The Italian brain drain: cream and milk

High Educ https://doi.org/10.1007/s10734-018-0292-8

The Italian brain drain: cream and milk Mattia Cattaneo 1,2 & Paolo Malighetti 1,2 & Stefano Paleari 1,2

# Springer Nature B.V. 2018

Abstract This study analyzes the scientific profiles of doctorate holders who migrate in search of better job opportunities as a response to the increased temporariness and uncertainty of employment in their native country. Analyzing the career trajectories of the population of research-active Italian doctorate holders in economics, finance, and business management who graduated in the years 2008–2010, we find that those who are more likely to move abroad are those with the relatively best and worst research performances before migrating. Instead, those remaining in Italy are found to be on average of good scientific quality. Further, we find that the best performers target cross-border reputable universities for more prestigious careers, while the relatively lower performers generally move internationally to non-research-oriented institutions simply to look for an occupation. Migrant doctorate holders with average research performances do not differ from those staying in their native country. Our results have important policy implications as they contradict the popular belief that only the cream of the talented brains flee Italy. Keywords Brain drain . Migration policy . Human capital . Italy . Doctorates . Academic career

Introduction The brain drain has increasingly become a critical phenomenon that policy makers cannot definitively ignore due to its important detrimental effects on the country of emigration (Carrington and Detragiache 1999). Although this concept was initially used to identify British engineers and scientists moving to the USA in the 1960s (Rhode 1991), it extends to describe the rapid growth of migration patterns of highly skilled human capital leaving the native

* Mattia Cattaneo [email protected]

1

Department of Management, Information and Production Engineering, University of Bergamo, via Pasubio 7b, 24044 Dalmine, BG, Italy

2

HERe (Higher Education Research), University of Bergamo, Dalmine, BG, Italy

High Educ

country in the most recent decades (Davenport 2004; Baruffaldi and Landoni 2012). Today, this phenomenon has been increasingly exacerbated, becoming a major concern even for some already developed countries (e.g., Southern European countries), which are not able to provide individuals with appropriate future opportunities. Rationally, educated individuals look for cross-border opportunities as a response to a lack of satisfactory salaries and permanent job positions in the home country. This effect is particularly true when considering the most educated component of society, i.e., students earning a PhD. Part-time positions and short-term contracts have indeed become more common in universities, causing increased uncertainty and temporariness during career progress (Cruz-Castro and Sanz-Menéndez 2010; Stephan 2012). Although previous studies have already pointed out that the migration of educated individuals represents a major concern for the country of emigration1 (e.g., Davenport 2004; Beine et al. 2011; Docquier and Rapoport 2012), the literature has almost neglected to investigate the extent to which educated migrants better embody the key to future innovation and economic development compared to people staying in their native country. By drawing upon economic development theories, understanding the quality of highly skilled people migrating to other countries is, however, essential to understanding the competitive advantage that countries might acquire in the long run (Agrawal et al. 2011). It is worth noting that unlike other fixed factors contributing to the development of a country, such as physical capital, human capital is a flow, a highly mobile factor that can and does relocate together with the individuals possessing it (Florida et al. 2008, p. 619).2 The key issue is therefore understanding to what extent countries are losing truly valuable human capital resources. In other words: Are highly skilled migrating individuals comparatively better than those who do not migrate? In this regard, Italy has been widely recognized as one of the most important countries affected by the brain drain phenomenon at high levels of education (Hellemans 2001). Since the early 2000s, crucial factors have contributed to highly skilled human capital’s decision to migrate, such as pathologically low levels of investments, a high level of bureaucracy, and difficulty in advancing in an academic career (Abbott 2001). In recent years, Italy has confirmed its underinvestment in research, positioning itself as an underdeveloped country in terms of gross domestic spending on R&D (1.33 Italy vs. 1.97 EU28 in 2015 excluding Italy according to the OECD) and the number of researchers per 1000 employed (4.93 Italy vs. 7.87 EU28 in 2015 according to the OECD). The opportunities for career growth today are not comparable to other developed European countries due to the temporariness and uncertainty of academic positions as well as bureaucratic difficulties. The potential for career advancement has further decreased as an effect of important turnover restrictions. After the passing of the 1

