooperative and cited patents tend to have lower local relatedness than general patents, making them more likely to facilitate innovative breakthroughs in the origin country. This phenomenon occurs because cooperation and citation networks can broaden the geographical scope of knowledge exchange (Breschi and Lissoni, 2001; Singh, 2005). Patents integrating knowledge from both countries differ from the local knowledge base, increasing the likelihood of breakthroughs along existing regional evolutionary paths. Hu and Jaffe (2003) also find that non-local knowledge-related models, such as knowledge-based cooperation or citations, can enhance breakthroughs in national knowledge innovation. Thus, Hypothesis 2 is confirmed, indicating that the flow of high-skilled talent from emerging countries to developed countries may reduce the dependence of technological evolution in emerging nations on their local knowledge base by promoting multinational cooperatives or cited patents, thereby achieving breakthroughs in technological evolution paths.
Table 7 The effect of patents collaboration (citation) on Density indicator
Density
(1) (2)
Collaboration -0.0054***
(0.0001)
Citation -0.0014***
(0.0001)
“Country-period” control YES YES
“Technology-period” control YES YES
N 5467312 670993
R2 0.803 0.911

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

5.2.4 Discussion on heterogeneity

Finally, the influence of talent outflow on the technological evolution of emerging countries may involve heterogeneity in technology types. Knowledge effectiveness and fluidity influence the knowledge return process triggered by talent mobility (Ma, 2017). Both complexity and originality are crucial attributes of different forms of technology. Existing research suggests that complex technologies often require a diverse combination of knowledge and are positively associated with regional economic benefits (Alesina et al., 2016). These technologies with higher economic value tend to have stronger incentives to transcend geographical boundaries and achieve transnational diffusion through talent mobility (Lv et al., 2022). On the other hand, technological novelty indicates the degree of innovation within a technology field, specifically whether it pertains to mature technologies (Corredoira and Banerjee, 2015). Novelty can be categorized into novelty in recombination (NR) and novelty in technological knowledge origins (NTO). Novelty in recombination refers to combinations of principles used in patents that differ from prior technologies, while novelty in technological knowledge origins indicates that a patent draws on knowledge from previously unrelated fields. Knowledge combinations in non-mature, novel technologies exhibit considerable uncertainty, making it more challenging to achieve a transnational return through talent mobility. This study utilizes indicators designed by Balland et al. (2020) and Verhoeven et al. (2016) to measure technological complexity based on the average number of inventors required per patent across various technology types. Additionally, the indicators for “novelty in recombination” and “novelty in technological knowledge origins” are combined, with the 1/4 quantile used to categorize low complexity, high complexity, low novelty, and high novelty. Heterogeneous regression analyses are then conducted for each group.
Table 8 presents the regression results of technological complexity heterogeneity. It shows that the interaction term Migration×Density is significantly negative for both high- and low-complexity technologies, with the absolute value of the coefficient being significantly greater for high-complexity technologies. This indicates that the outflow of high-skilled talent has a more pronounced impact on the evolution of high-complexity technologies in emerging countries. In fact, Kerr (2008) finds that the positive impact of mobile talent on the origin country is primarily evident in certain key industries. The reason is that the economic effects of high-complexity technologies are more likely to attract high-skilled individuals to learn and emulate (Ren et al., 2021), thereby increasing the likelihood of knowledge flowing back to emerging countries through talent mobility.
Table 8 Heterogeneity analysis on technological complexity
Entry Maintain
Low complexity High complexity Low complexity High complexity
(1) (2) (3) (4)
Migration 0.0351*** 0.0546*** 0.0582*** 0.0454***
(0.0109) (0.0069) (0.0188) (0.0128)
Density 12.9930*** 12.4044*** 15.7067*** 13.9160***
(0.1975) (0.1572) (0.3184) (0.2491)
Migration×Density -0.1193*** -0.2358*** -0.1990*** -0.1419***
(0.0323) (0.0213) (0.0497) (0.0343)
Trade -0.0190 -0.0290*** -0.0295* -0.0288**
(0.0117) (0.0067) (0.0166) (0.0117)
FDI 0.0663 0.0249 0.0684 -0.0440
(0.0433) (0.0186) (0.0690) (0.0272)
Total_pat 0.1420*** 0.0717*** 0.1057*** 0.0733***
(0.0106) (0.0051) (0.0104) (0.0093)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
N 7187 9946 3867 4494
R2 0.580 0.575 0.650 0.676

