Brain drain and economic growth: evidence of productivity growth European Journal of Government and Economics 10(2), December 2021, 128-145 European Journal of Government and Economics ISSN: 2254-7088 Brain drain and economic growth: evidence of productivity growth from brain circulation Woosik Yua* a Graduate School of International Studies, Seoul National University, Seoul, Republic of Korea * Corresponding author at: woosik@snu.ac.kr Abstract. This paper analyzes the effect of the so-called ‘brain drain’ on economic growth through the channel of growth in total factor productivity. We analyze panel data that measure the severity of brain drain, which are from IMD and the U.S. National Science Foundation. Our analysis shows that middle-income countries have more brain drain compared to the group of high-income countries. Also, emerging economies that grow fast tend to experience more brain drain. Our results from fixed effects regression models show that that brain drain has a significant and positive impact on economic growth, and the main channel is productivity growth. This can be considered as evidence of the positive effects of ‘brain circulation’, which is one of the brain drain phenomena that settlement of the talents in advanced countries can eventually help improve the productivity of home country by the sharing of advanced technologies and skills around them with colleagues in motherland. Therefore, a strategy of utilizing overseas resident talents should also be considered, alongside the brain-attraction policy. Keywords. brain drain, brain circulation, economic growth, total factor productivity JEL Codes. O11; O15; O47; F43 DOI. https://doi.org/10.17979/ejge.2021.10.1.7831 1. Introduction As globalization deepens, issues with brain drain are often discussed worldwide. Brain drain refers to a state or a phenomenon in which a country cannot utilize its effective labor of human resources or human capital. This phenomenon is largely divided into internal brain drain and external brain drain. Internal brain drain refers to the phenomenon in which domestic personnel with specialized skills are engaged in jobs unrelated to their majors, and external brain drain refers to the state in which competent talents of a country reside in other foreign countries, and their productivity contributes to the production of other countries. The brain drain to be addressed in this study is the second concept, the external brain drain. Whenever the ‘Brain Drain Index’ is announced at the International Institute for Management Development (IMD) in Switzerland every year, developing countries have continuously raised serious questions about the negative impact of brain drain on economic growth as well as national productivity and competitiveness. The argument is that the more serious the outflow of competent human resources to foreign countries, the less experts and research personnel in the home country, and this results in the weakening of the country’s research and technology competitiveness and reduction of innovation, which in turn adversely affects economic growth. mailto:woosik@snu.ac.kr https://doi.org/10.17979/ejge.2021.10.1.7831 Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 129 On the contrary, some argue that brain drain may have a positive effect on the national economy and competitiveness. In South Korea during the 1970s, the ‘brain drain’ had even been regarded as an act of ‘traitoring’ and the students residing abroad after the completion of their studies were harshly blamed. However, in recent days, as described in Saxenian (2005) and Teney (2021), the practical concept of ‘brain circulation’ came up in which the students or workers who studied and were employed abroad do not return to their home country right away, but return after working abroad and acquiring advanced learning and skills, or contributing to the motherland through various human networks built up while living abroad. In this regard, brain drain is considered a positive phenomenon as the first step for the brain circulation. As such, there are not many studies that quantitatively analyze the direct relationship between brain drain and economic growth using growth regression models, although brain drain can be a major factor in economic growth, either through the leakage of human capital or through the technological progress from brain circulation. Therefore, in this study, we intend to supplement the prior study by conducting an empirical analysis of the effects of external brain drain on economic growth using growth regression models, while looking at the time series trend for brain drain situations of several selected countries. 2. Literature review Studies on the negative effects of the flight of skilled workers on the national economy and competitiveness are conducted by Grubel and Scott (1966), Bhagwati and Hamada (1974), and McCulloch and Yellen (1977). In addition, Docquier and Rapoport (2008) presented empirical analysis that since the 1990s, the migration of skilled workers (brain drain) has increased by a greater margin than that of ordinary workers, and explained the factors behind the brain drain. However, studies by Mountford (1997), Stark, Helmenstein, and Prskawetz (1997), Beine, Docquier and Rapoport (2001), and Cinar and Docquier (2004) suggest that the migration of the talents to advanced technology-holding countries could have a positive impact on economic growth and competitiveness in the home country. According to Beine et al. (2001), immigration of talents can have both a ‘brain effect’ when returning home after acquiring advanced technology and a ‘drain effect’ that occurs when not returning home, and when brain effect is large enough under the open economy