Estimating the cost of educating a student from primary school to PhD graduation, the Italian higher education system has supported almost 165,000 €. Data for the primary school to tertiary education period relate to the resources used each year to operate schools/universities (e.g., the salaries of teachers and other staff, maintenance of school buildings, students’ meals, or the rental of school buildings and other facilities, according to Education at a Glance reports). At the PhD level, the estimate considers the full cost of a PhD scholarship and the standard cost of a student in the field of social science to account for the resources used to operate the graduating higher education institution. It should be noted that in some HE systems, like the Italian one, PhD students are also paid to do research, thus implying a further expenditure for the system. The expected return is therefore not completely equivalent to that of the other levels of education where the return of the state could be more disinterested (we thank an anonymous referee for this comment). 2 Of course, although human capital is Bless fixed^ than fixed capital, its international mobility implies some costs. It is indeed not perfectly and immediately transferrable across countries, even at higher education levels (language, country-specific contexts, expertise and knowledge, lack of social connection in the domain of study/ research, etc.).

High Educ

Tremonti Law (Law 133/2008) concerning turnover blocks,3 tenured academic staff fell by 20%. The most recent report of the Italian Statistical Office highlights that the migration of doctorate holders has continuously increased in recent years (ISTAT 2014) where 13% of all PhD students graduating in 2010 had migrated abroad to look for better opportunities 4 years after their graduation, which was 6 percentage points higher compared to those graduating in 2009. In order to analyze the difference in the scientific standing of internationally moving doctorate holders, we consider the entire population of Italian doctorate holders in economics, finance, and business management that graduated in the years 2008–2010. Focusing on 1523 doctorates, we explore the career trajectories of those who have been scientifically active during their careers, 30% of whom moved abroad after their PhD graduation. This relatively high mobility is one reason to focus on the subject areas of economics, finance, and business management. Doctorate holders in these fields are indeed more mobile compared to those engaged in other disciplines, with an average migration per discipline of 13% (ISTAT 2014). This choice also allows us to consider a really brain-intensive area of science and thus neglect the disparity in the average physical capital (e.g., laboratories) of universities located in different countries. Results suggest that the doctorate holders who are more likely to move abroad are those with the best and worst research performances, whereas those remaining in Italy are found to have average research performances. Indeed, we do not find evidence of a significant difference in the research performances of those migrating abroad and those staying in their native country. This result contradicts the general popular belief that only Bbrains^ are going abroad looking for better opportunities. The effect is instead curvilinear and U-shaped. The probability of internationally moving decreases as scientific performance increases up to a certain level, but at a high level of research performance, the relationship reverts. Interestingly, deepening the analyses, the findings highlight that those with better research performances move to more prestigious universities to accelerate their academic careers (cream) and increase their reputations, but lower performing doctorate holders are more likely to move to another country to find a job in a non-research-oriented institution (milk) given the lack of opportunities in the native country. The rest of this article is organized as follows. BLiterature review^ reviews the literature on the brain drain phenomenon. BResearch design^ describes the study’s data, variables, and methodology. BEmpirical analysis^ presents the results, and BConclusion^ concludes.

Literature review Understanding the decisions of well-educated people to move to another country after completing their education is currently important for the development of knowledge societies. This study is conceptually guided by three main themes. The first deals with the choice to migrate at an individual level, the second considers the perspective of native countries losing highly skilled human capital, and the third focuses on the nature of the people migrating abroad. 3

The Tremonti Law (Law 133/2008) imposed a limit for the recruitment of permanent researchers for the period 2009–2011. In particular, the total expenditure for hiring new researchers could not exceed 20% (50% in 2012) of that of the year before and, further, the total number of hired researchers could not exceed 20% (50% in 2012) of those recruited the year before. Limits were then relaxed from 20 to 50% for this period (2009–2010) under the Gemini Law (Law 1/2009).