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

Table 9 details the situation concerning different levels of technological novelty. Here, the interaction term Migration×Density is significantly negative for low-novelty technologies, indicating that the outflow of high-skilled talent primarily facilitates breakthroughs in technological evolution pathways within emerging countries, particularly in low-novelty technology types. This is because novel technologies require extensive combinations of local knowledge and have higher diffusion costs (Plunket and de Waldemar, 2023). For instance, Mascarini et al. (2023) find that new technologies tend to exhibit a higher level of regional stickiness. Therefore, high-skilled talent primarily contributes knowledge to mature technology fields, making it challenging for emerging countries to leap into more advanced technology areas. In summary, the discussion in this paper suggests that while the outflow of high-skilled talent to developed countries may benefit emerging countries in breaking through their knowledge bases and developing mature and complex technology fields, it remains difficult for them to become originators of new technologies or to develop original and novel technology types. Nonetheless, some studies propose that emerging countries can cultivate high-novelty technologies through self-cultivation or by attracting talent (Liu and Chen, 2015).
Table 9 Heterogeneity analysis on technological novelty
Entry Maintain
Low novelty High novelty Low novelty High novelty
(1) (2) (3) (4)
Migration 0.0393*** 0.0227** 0.0248** 0.0004
(0.0064) (0.0093) (0.0115) (0.0180)
Density 12.1587*** 11.1634*** 13.4935*** 13.8471***
(0.1432) (0.1802) (0.2239) (0.3237)
Migration×Density -0.1901*** -0.1141*** -0.1543*** -0.0254
(0.0196) (0.0271) (0.0308) (0.0467)
Trade -0.0201*** -0.0200** -0.0483*** -0.0455**
(0.0060) (0.0100) (0.0104) (0.0179)
FDI 0.0223 0.0602** -0.0595** 0.0085
(0.0164) (0.0305) (0.0282) (0.0540)
Total_pat 0.0515*** 0.1114*** 0.0798*** 0.1015***
(0.0051) (0.0127) (0.0094) (0.0131)
“Country-period” control YES YES YES YES
“Technology-period” control YES YES YES YES
N 10903 6491 4875 2567
R2 0.559 0.590 0.662 0.736

Note: *** p<0.01, ** p<0.05, * p<0.1; the values in parentheses represent robust standard errors.