model, brain drain can have a positive effect on economic growth. Stark et al. (1997) also presented a research that said the dissemination of skills and knowledge gained in advanced technology-holding countries could help improve the home country’s technological prowess when migrant talent returns. Cinar et al. (2004) also argued that the home remittance of competent people who moved to developed countries had a positive effect on economic growth of the home country, as is often seen in the cases of immigration of workers from developing countries. Moreover, Vidal (1998) explored brain drain’s positive effect on human capital formation, and Saxenian (2005) described the process of brain circulation from brain drain with lowered transaction costs associated with digitization, and how it ultimately helps the development of Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 130 technology in home country, using the data of Chinese- and Indian-born engineers in the United States (U.S.). More recently, Kang, Lim, and Hwang (2013) provided empirical evidence of brain drain’s indirect but positive influence on national competitiveness through R&D investment and accumulated human capital. Also, Lodigiani, Marchiori, and Shen (2015) showed that brain drain could have both positive and negative effects on GDP per capita through the technology diffusion at origin from the high-skilled diaspora, depending on the region’s relative distance to the technological frontier. Amid this widespread agreement and objection to the severity of brain drain, this study seeks to infer the effects of brain drain on economic growth and technological growth (a possible channel to economic growth) of home country, using the data that proxies the degree of brain drain. 3. Motivation We were intrigued by the concept of ‘brain circulation’, one of the positive results from brain drain. As described in the above sections, talents who studied and were employed abroad can benefit their motherland through their acquisition of frontier technology and skills and sharing of their knowledge to their colleagues living in motherland, or co-workings with colleagues in home country. With an awareness that brain drain can adversely affect society in the sense of human capital flight, we hypothesized that brain drain could also have a positive effect on economic growth, as many other researchers claim. However, our main argument is more comprehensive. We hypothesize that brain drain helps achieve economic growth through productivity growth. Brain drain can induce brain circulation, and brain circulation helps home country’s technological progress. Technological progress ultimately improves productivity of production, and results in growth of income. Particularly, the total factor productivity (TFP) is predicted to be mostly affected by the brain drain via brain circulation and technology diffusion. TFP refers to the productivity of an economy’s composite inputs that are used to produce national output, and in neoclassical growth theory, economic growth rate converges to the growth rate of TFP in the long run (steady state). Thus, TFP can be a proxy for a country’s productive efficiency and technological level. TFP growth is one of the key components and channels for the economic growth in a simple growth model of 𝑌𝑌 = 𝑇𝑇𝑇𝑇 where 𝑌𝑌 is national income, 𝑇𝑇 is total factor productivity, 𝑇𝑇 is composite input, and 𝑁𝑁 is population. From the model, the income growth can be derived: 𝑌𝑌 𝑁𝑁 = 𝑇𝑇 ∙ 𝑋𝑋 𝑁𝑁 [1] Then, 𝑌𝑌 𝑁𝑁 = 𝑦𝑦 is national income per capita and 𝑋𝑋 𝑁𝑁 = 𝑥𝑥 is composite input per capita. When converting to growth terms, national income per capita growth (𝑔𝑔𝑦𝑦) can be expressed as the sum Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 131 of growth of total factor productivity (𝑔𝑔𝑇𝑇 ) and growth of per-capita composite input (𝑔𝑔𝑥𝑥 ) as described in equation [2]: 𝑔𝑔𝑦𝑦 = 𝑔𝑔𝑇𝑇 + 𝑔𝑔𝑥𝑥 [2] This implies that there are two main channels to which economic growth can be achieved: increasing growth rate of TFP or growth of composite input per capita. As the key hypothesis of this paper is that brain drain may enhance technology and productivity of home country through technological diffusion, testing the effects of brain drain on TFP growth will be the focus of our empirical analysis, as a channel to achieve income growth. 4. Empirical landscape Before analyzing the data, we will look at the main explanatory variables of the paper along with the descriptive statistics (see Table 1). In this study, two indicators are used as key variables that can proxy the degree of brain drain. The first data was established using the ‘Brain Drain Index’ survey data from the World Competitive Executive Opinion Survey of the International Institute for Management Development (IMD) in Switzerland. The criteria number in IMD Competitiveness Yearbook is ‘3.2.21’ and its title is ‘Brain drain’. The survey question is “Brain drain ([of ]well- educated and skilled people) does not hinder competitiveness in your economy”1 with a scale from zero to ten, and is asked to influential entrepreneurs worldwide. Recent years of data can be downloaded online at website of IMD World Competitiveness Online2, and the past data can be found in IMD World Competitiveness Yearbook series. As mentioned above, the index values of the original data are distributed from 0 to 10, and the more severe the brain drain is, the more negative impact it has on economic and management activities, and the closer it becomes to zero. In this study, the corresponding brain drain index values were subtracted from 10, so that the higher the brain drain is, the higher the values are. This newly defined IMD brain drain index is referred to as the ‘brain drain index’ in this