High Educ

Early research has viewed migration as an important investment in human capital, where migrating individuals aim to maximize their expected utility and, thus, the difference between the benefits and costs of moving. The decision to internationally migrate, however, implies higher transaction costs (e.g., finding a new location) compared to interregional migration in the geographical national distribution of people between rural and urban areas (Massey et al. 1993; De Haas 2010) and a more forward decision for well-educated people. By drawing on human capital theory, the reasons to migrate to another country lie on the desire to find higher wages, a greater quality of life towards locations offering higher utilities and more economic and political stability (Davenport 2004; Güngör and Tansel 2014), and personal characteristics of the migrant (Sjaastad 1962). Yet, reputation, career rewards, and intrinsic motivations are crucial factors for migration. Contrary to lesseducated migrants, the highly skilled counterpart is more driven by Bpull^ factors (career prospects and lifestyle at the hosting country) rather than the Bpush^ ones, such as the economic and political conditions at destination (Globermaaa and Shapiro, 2008). At the extreme, in some contexts such as Africa, safety for families and the possibility to move towards places without violence or political unstable conditions are the most important motivations for all levels of education (Clemens 2009). Researchers are also found to be more prone to establish abroad because of a lack of quality jobs at home along with the existence of limited opportunities to grow in their field, also in terms of research equipment and infrastructure support (Krishna and Khadria 1997). Beyond the material and economic rationales, a further key determinant lies in the professional issue, whereas agglomeration forces have been found to play an important role in the research sector. The Bbrains go where the brains are^ (Mashelkar 1984) in search of an appropriate climate conducive for research and collaborations is a dynamic that has permeated science and other sectors over time. Moreover, Kwok and Leland (1982) interestingly show that information asymmetry might also play a role in addition to the other determinants, as employers in the host country can determine graduates’ potential productivity more precisely than those in the native country. From the perspective of the source country, doctorate holders remaining abroad represent a critical loss of investment in talented people, or an indirect subsidy to the host country (Altbach 1991), that cannot spur innovation at home (Kim et al. 2011). This is particularly crucial for countries showing low degrees of internationalization at all levels of education, thus entirely bearing the cost of education. Historically, this is different from the Anglo-Saxon contexts where, already from the 1990s, the percentage of foreign-born college graduates obtaining their highest degree in a US university was significant (Glytsos 2010). Today, due to the small set of opportunities that even some developed countries are able to offer to well-educated individuals,4 international migration patterns are becoming highly alarming due to a clear Bbrain drain^ phenomenon. Although scholars have commonly examined this dynamic by considering the international migration occurring from developing to developed countries (e.g., Docquier et al. 2007; Beine et al. 2011), the declining economic conditions of developed home countries (e.g., southern European countries) at the time of PhD graduation has become a strong predictor of the migration choice (Finn 2007). Today, not only do policy makers have to face the migration (and retention) of foreign doctorate holders and graduates to the USA 4

In Italy, the share of unemployed individuals who attained or completed tertiary education as the highest level has increased in the last decade from 6.40 in 2000 to 11.90 in 2014.

High Educ

(Gonzales 1992; Van Bouwel and Veugelers 2012; Wang et al. 2013; Altbach 2015; Roh 2015), but the limited budgets for innovation and research of developed countries make them increasingly concerned about the outflow of Bthis cream of the national talent^ (Welch and Zhen 2008, p. 520) to previously competing countries. These effects are even more exacerbated in fields where investments in new advanced technology equipment are regularly required, as in the case of the health professions (Pang et al. 2002; Dodani and LaPorte 2005, Arah et al. 2008). Notwithstanding the importance of figuring out the directionality of international human capital migration, the most critical issue is understanding the value of people leaving their native country (Schuster 1994). Indeed, the effects of human capital on innovation and economic development depend on both its quantity and, especially, its quality. Previous research pointed out that the quality dimension of human capital has a real impact on countries’ development (e.g., Hanushek and Kim 1995; Gennaioli et al. 2013). At a macro level, the difference in the quality of human capital across countries is demonstrated to systematically vary with the level of development (Manuelli and Seshadri 2014). This result is even more critical when assuming that the quality of human capital crucially influences areas’ absorptive capacities, which results in a greater ability of the hosting country to learn advanced technologies and new knowledge (e.g., Carr et al. 2001). To the best of our knowledge, no studies have already investigated the migration of doctorate holders from higher education systems that are currently underinvesting in research (but that invested in the cost of their education) to the strongest ones by accounting for researchers’ scientific quality.