6 Conclusion and discussion

Currently, the world is undergoing a historic transformation in the knowledge economy, with an increasing number of countries entering or preparing to enter an innovation-driven development phase. High-skilled human capital is gradually replacing traditional production factors, and the importance of knowledge accumulation and acquisition is rising. This paper examines the impact of high-skilled talent mobility from emerging to developed countries, focusing on how this movement induces knowledge reflow and influences technological evolution in the origin countries. The research findings indicate that talent flow accelerates the return of knowledge from the host country to the origin country, helping emerging countries expand their knowledge base and enter and maintain breakthrough technologies. This talent flow effect may be related to the formation of more cross-national collaborations or cited patents in the origin country, which often have a lower correlation with the knowledge base of the origin nation. Finally, the impact of knowledge reflow induced by talent mobility also exhibits technological heterogeneity, primarily concentrated in high-complexity and low-novelty technologies. This implies that while the migration of skilled individuals from developing to industrialized nations supports the development of mature and complex new technologies, thereby improving the economic level of the country. However, it remains challenging for these emerging countries to truly position themselves at the forefront of global technological development.
The discussion in this study regarding the impact of talent outflow on emerging countries is significant. Currently, immigration research focuses on four main areas: talent movement from origin countries to destination countries, from destination countries to origin countries, across destination countries, and the destination country dimension (Singh, 2005), with a primary focus on the host country while relatively neglecting the origin country. This study not only analyzes the positive knowledge spillover of high-skilled talent mobility from the perspective of the origin country, providing empirical evidence for the “brain gain” phenomenon, but it also further integrates the dynamic theory of technological evolution to explore the positive role of knowledge reflow in facilitating the breakthroughs in technological evolution paths within these countries. This approach complements the primarily static international migration innovation research and provides insights into the factors of talent mobility that contribute to advancements in technological evolution, setting the stage for future research to examine the roles of other cross-regional channels. Furthermore, this paper’s findings also carry a range of policy implications. The increasing mobility of high-skilled talents has fostered the globalization of scientific research and development, benefiting emerging countries as the primary origins of talent. Thus, emerging nations can stimulate the movement of skilled individuals by implementing appropriate policies, to access non-local knowledge from more advanced countries, thereby enhancing the potential for technological breakthroughs. For example, China could consider relaxing certain mobility restrictions in its overseas talent policies, and actively establishing and strengthening domestic connections with overseas talent to enhance knowledge reflow and facilitate transnational innovation activities. However, since talent mobility is less likely to bring back novel knowledge, it is crucial for China not to overly depend on overseas talent. While continuing to attract talent through programs like the Peacock Plan and the Thousand Talents Program, it should also focus on cultivating local talents to create positive talent circulation and strengthen the country’s internal capacity for independent innovation.
Although this research provides valuable insights into the influence of high-skilled talent mobility on the technological evolution of emerging nations, several limitations remain. First, this paper primarily focuses on global patent data from a short period, which restricts the range of innovations considered and overlooks internal innovation activities in lower-tier countries, thereby reducing the applicability of the findings to real-world policies. Future studies could benefit from exploring non-patent innovation behaviors and analyzing dynamic changes in national innovation over a longer time series at a macro level. Second, the research scope and the subjects of analysis in this study are relatively narrow, lacking comparisons of impacts across different geographical scales and a detailed analysis of mechanisms at the micro level. Future research could enhance the analysis by incorporating the perspectives of enterprises and universities. Lastly, while emerging countries are significantly impacted by talent mobility, developed countries also benefit from it, potentially through different mechanisms. Future studies could explore these differences, particularly as some developed nations begin implementing specific policies to attract high-skilled talents. This comparative approach will broaden the application of the studies’ findings, providing valuable insights for policymakers and scholars involved in innovation and migration topics.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Agrawal A, Kapur D, McHale J et al., 2011. Brain drain or brain bank? The impact of skilled emigration on poor-country innovation. Journal of Urban Economics, 69(1): 43-55.

[2]
Alesina A, Harnoss J, Rapoport H, 2016. Birthplace diversity and economic prosperity. Journal of Economic Growth, 21(2): 101-138.

[3]
Arrow K J, 1969. Classificatory notes on the production and transmission of technological knowledge. The American Economic Review, 59(2): 29-35.

[4]
Bahar D, Choudhury P, Rapoport H, 2020. Migrant inventors and the technological advantage of nations. Research Policy, 49(9): 103947. doi: 10.1016/j.respol.2020.103947.

[5]
Bahar D, Hausmann R, Hidalgo C, 2014. Neighbors and the evolution of the comparative advantage of nations: Evidence of international knowledge diffusion? Journal of International Economics, 92(1): 111-123.

[6]
Bahar D, Rapoport H, Turati R, 2022. Birthplace diversity and economic complexity: Cross-country evidence. Research Policy, 51(8): 103991. doi: 10.1016/j.respol.2020.103991.

[7]
Balland P A, Jara-Figueroa C, Petralia S G et al., 2020. Complex economic activities concentrate in large cities. Nature Human Behaviour, 4(3): 248-254.

[8]
Bhagwati J, Hamada K, 1974. The brain drain, international integration of markets for professionals and unemployment: A theoretical analysis. Journal of Development Economics, 1(1): 19-42.

[9]
Boeri T, Brücker H, Docquier F et al., 2012. Brain Drain and Brain Gain: The Global Competition to Attract High-skilled Migrants. Oxford: Oxford University Press.