paper. The indicator covers data from 61 countries and provides a total of 22 years of time series values from 1995 to 2016. This data is important in the sense that it can give information on how the entrepreneurs feel about brain drain of the country, as the entrepreneurs are the front-line agents who put the up-to-date technologies into practical uses, and they are eager to employ the talented the most. 1 The IMD World Talent Ranking Methodology (2019) 2 https://worldcompetitiveness.imd.org/CustomSearch https://worldcompetitiveness.imd.org/CustomSearch Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 132 Table 1. Descriptive statistics of brain drain index and share of doctorate recipients staying in the U.S. Sample Size Mean Standard Deviation Minimum Maximum No. of Countries Period Brain drain index3 1,174 4.779 1.582 1.04 8.85 61 1995~2016 PhD ratios staying in the U.S. 287 0.628 0.167 0.10 0.92 41 2010~2016 Brain drain index 210 5.379 1.322 2.75 8.85 30 (common sample counties) 2010~2016 (common sample period) PhD ratios staying in the U.S. 210 0.609 0.145 0.21 0.92 Note. Third and fourth rows are for common sample period (2010~2016) and sample country list (30 countries). Source: IMD World Competitiveness Executive Opinion Survey, US National Science Foundation Table 2. Rankings of IMD Brain Drain Index by Country in 20164. Rank Country B.D. Index Rank Country B.D. Index Rank Country B.D. Index 1 Hungary 8.44 22 Lithuania 5.97 43 Canada 4.20 2 Venezuela 8.44 23 Argentina 5.90 44 Israel 4.17 3 Bulgaria 8.04 24 Mongol 5.83 45 Iceland 4.08 4 Ukraine 8.00 25 Jordan 5.78 46 Chile 3.94 5 Croatia 7.84 26 Brazil 5.73 47 Belgium 3.89 6 South Africa 7.23 27 Japan 5.58 48 Indonesia 3.76 7 Slovakia 7.13 28 Spain 5.49 49 Germany 3.64 8 Kazakhstan 7.05 29 Mexico 5.23 50 Britain 3.58 9 Slovenia 6.85 30 France 5.22 51 Singapore 3.51 10 Russia 6.75 31 Italy 5.14 52 Finland 3.46 11 Greece 6.68 32 India 5.09 53 UAE 3.38 12 Portugal 6.61 33 Ireland 4.93 54 Luxembourg 3.35 13 Latvia 6.58 34 Rumania 4.55 55 Netherlands 2.80 14 Poland 6.58 35 Turkey 4.49 56 Hong Kong 2.78 15 Estonia 6.51 36 Cech 4.46 57 Denmark 2.75 16 South Korea 6.40 37 Thailand 4.41 58 Sweden 2.70 17 Taiwan 6.31 38 Qatar 4.41 59 USA 2.67 18 Philippines 6.23 39 Austria 4.35 60 Swiss 2.03 19 Columbia 6.07 40 Australia 4.31 61 Norway 1.68 20 China 6.05 41 New Zealand 4.30 21 Peru 5.98 42 Malaysia 4.28 Note. Source: IMD World Competitiveness Executive Opinion Survey. 3 Values were subtracted from 10. 4 The higher the value, the more serious the degree of brain drain is. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 133 The second brain drain proxy data are from the U.S. National Science Foundation's ‘Doctorate Recipients from U.S. Universities’ dataset, which can be downloaded from NSF website5. More specifically, the data we used are taken from the ‘Share of doctorate recipients with temporary visas intending to stay in the U.S. after doctorate receipt’. In this paper, we will refer all of the doctorate degrees in this data as ‘Doctor of Philosophy (Ph.D.)’, though we are aware that not all doctorate-level degrees are Ph.D. degrees. Thus, we will call this variable as ‘Share of Ph.D.’s staying in the U.S.’ for convenience. The variable takes a value of zero to one, and covers data of U.S. Ph.D. recipients from a total of 41 countries over the world with the seven-year time series availability from 2010 to 2016. The U.S. universities’ Ph.D. graduates, who are considered to be the frontiers of advanced technology, can be classified as the ‘top-class’ professionals in their fields, and therefore, we postulate that the higher the percentage of them remaining in the U.S. is, the higher the level of brain drain their home countries face. 4.1. IMD Brain Drain Index data First, Table 2 of IMD brain drain index rankings by country shows that Hungary, Venezuela, and Bulgaria are ranked the first, second and the third as of 2016. Not surprisingly, the upper-ranked countries are mostly middle-income or lower-middle-income countries, rather than the richest or poorest countries. As described in the table above, seven out of the top 10 countries, i.e., Hungary, Bulgaria, Ukraine, Croatia, Slovakia, Slovenia, and Russia, are Eastern European countries. This is likely to be the result of the recent increase in joins of these East European countries into the European Union, acquiring domestic labor’s freedom of movement and employment in advanced countries such as U.K., Germany, and France. We chose four countries, South Korea, China, Japan, and U.S. as samples to review time- series trends of individual economies. We chose China because it is one of the fast-growing countries with largest population size in the world, U.S. because of its largest GDP and highest technology level with top-class quality of human capital, South Korea because it is known as a miraculous example case where education and human capital played important roles to escape from the ‘middle-income trap’ and achieve sustained rapid growth, i.e., ‘six-percent six-decade’ growth. Finally, Japan was selected because it is widely known that it has become a trend in Japan that Japanese students do not pursue doctorate program abroad, and at the same time, the economy has been slowing down for decades despite its large size. 5 https://www.nsf.gov/statistics/doctorates/ https://www.nsf.gov/statistics/doctorates/ Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 134 Figure 1. Time series of IMD brain drain index of South Korea, China, Japan, and U.S.A. Source: IMD World Competitiveness Executive Opinion Survey Figure 1 shows that China’s IMD brain drain index has been somewhat flat with some up and downs, but decreasing since 2013, while that of Japan has been rising. Moreover, in recent years, South Korea’s IMD brain drain index exceeded that of China (in years 2012, 2014, 2015, and 2016). In 2016 South Korea was ranked 16th at 6.40, surpassing China’s 6.05 (20th). Also, it is interesting to notice that South Korea’s brain drain has leaped dramatically during the global financial crisis, in years 2007 and 2008, and since then, the trend is more or less flat, aside from the one-time negative shock in 2013. 