Research design Sample and data sources In order to understand the profile of doctorate holders moving to a foreign country after their PhD graduation, we consider the entire population of Italian doctorate holders in economics, finance, and business management graduating in the period 2008–2010 (PhD cycles XXIII–XXVI).5 Focusing on 1,523 doctorate holders (5% of the entire population), we investigate the career trajectories of those who have been scientifically active on at least one occasion as of December 31, 2016, i.e., those registering at least one scientific product (article, book, or book chapter) in the Scopus Sciverse bibliometric dataset. The data reveal that 35% of all doctorate holders published at least one scientific product (521 individuals), and that, among them, 30% (160) moved abroad after their PhD graduation. Figure 1 maps the percentage of new doctorate holders in these disciplines over the entire population of new doctorate holders in Italy in the period 2008–2010, reporting the average value at a provincial level (average value of the universities located in the same province). Interestingly, these PhD programs are widely offered in Italy. Indeed, 56 of 96 universities provide doctoral programs in economics, finance, and business management,

5

The data were provided by the Ministry of University and Research (MIUR) on February 19, 2016.

High Educ Avg. value of PhDs in economics per province NO doctorates

Fig. 1 Percentage of doctorate holders in economics, finance, and business management in each Italian university, aggregated at a province level

including also telematics universities (see BAppendix table A1^ for the list of new PhDs per university in the period 2008–2010). The distribution of the share of Italian doctorate holders in economics, finance, and business management is quite homogenous across the entire peninsula despite the existing socio-economic disparities among Italian macro-regions. We constructed scientific output measures by matching the list of 521 PhD holders to publication data on Scopus Sciverse. In order to minimize errors, we identified the ID of each specific doctorate holder if the following criteria were successfully met: (1) the last name; (2) the first name; (3) the ORCID digital identifier, if any; (4) the correct university and departmental (if any) affiliation name; and (5) the subject area of the majority of indexed products to that specific author ID. Subsequently, each scientific product was matched to an author if it was published during the doctoral program and no later than 1 year after the year of the PhD graduation.

High Educ

Methodology and variable description Methodology In order to analyze whether more or less promising (in scientific terms) doctorate holders have a higher/lower probability of moving abroad after their doctorate, we first perform a probit regression on the probability of going abroad after the PhD graduation, controlling for both individual, departmental, and university level characteristics that might influence their decisions. In the second stage, to better explore the international pattern of Italian doctorate holders, we perform a multinomial logit regression,6 estimated with a maximum likelihood procedure, to highlight the profiles of doctorate holders moving internationally toward (1) a ranked university; (2) a non-ranked university; or (3) another non-research-oriented institution (e.g., Ernst and Young, or McKinsey Corporation); or (4) staying in their home country (our reference case). Ranked universities have been defined as those ranked in the worldwide university ranking published by the Jiao Tong University in Shanghai (ARWU - Academic Ranking of World Universities), as in Horta et al. (2016). The outcome variable of the multinomial logit regression is the probability of falling into one of the four categories based on a non-linear function considering four outcomes (Maddala 2001).

Definition of the variables To account for the scientific potential and standing of Italian doctorate holders just after their PhD graduations, we first download all authored scientific products that each author published during the doctoral program and no later than 1 year after the year of PhD graduation (Research performance).7 To also account for the quality of their scientific production, each scientific product is weighted for the journal impact factor of the journal where it was published (e.g., Abramo et al. 2009; Gonçalves et al. 2009; Gaulé and Piacentini 2013; Rauhvargers 2014). In this regard, we use the Scimago journal rank indicator to account for the quality of each scientific product. Although the literature acknowledges the citations that a scholar receives across her career as an indicator of research quality and visibility (Horta et al. 2016), we decide not to consider the number of citations of papers published during the doctorate (and a year after) in the following years. Doing so allows us to avoid problems of endogeneity associated with the citation-gains resulting from migration trajectories (Franzoni et al. 2014). People moving to international research groups may indeed benefit from greater visibility, more contacts, and better reputations that can translate into more citations and publications (e.g., De Filippo et al. 2009); these aspects are not directly associated with the doctorate holders’ initial scientific potential and could potentially bias our analysis. To analyze whether more or less promising (in scientific terms) doctorate holders have a higher or lower probability of moving abroad after their doctorates, our dependent variable is a 6

In order to test for the validity of implementing a multinomial logit regression, we test for the independence of irrelevant alternatives (IIA) condition, which has to be met when using this model (e.g., Greene 2012). The results from the Hausman-McFadden test show that we cannot reject the null hypothesis H0, which states that the odds (alternative/outcome j vs. alternative/outcome k) are independent of other alternatives. The condition is therefore satisfied. 7 As a robustness test, we investigate the relationship between cross-border mobility and research productivity considering papers published within the end of the PhD and also after 2 years from PhD graduation. In both cases, findings corroborate our evidence (see BEmpirical analysis^).