[10]
Boschma R, Frenken K, 2010. The spatial evolution of innovation networks:A proximity perspective. In: The Handbook of Evolutionary Economic Geography. Cheltenham: Edward Elgar Publishing.

[11]
Bosetti V, Cattaneo C, Verdolini E, 2015. Migration of skilled workers and innovation: A European perspective. Journal of International Economics, 96(2): 311-322.

[12]
Breschi S, Lissoni F, 2001. Knowledge spillovers and local innovation systems: A critical survey. Industrial and Corporate Change, 10(4): 975-1005.

[13]
Breschi S, Lissoni F, Miguelez E, 2017. Foreign-origin inventors in the USA: Testing for diaspora and brain gain effects. Journal of Economic Geography, 17(5): 1009-1038.

[14]
Cao G Q, Xie J G, 2023. Path-dependence or path-breakthrough: The evolution model of China’s regional comparative advantage. Inquiry into Economic Issues, (1): 87-104. (in Chinese)

[15]
Card D, 2001. Immigrant inflows, native outflows, and the local labor market impacts of higher immigration. Journal of Labor Economics, 19(1): 22-64.

[16]
Caviggioli F, Jensen P, Scellato G, 2020. Highly skilled migrants and technological diversification in the US and Europe. Technological Forecasting and Social Change, 154: 119951. doi: 10.1016/j.techfore.2020.119951.

[17]
Clemens M A, 2015. Losing Our Minds? New Research Directions on Skilled Migration and Development. >Washington, DC: Center for Global Development.

[18]
Colombelli A, Krafft J, Quatraro F, 2014. The emergence of new technology-based sectors in European regions: A proximity-based analysis of nanotechnology. Research Policy, 43(10): 1681-1696.

[19]
Constant A F, Zimmermann K F, 2016. Diaspora economics: New perspectives. International Journal of Manpower, 37(7): 1110-1135.

[20]
Corredoira R A, Banerjee P M, 2015. Measuring patent’s influence on technological evolution: A study of knowledge spanning and subsequent inventive activity. Research Policy, 44(2): 508-521.

[21]
D’Ambrosio A, Montresor S, Parrilli M D et al., 2019. Migration, communities on the move and international innovation networks: An empirical analysis of Spanish regions. Regional Studies, 53(1): 6-16.

[22]
Ding X Y, Li J X, 2022. Bilateral trade, product relatedness and the upgrading of the intra-product specialization of China. Economic Geography, 42(7): 106-116. (in Chinese)

[23]
Docquier F, Rapoport H, 2012. Globalization, brain drain, and development. Journal of Economic Literature, 50(3): 681-730.

[24]
Duan D Z, Du D B, 2020. Structural evolution of global high-tech trade system: Products, networks and influencing factors. Acta Geographica Sinica, 75(12): 2759-2776. (in Chinese)

DOI

[25]
EBRD, 1998. European Bank for Reconstruction and Development. Transition Report 1998.

[26]
Esposito C R, Rigby D L, 2019. Buzz and pipelines: The costs and benefits of local and non-local interaction. Journal of Economic Geography, 19(3): 753-773.

[27]
Fackler T A, Giesing Y, Laurentsyeva N, 2020. Knowledge remittances: Does emigration foster innovation? Research Policy, 49(9): 103863. doi: 10.1016/j.respol.2019.103863.

[28]
Fagerberg J, Verspagen B, 2020. Innovation, diffusion, the economy and contemporary challenges: A comment. Industrial and Corporate Change, 29(4): 1067-1073.

[29]
Fassio C, Montobbio F, Venturini A, 2019. Skilled migration and innovation in European industries. Research Policy, 48(3): 706-718.

DOI

[30]
Fink C, Miguelez E, 2017. The International Mobility of Talent and Innovation:New Evidence and Policy Implications, Intellectual Property, Innovation and Economic Development. Cambridge: Cambridge University Press.