4.2. NSF Doctorate Recipients from U.S. Universities data Next, Table 3, the 2016 U.S. NSF’s ranking of the share of Ph.D. recipients staying in the U.S. after their completion in the course, tells that Iran, Bangladesh and Venezuela are at the top of the list. Unlike the IMD brain drain index, six of the top 10 countries are made up of Asian countries, Iran, Bangladesh, Nepal, India, China, and Taiwan, while only two Eastern European countries, Bulgaria and Romania, are listed in the top 10. Also, Venezuela, a South American country, and Nigeria, an African country, are ranked third and eighth, respectively. This is attributed to the absence of the effect of Eastern European countries’ membership status in the European Union. In other words, there are no visa problems for U.S. Ph.D. holders from Eastern Europe who have gone to settle down in the advanced European countries such as U.K. and Germany, and they are not counted and not reflected in this ranking. This is because the top universities and research institutes in Europe do not lag far behind compared to those of the U.S. in terms of research environment and the quality of co-workers. Global Financial Crisis 1 2 3 4 5 6 7 8 IM D B ra in D ra in In de x (1 0 - o rig in al v al ue ) 19 95 19 98 20 01 20 04 20 07 20 10 20 13 20 16 Year South Korea China Japan USA Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 135 Table 3. Rankings of Share of new Ph.D.’s Staying in the U.S. by Country in 20166. Rank Country No. of US PhD Recipients Share of Staying in the US Rank Country No. of US PhD Recipients Share of Staying in the US 1 Iran 695 0.911 22 Italy 167 0.659 2 Bangladesh 185 0.903 23 Kenya 61 0.656 3 Venezuela 27 0.889 24 Argentina 63 0.635 4 Bulgaria 44 0.886 25 Spain 73 0.616 5 Nepal 226 0.885 26 Columbia 187 0.615 6 Romania 51 0.882 27 Mexico 222 0.613 7 India 2,203 0.872 28 Germany 183 0.612 8 Nigeria 111 0.847 29 Turkey 472 0.576 9 China 5,534 0.809 30 Brazil 155 0.568 10 Taiwan 593 0.779 31 Canada 407 0.565 11 Philippines 51 0.765 32 Britain 115 0.548 12 Vietnam 124 0.718 33 Japan 166 0.530 13 Greece 84 0.714 34 Israel 83 0.518 14 Ukraine 31 0.710 35 Pakistan 102 0.451 15 Lebanon 58 0.707 36 Jordan 98 0.449 16 Ghana 87 0.701 37 Indonesia 92 0.435 17 Russia 108 0.694 38 Singapore 90 0.389 18 South Korea 1,228 0.668 39 Chile 130 0.323 19 Australia 42 0.667 40 Thailand 185 0.303 20 France 105 0.667 41 Saudi Arabia 238 0.101 21 Egypt 118 0.661 Note. Source: U.S. National Science Foundation Interestingly, Figure 2 shows a different pattern in the number of Ph.D.’s between South Korea and China. Korea’s share of doctorate recipients staying in the U.S. sharply increases since 2013 while that of China continuously drops since 2012. Moreover, the number of new Ph.D. graduates from Korea decreases since 2012, while that of China has risen continuously and significantly at least from the year 2010, showing the seemingly opposite trends. In Korea, a high percentage (52.4%) of doctorate holders are starting their research career as non-regular workers (Song et al., 2016)7, which are often discriminated against not only by the threat of job security but also by limitations to research activities, wages, children’s education supports, and welfare systems compared to the full-time positioned workers. On the other hand, since the 1990s, Chinese government started an initiative called ‘Thousand Talents Plan (TTP)’ or so-called ‘The Salmon Project’, of which the goal is to bring back 1000 talents to China who have studied in advanced countries by providing exceptional benefits 6 The higher the value, the more serious the degree of brain drain is. 7 In 2016, 75.5% of South Korea’s new doctorate recipients are employed or confirmed to be employed, and among them, only 63.1% are employed for full-time regular positions, and 36.9% are employed for temporary (non-regular) positions. In sum, more than half (52.4%) of South Korea’s doctorate recipients are either unemployed or employed for temporary jobs. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 136 including financial reward of one million yuan per person, housings, medical insurance, and even the high-quality education opportunities for their children. (Oh, 2016). This plan, which is also translated as the ‘The Recruiting Program of Global Experts’, aims to recruit world-class scholars, entrepreneurs, professional technicians and manager-level talents over the next five to ten years and deploy them to state-driven research projects, state-owned enterprises and banks, and industrial technology complexes (Lee, 2018). As a result of these efforts, many Chinese doctorate recipients appear to be returning to their home countries to settle down, and graduate students actively seek for Ph.D. study opportunities in the U.S. because they are aware of these rewards. Figure 2. Number of U.S. Ph.D. recipients and the share of them intending to stay in the U.S. Source: U.S. National Science Foundation On the other hand, for Japan and Iran, the number of Ph.D. recipients and the share of Ph.D.’s staying in the U.S. move in the same direction. In Japan, both the number of Ph.D. recipients and the share of Ph.D.’s staying in the U.S. show downward patterns by 2015. In the case of Iran, both variables show a pattern of upward movement, and the gradients are steep. In particular, the number of U.S. doctorate recipients from Iran grows rapidly from 147 in year 2010 to 695 in 2016. Iran was selected as a new sample for analysis simply because its ‘share of Ph.D.’s staying in the U.S.’ was the highest among the sample countries as of 2016. 