High Educ

dummy variable equal to one in the case where the doctorate holder moved abroad after graduation. In order to find information on career trajectories after the PhD, we relied on different sources: LinkedIn and Google scholar profiles, online CVs, and the official websites of the hosting cross-border institutions/firms. Each profile is considered reliable if it clearly reports where the individual received her PhD and the name of the program. Because the probability of moving abroad is influenced by different dimensions, our analysis includes known controls for mobility, such as gender, age (at the beginning of the PhD program), and the specific subfield of science (economics, finance, or business management). Additional variables of interest are included to account for the context at the origin as follows: International mobility: We control for whether the student attending the PhD program in Italy is foreign and, thus, has a higher probability of returning to her home country after graduation for reasons other than the uncertainty associated with the academic job market in Italy. Intra-national mobility: Similarly, we control for whether an Italian student decided to leave her household and relocate to another area to attend her PhD studies. To this extent, a dummy variable is included in the model and is equal to one for intra-nationally mobile PhD students, namely those attending the PhD program at a university located in a different Italian region from that of their household. PhD program internationalization: Attending more internationalized PhD programs, measured as the percentage of international students attending doctoral programs at a university, would offer more opportunities to interact with both public and private institutions in other countries and would also be beneficial in terms of curriculum internationalization (Seeber et al. 2016). Doing so might indeed increase the probability that students can create linkages with other cross-border institutions to be exploited after graduation. International writing experience: Having mainly published in Italian journals rather than in the internationally recognized ones could not only signal a lower scientific potential of researchers, but also their lower propensity towards the international context. We account for this dynamic by identifying the percentage of articles published in Italian journals indexed in Scopus, such as Economica, Economia Politica, Rivista Geografica Italiana, and Stato e Mercato up to 1 year after the PhD program. A set of other variables is included to control for the context and the opportunities at the department/university of PhD graduation: Doctorate holders’ career competition: As the competition to become a researcher within the university of PhD graduation increases, so does the probability of a doctorate holder finding a job elsewhere (Cruz-Castro and Sanz-Menéndez 2010). Increasing mobility might indeed positively increase the chance of receiving tenure in the future (Lutter and Schröder 2016). As long as becoming a researcher represents the first step in the academic career of a doctorate holder, we control for the percentage of graduated PhDs in the same year, field, and university relative to the number of active researcher positions. Career advancement opportunities: In addition to the level of competition that PhD graduates face after their graduation, we consider a measure of career growth opportunity within the area of economics at the university of PhD graduation. We include the

High Educ

difference in the number of researchers between the year of PhD graduation and the year before, in the specific fields of economics, finance, and business management at the university of graduation, relative to the number of professors in the year before. As long as undertaking an academic career is increasingly competitive for postdocs (Bäker 2015), we expect that having scarce opportunities at the graduating university would increase the probability of going abroad to seek a higher salary or a faster path to obtain a tenured position compared to students at other Italian universities. Geographic area of graduation: We also consider a set of three dummy variables to account for the different Italian contexts and inequalities of opportunity (Checchi and Peragine 2010) and the associated practices typical of the area where the university is located, namely the Southern, Northern, and Central Italian macro-areas, where the Southern macro-area is the reference case. University size: The size of the university of PhD graduation, measured as the total number of registered students (bachelors and masters students), helps to account for the breadth of the network and international academic contacts that the institution has been able to develop over time. Teaching-orientation: We control for the ratio of the number of registered bachelors and masters students to the number of professors. It is indeed recognized that more teachingoriented universities, i.e., universities with more students per professor, are less researchoriented, increasing PhD students’ teaching commitments during the doctoral program as resources for professors instead of fostering students’ research and allowing them to build an international network. Private and doctoral university: We control for whether a PhD student graduates from one of the six doctoral universities in Italy (e.g., IMT Institute for Advanced Studies Lucca and Sant’Anna School of Advanced Studies) or a private university (Private University), as these institutions are known to be more equipped to provide students consolidated international research linkages and future cross-border opportunities during and after the program. Further, these institutions are less prone to immediately employ doctorate holders graduating from their PhD programs, discouraging inbreeding practices (Horta 2013). Lastly, we include the ratio of the research and development expenditure (% of GDP) of the destination country to that of the origin country (R&D expenditure D-O ratio) to account for the attractiveness of countries in the current knowledge society (Roh 2015).