[31]
Gui Q C, Du D B, Liu C L, 2023. Global scientific knowledge production: Process, pattern and influential factors. Scientia Geographica Sinica, 43(7): 1185-1194. (in Chinese)

DOI

[32]
He C F, Guo J H, Xie Y H, 2023. The impact of import spillover effect on the reconstruction of regional export industries: From the perspective of cognitive proximity. Acta Geographica Sinica, 78(6): 1323-1338. (in Chinese)

DOI

[33]
He C F, Zhu S J, 2019. Evolutionary Economic Geography in China. Singapore: Springer Nature Singapore Pte Ltd.

[34]
He C F, Zhu S J, 2020. The principle of relatedness in China’s regional industrial development. Acta Geographica Sinica, 75(12): 2684-2698. (in Chinese)

[35]
Hidalgo C, Klinger B, Barabási A et al., 2007. The product space conditions the development of nations. Science, 317: 482-487.

PMID

[36]
Hoskisson R E, Hitt M A, Johnson R A et al., 2002. Conflicting voices: The effects of institutional ownership heterogeneity and internal governance on corporate innovation strategies. Academy of Management Journal, 45(4): 697-716.

[37]
Hou C G, Du D B, Liu C L et al., 2020. Spatio-temporal evolution and factors influencing international student mobility networks in the world. Acta Geographica Sinica, 75(4): 681-694. (in Chinese)

DOI

[38]
Hu A, Jaffe A B, 2003. Patent citations and international knowledge flow: The cases of Korea and Taiwan. International Journal of Industrial Organization, 21(6): 849-880.

[39]
International Finance Corporation, 1999. Draft Education Strategy. Washington, DC.

[40]
Isaksen A, Trippl M, 2017. Exogenously led and policy-supported new path development in peripheral regions: Analytical and synthetic routes. Economic Geography, 93(5): 436-447.

[41]
Jin W W, Yang B F, Zhu S J, 2022a. Progress on the regional consequences of international high-skilled migration. Human Geography, 37(5): 8-14, 182. (in Chinese)

[42]
Jin W W, Zhu S J, Wang C, 2022b. Technological relatedness, regional accessibility and firm productivity. Geographical Research, 41(2): 509-526. (in Chinese)

[43]
Kerr S, Kerr W, Özden Ç et al., 2016. Global talent flows. Journal of Economic Perspectives, 30(4): 83-106.

[44]
Kerr W R, 2008. Ethnic scientific communities and international technology diffusion. Review of Economics and Statistics, 90(3): 518-537.

[45]
Kogler D F, Rigby D L, Tucker I, 2013. Mapping knowledge space and technological relatedness in US cities. European Planning Studies, 21(9): 1374-1391.

[46]
Li D D, Wang T, Zhou H, 2013. The structural characteristics of knowledge spillover networks based on different spatial and temporal scales. Scientia Geographica Sinica, 33(10): 1180-1187. (in Chinese)

[47]
Lissoni F, 2018. International migration and innovation diffusion: An eclectic survey. Regional Studies, 52(5): 702-714.

[48]
Liu Y, Shen J F, Liu Y Q, 2013. Transnational mobility of the highly skilled: A review. Human Geography, 28(2): 7-12. (in Chinese)

[49]
Liu Y G, Chen Y, 2015. Remarks on Chinese migration policy within globalizing urbanization. World Regional Studies, 24(1): 1-10, 37. (in Chinese)

[50]
Lv S, Sun T S, Sun Y K, 2022. Types of inter-city technological proximity and differences in the knowledge spillovers: From the perspective of knowledge complexity. Geographical Research, 41(11): 3036-3050. (in Chinese)

[51]
Ma H T, 2017. Triangle model of Chinese returnees: A tentative method for city networks based on talent flows. Geographical Research, 36(1): 161-170. (in Chinese)

[52]
Ma H T, 2020. The theoretical construction and network simulation of intercity innovative relationships in knowledge flow space. Acta Geographica Sinica, 75(4): 708-721. (in Chinese)

DOI

[53]
Ma L, Xie Y Y, Zhang B, 2022. Place embeddedness of Chinese academic returnees from the perspective of transnationalism. Acta Geographica Sinica, 77(6): 1430-1445. (in Chinese)

DOI

[54]
Mao X Y, He C F, 2019. Export expansion and regional diversification: Learning from the changing geography of China’s exports. The Professional Geographer, 71(4): 692-702.