12 00 13 00 14 00 15 00 .5 8 .6 .6 2 .6 4 .6 6 2010 2012 2014 2016 South Korea 35 00 40 00 45 00 50 00 55 00 .8 1 .8 15 .8 2 .8 25 2010 2012 2014 2016 China 16 0 18 0 20 0 22 0 24 0 .4 .4 5 .5 .5 5 2010 2012 2014 2016 Year Japan 0 20 0 40 0 60 0 80 0 .8 4 .8 6 .8 8 .9 .9 2 2010 2012 2014 2016 Year Iran - No. of New Ph.D.s (Straight) - Share of Staying (Dotted) Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 137 Table 4. Pearson correlation analysis between brain drain index and selected macroeconomic variable. Variables8 Correlation Coefficient P-value Brain Drain Index Share of Staying in the US 0.519 0.000 Brain Drain Index Total Factor Productivity -0.477 0.000 Brain Drain Index GDP per Capita -0.452 0.000 Brain Drain Index Human Capital Index -0.271 0.000 Share of Staying in the US Total Factor Productivity -0.283 0.000 Share of Staying in the US GDP per Capita -0.415 0.000 Share of Staying in the US Human Capital Index -0.255 0.000 4.3. Correlation Analysis of Brain Drain Indicators Table 4 shows the results of Pearson correlation tests between the brain drain variables and key macroeconomic variables. We are presenting this correlation analysis to show the landscape of macroeconomic data and their relationships with brain drain variables before moving on to the regression analysis. First, there was a strong correlation between IMD brain drain index and the share of Ph.D. recipients staying in the U.S. The higher the share of Ph.D.’s staying in the U.S. is, the higher the brain drain index is. In addition, both variables have negative correlation with the key macroeconomic growth variables, i.e., total factor productivity, per capita GDP, and human capital index. The more technically advanced, or the more productive the countries are, the richer the countries are, and the higher the human capital index is, the less brain drain there is. 5. Empirical analysis 5.1. Model Specification In this study, we use these two variables described earlier, IMD’s ‘brain drain index’ and NSF’s ‘share of Ph.D.’s staying in the U.S.’ as key explanatory variables and explain the economic growth and total factor productivity growth using fixed-effects linear regression models. The growth regression models are represented by the following six reduced-form equations such that: 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 1: 𝑔𝑔_𝑦𝑦𝑖𝑖 ,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖 ,𝑡𝑡 + 𝛾𝛾𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [3] 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 2: 𝑔𝑔_𝑦𝑦𝑖𝑖 ,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛾𝛾𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [4] 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 3: 𝑔𝑔_𝑦𝑦𝑖𝑖 ,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛾𝛾0𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑡𝑡 + 𝛾𝛾1𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [5] 8 Total factor productivity index is ‘ctfp’ from Penn World Table (PWT) 9.1 (Feenstra RC, Inklaar R & Timmer MP, 2015), GDP per capita is ‘rgdpe/pop’ from PWT, human capital index is ‘hc’ from PWT. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 138 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 4: 𝑔𝑔_𝑇𝑇𝑖𝑖,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡 + 𝛾𝛾𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [6] 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 5: 𝑔𝑔_𝑇𝑇𝑖𝑖,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛾𝛾𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [7] 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 6: 𝑔𝑔_𝑇𝑇𝑖𝑖,𝑡𝑡~𝑡𝑡+5 = 𝛽𝛽0 + 𝛽𝛽1𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 + 𝛾𝛾0𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑡𝑡 + 𝛾𝛾1𝑇𝑇𝑖𝑖,𝑡𝑡 + 𝜈𝜈𝑡𝑡 + 𝜐𝜐𝑖𝑖 + 𝜀𝜀𝑖𝑖.𝑡𝑡 [8] where 𝑔𝑔_𝑦𝑦𝑖𝑖,𝑡𝑡~𝑡𝑡+5 and 𝑔𝑔_𝑇𝑇𝑖𝑖,𝑡𝑡~𝑡𝑡+5 are average annual growth rates of home country 𝑖𝑖 ’s income (per-capita GDP) and total factor productivity from year 𝑡𝑡 to year 𝑡𝑡 + 5, respectively. 𝐵𝐵𝐵𝐵𝐵𝐵𝑖𝑖,𝑡𝑡 and 𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖,𝑡𝑡 denote country 𝑖𝑖’s Brain Drain Index and share of doctoral graduates staying in the U.S. at year 𝑡𝑡, respectively. 𝑁𝑁𝑁𝑁𝑁𝑁𝑖𝑖,𝑡𝑡 represents the number of U.S. Ph.D. recipients of country 𝑖𝑖 in year 𝑡𝑡. This term is included to control for the differences in capacities of sample countries to send their students for U.S. doctoral programs. 𝑇𝑇𝑖𝑖,𝑡𝑡 is a set of external environmental and policy variables that may affect the dependent variables, namely, lagged income (lagged GDP per capita), years of schooling, fertility rate, investment to GDP ratio, government spending to GDP ratio, trade openness, and terms of trade change. These variables are often used in literature with conventional growth regression models such as Barro (2016) and Lee (2017). Moreover, 𝜈𝜈𝑡𝑡 and 𝜐𝜐𝑖𝑖 are included to control for the year- and country-specific fixed effects, respectively. 𝜀𝜀𝑖𝑖.