Descriptive statistics Table 1 describes the characteristics of our sample of Italian doctorate holders, reporting both individual- and university-level features. Our sample is equally composed of male (50.3%) and female doctorate holders, and the average age is 27 years at the beginning of the PhD, with a maximum of 56 years. Considering their research performance during and within a year after the program, doctorate holders in economics, finance, and business management published a total of 0.81 articles, weighted for their quality, with a maximum of 56.2 for a star PhD student publishing a paper in Science (AAAS), among other outstanding scientific contributions. On average, doctorate holders published 1.6% papers in Italian journals (still indexed in Scopus). Analyzing the competition for a career at the doctoral level, new doctorate holders on average accounted for 50.3% of researchers in the same university, area of science, and year, with some

High Educ Table 1 Descriptive statistics Variables Independent variable Research performance Control variables Individual level Gender (male = 1) Age (at the beginning of the PhD) Intra-national mobility International mobility International writing experience University/departmental level PhD program internationalization Doctorates’ career competition Career advancement competition University size Student to faculty ratio Private university Doctoral university Country level R&D expenditure D-O ratio Fields of science Economics Finance Management Geographic area of graduation North Centre South

Obs.

Mean

S.D.

Min

Max

521

0.809

3.157

0

56.234

521 521 521 521 521

0.503 26.975 0.494 0.071 0.016

0.500 4.010 0.500 0.257 0.093

0 22 0 0 0

1 56 1 1 1

521 521 521 521 521 521 521

0.100 0.503 0.018 37,528.44 32.857 0.129 0.015

0.269 0.324 0.070 27,843.3 14.301 0.335 0.123

0 0.040 −0.054 273 0 0 0

1 1.500 0.800 142,796 109.901 1 1

521

1.171

0.384

0.160

2.838

521 521 521

0.818 0.088 0.094

0.386 0.284 0.292

0 0 0

1 1 1

521 521 521

0.484 0.330 0.186

0.500 0.471 0.390

0 0 0

1 1 1

cases where the absorption of new researchers would be critical, as at the Napoli - Università degli studi BL’ Orientale^ in 2010, where there were 150% more new doctorate holders than already hired researchers. Nevertheless, in terms of career advancement opportunities, the data reveal that on average the number of researchers increased by 1.8% between the year of PhD graduation and the previous year in the specific fields of economics, finance, and business management compared to the number of professors in the year before PhD graduation. The minimum value was that of the University of Varese - Insubria, where the number of researchers decreased by 5.4% between 2008 and 2009 compared to the number of professors in 2008 (mainly due to turnover blocks). Once doctorate holders decided to migrate abroad, they usually chose countries registering a higher R&D expenditure compared to Italy (more than 15%), as in the case of South Korea, which has highly invested in research and innovation in recent years (R&D expenditure D-O ratio: 2.836). In terms of pre-PhD mobility, 49.4% of doctorate holders moved from their households to attend PhD programs in a university located in a different Italian region, whereas 37 doctorate holders (7.1%) came from abroad to attend a PhD course in Italy. At an institutional level, the PhD programs offered at the graduating universities were not highly internationalized, with on average 10% of enrolled PhD students coming from foreign countries. On average, graduating universities enrolled approximately 37.5 thousand students per year. In terms of teaching resources, universities in our sample suffered from a relatively high student to faculty ratio (almost 33 students per professor), potentially decreasing the research focus of the academic staff.

High Educ

Almost 48% of doctorate holders graduated from a university located in a Northern Italian macro-region, whereas only 18.6% graduated from a university in the South, and 82% of graduates are from the pure field of economics. 12.9% of PhD students graduated from a private Italian university and 1.5% from a specific doctoral university.