[55]
Mascarini S, Garcia R, Quatraro F, 2023. Local knowledge spillovers and the effects of related and unrelated variety on the novelty of innovation. Regional Studies, 57(9): 1666-1680.

[56]
McAuliffe M, Triandafyllidou A, 2021. World Migration Report 2022. International Organization for Migration (IOM).

[57]
McCulloch R, Yellen J, 1977. Factor mobility, regional development, and the distribution of income. Journal of Political Economy, 85(1): 79-96.

[58]
Miguelez E, Temgoua C, 2020. Inventor migration and knowledge flows: A two-way communication channel? Research Policy, 49(9): 103914. doi: 10.1016/j.respol.2019.103914

[59]
Montobbio F, Sterzi V, 2013. The globalization of technology in emerging markets: A gravity model on the determinants of international patent collaborations. World Development, 44: 281-299.

[60]
Ozgen C, 2021. The economics of diversity: Innovation, productivity and the labour market. Journal of Economic Surveys, 35(4): 1168-1216.

[61]
Plunket A, de Waldemar F S, 2023. Regional recombinant novelty, related and unrelated technologies: A patent-level approach. Regional Studies, 57(7): 1267-1288.

[62]
Qian X Y, Yang Y, 2022. From path creation to path locking: A comparative analysis of Hong Kong, Macao, and Taiwan investment and foreign investment in the Pearl River Delta region. Progress in Geography, 41(9): 1635-1646. (in Chinese)

[63]
Qiu S M, Liu X L, Gao T S, 2017. Do emerging countries prefer local knowledge or distant knowledge? Spillover effect of university collaborations on local firms. Research Policy, 46(7): 1299-1311.

[64]
Ren Z R, He C F, Wang W Y, 2021. Progress of research on economic complexity and regional economic development in the perspective of evolutionary economic geography. Progress in Geography, 40(12): 2101-2115. (in Chinese)

DOI

[65]
Saxenian A, 2008. The new argonauts: Regional advantage in a global economy. Economic Geography, 84(1): 105-108.

[66]
Schmoch U, 2008. Concept of a technology classification for country comparisons. Final Report to the World Intellectual Property Organisation.

[67]
Singh J, 2005. Collaborative networks as determinants of knowledge diffusion patterns. Management Science, 51(5): 756-770.

[68]
Sun J, Ji J Y, Wang D, 2005. The research on regularity in the returning of overseas technological talents. China Soft Science, (8): 6-10. (in Chinese)

[69]
Trippl M, Tödtling F, Lengauer L, 2009. Knowledge sourcing beyond buzz and pipelines: Evidence from the Vienna software sector. Economic Geography, 85(4): 443-462.

[70]
Valentina D, Miguelez E, 2022. The ties that bind and transform: Knowledge remittances, relatedness and the direction of technical change. Journal of Economic Geography, 222(2): 423-448.

[71]
Verhoeven D, Bakker J, Veugelers R, 2016. Measuring technological novelty with patent-based indicators. Research Policy, 45(3): 707-723.

[72]
Wan W, Long X N, 2023. FDI access policies and technological innovation of domestic enterprises. Economic Review, (2): 75-91. (in Chinese)

[73]
Whittle A, 2020. Operationalizing the knowledge space: Theory, methods and insights for smart specialisation. Regional Studies, Regional Science, 7(1): 27-34.

[74]
Wigger C, 2022. Who with whom? Untangling the effect of high-skilled immigration on innovation. Journal of Economic Geography, 22(2): 449-476.

[75]
World Bank, 2010. Innovation Policy: A Guide for Developing Countries. World Bank.

[76]
Zhou Y, He C F, 2019. Evolution of urban export product in China. Acta Geographica Sinica, 74(6): 1097-1111. (in Chinese)

DOI

[77]
Zhu S J, Wang C, He C F, 2019. High-speed rail network and changing industrial dynamics in Chinese regions. International Regional Science Review, 42(5/6): 495-518.

Options
Outlines

/

〈 〉