𝑡𝑡 represents the error term. The followings are summary statistics of variables used in regression analysis. Table 5 is a summary statistics table for the regression models that use IMD brain drain data, and Table 6 is the table for the models that use NSF Doctorate Recipients from U.S. Universities data. Since IMD data contains more country samples with a longer time span, the regression models that use this data have more observations than those that use NSF data. The correlation matrices of all used variables are presented in the appendix (see Table A1 and Table A2). Table 5. Summary Statistics for Model 1 and Model 4 (1993~2017, 61 Countries).9 Variable Obs. Mean Std. Dev. Min Max GDP per Capita Growth* 1,342 0.034 0.035 -0.133 0.230 TFP Growth* 1,320 0.003 0.028 -0.146 0.125 Brain drain index 1,361 4.80 1.52 1.33 8.56 Lagged income** (USD) 1,525 23,589 17,852 1,170 136,890 Years of schooling 1,464 9.95 2.20 3.29 13.55 Fertility rate 1,500 1.90 0.64 0.93 4.93 Investment/GDP 1,525 0.248 0.069 0.002 0.640 Government spending/GDP 1,525 0.182 0.064 0.037 0.423 Trade openness 1,525 0.800 0.693 0.068 5.457 Terms of trade change 1,525 0.005 0.064 -0.060 1.789 *: Average annual growth rate from year t to year t+5. **: Value of income for year t-5 9 GDP per capita (or income) is ‘rgdpe/pop’ from PWT, years of schooling is ‘yr_sch’ from PWT, investment to GDP ratio is ‘csh_i’ from PWT, government spending to GDP ratio is from ‘csh_g’ from PWT, trade to GDP ratio is ‘csh_x-csh_m’ from PWT, the terms of trade is ‘pl_x/pl_m’ from PWT, and fertility rate is from World Development Indicator’s ‘Fertility Rate’ variable. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 139 Table 6. Summary Statistics for Models 2, 3, 5, and 6 (2008~2017, 41 Countries) Variable Obs. Mean Std. Dev. Min Max GDP per Capita Growth* 287 0.022 0.033 -0.133 0.101 TFP Growth* 245 -0.002 0.027 -0.146 0.056 Share of staying in US 410 0.626 0.160 0.122 0.904 Number of US Ph.D. recipients 410 333 780 28 5,454 Lagged income** 410 18,570 14,789 1271 72,583 Years of schooling 360 9.11 2.69 3.11 13.55 Fertility rate 400 2.23 0.96 1.19 5.90 Investment/GDP 410 0.238 0.075 0.076 0.501 Government spending/GDP 410 0.165 0.047 0.059 0.336 Trade openness 410 0.528 0.451 0.084 3.251 Terms of trade change 410 0.013 0.122 -0.042 1.789 *: Average annual growth rate from year t to year t+5 **: Value of income for year t-5 5.2. Results Table 7 shows the results of linear regression analysis that explain the growth rate of GDP per capita and the growth rate of total factor productivity. All variables take five-year (plus and minus 2 years) moving average forms to rule out the short-term shocks and capture the long-term trend. As mentioned above, the dependent variables, GDP per capita growth and TFP growth, are generated by calculating growth rates from year 𝑡𝑡 to year 𝑡𝑡 + 5 to see the future (five-year-later) effects of brain drain. The control variable ‘Ln (Lagged Income)’ took the lagged form, which uses the value of the year five years ahead (𝑡𝑡 − 5). The analysis shows that both the brain drain index and the share of Ph.D.’s staying in the U.S. have significant and positive effects on economic growth and productivity growth. Models that have TFP growth as the dependent variable (models 4, 5, and 6) show greater regression coefficients for key explanatory variables than the models with GDP per capita growth. This implies that brain drain more affects TFP growth and suggest that the main channel of brain drain affecting economic growth is likely to be the productivity growth. In other words, since brain drain does not directly affect income growth, but via TFP growth, the regression coefficient on TFP turns out to be stronger, and this is on the same line with our hypothesis. Also, the differences of fit (R- squared and adjusted R-squared) are higher for models with ‘share of staying in U.S.’ variable (models 2, 3, 5 and 6), implying that these models better explain the effects of brain drain compared to other models. In addition, the inclusion of controls for the number of U.S. Ph.D. recipients does not alter the regression result, suggesting that the possible sample selection issues related to the magnitude of doctorate recipients are well-controlled. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 140 Table 7. Results of linear regression of economic and total productivity growth. Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 GDP per Capita Growth (t+5) TFP Growth (t+5) Brain Drain Index 0.012** (0.012) 0.013*** (0.003) Share of Staying in US 0.062** (0.015) 0.063** (0.021) 0.080*** (0.003) 0.082*** (0.005) Ln (Number of US Ph.D. Recipients) 0.0004 0.002 Ln (Lagged Income) -0.041*** 0.010 0.010 -0.028*** -0.003 -0.003 Ln (Years of Schooling) -0.033 -0.003 -0.004 -0.015 0.011 0.006 Ln (Fertility Rate) -0.045*** -0.014 -0.014 -0.021* -0.008 -0.006 Ln (Investment/GDP) -0.031*** -0.016 -0.016 0.013** 0.017 0.018 Ln (Gov. Sp. /GDP) -0.003 0.070*** 0.070*** 0.014*** 0.081*** 0.081*** Ln (Trade Openness) 0.031*** -0.073*** -0.073*** 0.030*** -0.046* -0.047* Terms of Trade Chg. -0.096 -0.257** -0.257** 0.164* -0.072 -0.071 Country & Year FE Included Included Included Included Included Included Observations 1,156 273 273 1,150 238 238 R-squared 0.406 0.801 0.801 0.308 0.735 0.735 Adjusted R-squared 0.357 0.754 0.753 0.252 0.669 0.667 Note. p < 0.1, ** p < 0.05, *** p < 0.01; Values in parenthesis are p-values; ‘Brain Drain Index’ and ‘Share of Staying in U.S.’ are in natural logarithm terms. However, in model 1, investment to GDP ratio shows a strong and negative relationship with per capita GDP growth rate, and in models 2 and 3, trade openness has strong and negative regression coefficients, and thus, these models seem to be disqualifying to be established as benchmark models and show possibilities of needs for other control variables. On the other hand, the correlation coefficients of models 4, 5, and 6 seem to be consistent with the conventional growth regression models. Lastly, the authors are acknowledging the potential weakness of the result that even with the year-specific fixed effects method, due to the moving average forms of the variables, there may be a possibility of issues with autocorrelation across time. Also, a more thorough analysis can be conducted with the inclusion of factors that determines the decision of the talented to study abroad in the U.S., as well as the decision to stay in the U.S. or return to the home country. For example, the following factors, which are not considered in this paper’s empirical analysis, may improve the models: the baseline living conditions of home country, job and income prospects in the U.S. compared to those of home countries, cultural differences, cost of living in the home countries. Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 141 6. Conclusion According to our international comparisons of brain drain, the patterns of brain drain vary from country to country. Particularly, South Korea and China show stark differences. South Korea’s human resource outflow continuously increases and deepens while China's talent outflow is showing a marked decline. Korean media often argue that the deepening of this pattern will reduce the relative quality of South Korea’s human capital, and that of China will continue to accumulate and surpass Korea’s level soon. In general, middle-income countries rather than high- income group have more brain drain. Also, emerging economies that grow fast tend to experience more brain drain. However, the results of growth regression analysis show that so-called brain drain has a significant impact on both economic growth and productivity (TFP) growth, implying that brain drain does not always affect the economy in a negative way. Moreover, brain drain has shown to have a more positive association with productivity growth, suggesting that the main channel of brain drain inducing economic growth is likely to be the productivity growth. This can be considered as evidence of the positive effects of ‘brain circulation’, which is one of the brain drain phenomenon that settlement of the talents in advanced countries can eventually help improve the productivity of home country by sharing of advanced technologies and skills around them with colleagues in motherland. Therefore, from a policy point of view, a two-track strategy is suggested to simultaneously pursue measures to maximize the improvement of national level of technology, which emphasizes the utilization of overseas resident talents alongside the brain-attraction policy. In this context, developing nations should try avoiding the unconditional brain-attraction policy. Furthermore, the support policies to raise the quality of research institutes and universities to the world-class level, as well as policies that encourage talents-sharing with global frontier firms will also help improve the home country’s technological prowess through technology and knowledge diffusion in the long run, as these policies can attract and embrace skilled overseas brain groups. References Barro, R. J. (2016). Economic Growth and Convergence, Applied to China. China and World Economy. 5(24):5-19. https://doi.org/10.1111/cwe.12172 Beine, M., Docquier, F., & Rapoport, H. (2001). Brain Drain and Economic Growth: Theory and Evidence. Journal of Development Economics, 64(1), 275-289. https://doi.org/10.1016/S0304- 3878(00)00133-4 Bénassy, J. P., & Brezis, E. S. (2013). Brain Drain and Development Traps. Journal of Development Economics, 102, 15-22. https://doi.org/10.1016/j.jdeveco.2012.11.002 https://doi.org/10.1111/cwe.12172 https://doi.org/10.1016/S0304-3878(00)00133-4 https://doi.org/10.1016/S0304-3878(00)00133-4 https://doi.org/10.1016/j.jdeveco.2012.11.002 Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 142 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. https://doi.org/10.1016/0304-3878(74)90020-0 Cinar, D., & Docquier, F. (2004). Brain Drain and Remittances: Implications for the Source Country. Brussels Economic Review, 47(1), 103-118. Docquier, F., & Rapoport, H. (2008). Brain Drain and Human Capital Formation in Developing Countries: Winners and Losers. The Economic Journal, 118(528), 631-652. https://doi.org/10.1111/j.1468-0297.2008.02135.x Feenstra R. C., Inklaar R., & Timmer M. P. (2015). The Next Generation of the Penn World Table. American Economic Review, 105(10), 3150–3182. https://doi.org/10.1257/aer.20130954 Grubel, H. B., & Scott, A. D. (1966). The International Flow of Human Capital. American Economic Review, 56(1/2), 268-274. IMD World Competitiveness Executive Opinion Survey IMD World Competitiveness Yearbook Kang, D. K., E.-S Lim, & J. Hwang. (2013). An Exploratory Study on the Relationship between Brain Drain and National Competitiveness. IOM MRTC Working Paper Series, No. 2013-05. IOM Migration Research and Training Centre. Republic of Korea. Lee, J. W. (2017). China's Economic Growth and Convergence. World Economy. 40:2455-2474. https://doi.org/10.1111/twec.12554 Lee, K. (2018). Chinese Science Paper Results Exceed the U.S. The Science Times. URL: https://www.sciencetimes.co.kr/?news=%EC%A4%91%EA%B5%AD-%EA%B3%BC%ED%9 5%99%EB%85%BC%EB%AC%B8-%EC%84%B1%EA%B3%BC-%EB%AF%B8%EA%B5% AD-%EC%B6%94%EC%9B%94%ED%95%9C%EB%8B%A4 (Retrieved on June 3, 2019. In Korean, title translated by the author). Lodigiani, E., Marchiori, L., & Shen, I.