Empirical analysis We first examine the equality of means in the research performances of those going abroad after the doctorate. The t test result indicates that there is no significant difference (P value = 0.585) between the scientific potential of those migrating abroad (0.923 is the sum of the impact factors of the articles published) and that of those staying in their native country (0.759). In other words, this result implies that research performance seems not to be a determinant of the probability of going abroad after receiving a PhD. To investigate the relationship between the scientific standing of PhD holders and the probability of moving abroad after their doctoral graduation in depth, we run two separate regression analyses. The first model regresses the probability of internationally moving after the PhD on the research performance of students during and a year after the program (Table 2). The most striking result is that the effect of the scientific standing of doctorate holders is curvilinear, with a U-shaped (convex) relationship, meaning that those who are more likely to move abroad are doctorate holders with the worst and best research performances, whereas those remaining in Italy have an average scientific quality (at the 1% significance level). In other words, the probability of moving decreases when research performance increases up to a certain level, but at a greater level of research performance, the relationship reverts. Among other factors, younger PhD holders have a higher probability of moving abroad and, on average, this result is also true for males. In detail, this corroborates the evidence that females are less mobile than males, who generally differ in the reasons behind mobility (Keith and McWilliams 1997). Overall, females are often tied to the mobility decision of their spouses (e.g., Nivalainen 2005), and further, in many countries (e.g., southern European countries), they may be more family-centered due to the strong societal roles that are associated with family responsibilities (Hakim 2006). In terms of career opportunities, the results suggest that the probability of moving to another country after graduation is positively associated with the presence of more colleagues finishing the PhD in the same year and in the same area of research relative to the number of researchers in the department (Career PhD competition). Interestingly, considering the number of newly opened researcher positions at the affiliated department, the results suggest that doctorate holders are not as concerned about the evolution of the department in the medium term (Career advancement competition), whereas their decision to move is affected by the turbulence and competition resulting from the presence of other colleagues (Career PhD competition). This is indeed exacerbated in systems exhibiting various degrees of localism in the academic promotion patterns as a result of inbreeding and provincialism dynamics, such as in Italy and France (Pezzoni et al. 2012). Especially in periods of funding restrictions, this of course intensifies the competition among locals as they all know that the arena is restricted and the competitors are their classmates. Further, we also found evidence that doctorate holders who move to Italy to attend the PhD program are highly likely to return to their native countries after receiving the title.

High Educ Table 2 The probability to move abroad after the PhD in Italy Variables

Probit model

Research performance

− 0.045*** (0.012) 0.002*** (0.000) 0.394*** (0.075) − 0.023*** (0.001) 0.042 (0.036) 1.307*** (0.140) − 0.340 (0.836) 0.303*** (0.050) 0.393*** (0.092) 0.455 (1.938) 0.001 (0.001) 0.003 (0.002) − 0.343 (0.298) 0.271*** (0.091) 3.233*** (0.541)

Research performance squared Gender (male = 1) Age (at the beginning of the PhD) Intra-national mobility International mobility International writing experience PhD program internationalization Career PhD competition Career advancement competition University size Student to faculty ratio Private university Doctoral university R&D expenditure D-O ratio Fields of science Finance Management Geographic area of graduation North Centre Constant Pseudo R-squared Log-pseudolikelihood Observations

− 0.095 (0.095) 0.201*** (0.024) 0.632*** (0.030) 0.507*** (0.028) − 4.799*** (0.352) 0.459 − 173.321 521

Investigating the pattern of the international migration decision, the results suggest that doctorate holders generally choose countries where R&D expenditure is higher than in Italy. Greater investment in creative work undertaken to foster knowledge in its different forms, i.e., humanity, culture, and society, might increase countries’ attractiveness in the eyes of individuals working on research at different levels (basic vs. applied) and also dealing with experimental development. Similar to foreign students coming in Italy to enroll in a PhD program (usually from eastern and less developed countries) primarily to