-L. (2015). Revisiting the Brain Drain Literature with Insights from a Dynamic General Equilibrium World Model. The World Economy, 39(4), 557–573. https://doi.org/10.1111/twec.12269 McCulloch, R., & Yellen, J. T. (1977). Factor Mobility, Regional Development and the Distribution of Income. Journal of Political Economy, 85(1), 79-96. https://doi.org/10.1086/260546 Mountford, A. (1997). Can a Brain Drain Be Good for Growth in the Source Economy? Journal of Development Economics, 53(2), 287-303. https://doi.org/10.1016/S0304-3878(97)00021-7 Oh, H. (2016). Leakage of Korean Talents. Chosun Ilbo. http://news.chosun.com/site/data/html _dir/2016/08/08/2016080801521.html (Retrieved on September 26, 2019. In Korean, title translated by the author). Saxenian, A. (2005). From Brain Drain to Brain Circulation: Transnational Communities and Regional Upgrading in India and China. Studies in Comparative International Development, 40(2), 35-61. https://doi.org/10.1007/BF02686293 Song, C. Y., Oh, H. Y., Kim, H. J., & Lee, E. H. (2016). Korea Survey of Earned Doctorates, 2016. Korea Research Institute for Vocational Education & Training, KRIVET. Republic of Korea. https://doi.org/10.1016/0304-3878(74)90020-0 https://doi.org/10.1111/j.1468-0297.2008.02135.x https://doi.org/10.1257/aer.20130954 https://doi.org/10.1111/twec.12554 https://www.sciencetimes.co.kr/?news=%EC%A4%91%EA%B5%AD-%EA%B3%BC%ED%95%99%EB%85%BC%EB%AC%B8-%EC%84%B1%EA%B3%BC-%EB%AF%B8%EA%B5%AD-%EC%B6%94%EC%9B%94%ED%95%9C%EB%8B%A4 https://www.sciencetimes.co.kr/?news=%EC%A4%91%EA%B5%AD-%EA%B3%BC%ED%95%99%EB%85%BC%EB%AC%B8-%EC%84%B1%EA%B3%BC-%EB%AF%B8%EA%B5%AD-%EC%B6%94%EC%9B%94%ED%95%9C%EB%8B%A4 https://www.sciencetimes.co.kr/?news=%EC%A4%91%EA%B5%AD-%EA%B3%BC%ED%95%99%EB%85%BC%EB%AC%B8-%EC%84%B1%EA%B3%BC-%EB%AF%B8%EA%B5%AD-%EC%B6%94%EC%9B%94%ED%95%9C%EB%8B%A4 https://doi.org/10.1111/twec.12269 https://doi.org/10.1086/260546 https://doi.org/10.1016/S0304-3878(97)00021-7 http://news.chosun.com/site/data/html_dir/2016/08/08/2016080801521.html http://news.chosun.com/site/data/html_dir/2016/08/08/2016080801521.html https://doi.org/10.1007/BF02686293 Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 143 Stark, O., Helmenstein, C., & Prskawetz, A. (1997). A Brain Gain with a Brain Drain. Economics letters, 55(2), 227-234. https://doi.org/10.1016/S0165-1765(97)00085-2 Survey of Earned Doctorates, Doctorate Recipients from U.S. Universities, National Science Foundation (NSF), URL: https://www.nsf.gov/statistics/doctorates/ Teney, C. (2021). Immigration of Highly Skilled European Professionals to Germany: Intra-EU Brain Gain or Brain Circulation? Innovation: The European Journal of Social Science Research, 34(1), 69-92. https://doi.org/10.1080/13511610.2019.1578197 The IMD World Talent Ranking Methodology (2019). https://www.imd.org/globalassets/wcc/ docs/release-2019/talent/imd_world_talent_ranking_methodology_2019.pdf (Retrieved on November 12, 2021). Vidal, J.-P. (1998). The Effect of Emigration on Human Capital Formation. Journal of Population Economics, 11(4), 589-600. https://doi.org/10.1007/s001480050086 https://doi.org/10.1016/S0165-1765(97)00085-2 https://www.nsf.gov/statistics/doctorates/ https://doi.org/10.1080/13511610.2019.1578197 https://www.imd.org/globalassets/wcc/docs/release-2019/talent/imd_world_talent_ranking_methodology_2019.pdf https://www.imd.org/globalassets/wcc/docs/release-2019/talent/imd_world_talent_ranking_methodology_2019.pdf https://doi.org/10.1007/s001480050086 Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 144 Appendix Table A1. Correlation Matrix for Variables of Models 1 and 4. Var1 Var2 Var3 Var4 Var5 Var6 Var7 Var8 Var9 Var2 0.667 1.000 (0.000) Var3 0.136 0.060 1.000 (0.000) (0.041) Var4 -0.327 -0.219 -0.448 1.000 (0.000) (0.000) (0.000) Var5 -0.170 0.039 -0.261 0.427 1.000 (0.000) (0.153) (0.000) (0.000) Var6 0.041 0.008 0.163 -0.231 -0.509 1.000 (0.136) (0.779) (0.000) (0.000) (0.000) Var7 -0.181 -0.085 -0.353 0.409 0.008 -0.087 1.000 (0.000) (0.002) (0.000) (0.000) (0.766) (0.001) Var8 0.268 0.143 0.361 -0.405 0.094 -0.141 -0.413 1.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Var9 -0.022 -0.025 -0.244 0.397 0.266 -0.305 0.391 -0.155 1.000 (0.412) (0.371) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Var10 -0.069 -0.105 0.128 -0.028 -0.035 0.038 -0.017 0.038 -0.045 (0.012) (0.000) (0.000) (0.283) (0.177) (0.139) (0.503) (0.134) (0.077) Note. Values in parentheses are p-values; Var1: GDP per capita growth; Var2: TFP growth; Var3: IMD brain drain index; Var4: lagged income; Var5: years of schooling; Var6: fertility rate; Var7: investment to GDP ratio; Var8: government spending to GDP ratio; Var9: trade openness; Var10: terms of trade change Woosik Yu / European Journal of Government and Economics 10(2), December 2021, 128-145 145 Table A2. Correlation Matrix for Variables of Models 2, 3, 5, and 6. Var1 Var2 Var3 Var4 Var5 Var6 Var7 Var8 Var9 Var10 Var2 0.866 1.000 (0.000) Var3 0.154 0.057 1.000 (0.009) (0.378) Var4 0.198 0.031 0.251 1.000 (0.001) (0.635) (0.000) Var5 -0.084 0.094 -0.400 -0.098 1.000 (0.155) (0.144) (0.000) (0.048) Var6 -0.032 0.209 -0.252 -0.112 0.747 1.000 (0.592) (0.001) (0.000) (0.034) (0.000) Var7 -0.196 -0.294 0.082 -0.145 -0.487 -0.560 1.000 (0.001) (0.000) (0.104) (0.004) (0.000) (0.000) Var8 0.157 0.003 -0.079 0.568 0.222 -0.005 -0.330 1.000 (0.008) (0.967) (0.112) (0.000) (0.000) (0.930) (0.000) Var9 -0.058 0.287 -0.084 -0.156 0.233 0.523 -0.361 -0.222 1.000 (0.328) (0.000) (0.090) (0.002) (0.000) (0.000) (0.000) (0.000) Var10 0.019 0.042 -0.323 -0.058 0.688 0.488 -0.361 0.320 0.009 1.000 (0.753) (0.512) (0.000) (0.246) (0.000) (0.000) (0.000) (0.000) (0.862) Var11 -0.034 0.060 0.108 -0.036 -0.020 -0.004 0.002 -0.003 0.072 -0.072 (0.570) (0.349) (0.029) (0.463) (0.694) (0.934) (0.963) (0.948) (0.146) (0.144) Note. Values in parentheses are p-values; Var1: GDP per capita growth; Var2: TFP growth; Var3: share of staying in U.S.; Var4: number of U.S. Ph.D. recipients; Var5: lagged income; Var6: years of schooling; Var7: fertility rate; Var8: investment to GDP ratio; Var9: government spending to GDP ratio; Var10: trade openness; Var11: terms of trade change Brain drain and economic growth: evidence of productivity growthfrom brain circulation 1. Introduction 2. Literature review 3. Motivation 4. Empirical landscape 5. Empirical analysis 6. Conclusion References