High Educ

increase their credentials when coming back home, Italian doctorate holders performing lower than the average may want to undertake an experience abroad. In the Italian context, this has been always evaluated as a positive driver to overcome any kind of job mismatch after returning at home (Ermini et al. 2017). At an institutional level, doctoral universities, which are acknowledged to dominate the international competition for students primarily focusing on the research mission (Marginson 2006), are associated with a higher probability for doctorate holders to go abroad, while this relationship does not occur when considering more teaching-oriented and larger academic institutions. In our sample, doctorate holders in management are more prone to move across borders than are PhD graduates in economics. This is consistent with the fact that, to a large extent, economics adopted a non-bibliometric approach while managing a bibliometric one for regulating the access to professorial positions in Italy, meaning that in management the access is more competitive on candidates’ research metrics (Seeber et al. 2017). Further, the probability of moving internationally is higher for universities located in the Northern and Central regions compared to those located in the Southern region. Finally, results suggest a negative relationship between the percentage of articles published in Italian journals up to 1 year after the PhD and the probability to move abroad after graduation, although it is found to be not significant. The second regression model aims to better describe the mobility pattern of PhD holders in relation to different types of destinations (Table 3). Considering the choice of staying in Italy (the native country) as the reference case, the results suggest that those with higher scientific profiles move to prestigious universities, whereas, on the other hand, those relatively lower scientific performers are significantly more likely to move internationally to a non-research-oriented institution. We do not find a significant difference (relative to those remaining at home) when considering doctorate holders moving to non-ranked universities. In summary, profiling PhD holders, those moving internationally generally belong to two categories: (1) the best performing PhD students, who aim to establish a faster academic career to achieve a reputable position in the scientific market (cream) and (2) the relatively lower performing PhD holders, who move across borders due to the lack of opportunities in their home country (milk), where scientific research performance has become a stringent requirement in recent years. On average, those hired at reputable universities are male and younger. Notably, the competition at the PhD level increases the mobility of doctorate holders to both ranked and non-ranked universities and also to other institutions rather than universities. Interestingly, in terms of career advancement opportunities, the coefficient is negative and significant when considering mobility to reputable universities, suggesting that as the number of positions increases (relative to the number of professors in the year before), the probability of going abroad decreases. On the contrary, the effect is positive in the case of international mobility to institutions other than universities. Doctorate holders at these institutions, who generally have lower scientific research performance, seem to be more concerned than others about recent openings of new researcher positions at the department of PhD graduation. These openings could make them conscious of the impossibility of entering academia due to the gap between the required scientific level (defined at a Ministerial level) and their research performance. Ultimately, the findings show that doctorate holders in finance are moving to non-ranked universities and other institutions less as compared to PhD graduates in economics.

High Educ Table 3 Estimated multinomial logistic regression results Variables

Ranked university

Non-ranked university

Non-research oriented institutions

Research performance

0.129*** (0.020) 0.408*** (0.040) − 0.053* (0.028) 0.018 (0.259) 1.975*** (0.461) − 0.504 (1.337) 0.933*** (0.233) − 2.814*** (0.530) − 0.001 (0.001) 0.005 (0.014) − 0.935 (0.722) 1.223*** (0.190) 8.342*** (2.935)

0.044 (0.074) 0.784* (0.425) − 0.051** (0.023) − 0.405** (0.205) 2.810*** (0.146) − 1.256 (1.291) 0.122*** (0.039) 1.880 (4.539) 0.007*** (0.001) 0.004 (0.026) 0.148 (0.383) 0.034 (0.533) 7.285** (3.258)

− 0.211*** (0.013) 0.723*** (0.180) 0.003 (0.048) 0.613*** (0.215) 2.225*** (0.686) − 8.074 (8.085) 0.419*** (0.159) 1.942*** (0.322) − 0.003*** (0.000) 0.009*** (0.002) − 0.446 (0.862) − 12.230*** (1.473) 8.163*** (2.431)

0.305 (0.368) 0.425*** (0.082)

− 0.339*** (0.002) 0.579*** (0.005)

− 0.858*** (0.188) − 0.583 (0.605)

1.107*** (0.206) 0.336** (0.131) − 11.839*** (4.306) 0.368 − 312.623 521

0.153 (0.561) 0.778*** (0.176) − 10.203*** (1.881)

1.752*** (0.148) 1.703*** (0.323) − 14.346*** (3.316)

521

521

Gender (male = 1) Age (at the beginning of the PhD) Intra-national mobility International mobility International writing experience Career PhD competition Career advancement competition University size Student to faculty ratio Private university Doctoral university R&D expenditure D-O ratio Fields of science Finance Management Geographic area of graduation North Centre Constant Pseudo R-squared Log-pseudolikelihood Observations

Conclusion The international migration of researchers has been a growing dynamic in recent decades and has been interpreted as a continuous Bdrain^ of talented (Bbrain^) people from developing to developed countries. Although previous research has largely accounted for an increase in internationally moving human capital (e.g., Davenport 2004), it has almost neglected to examine the associated quality of highly skilled migrants departing from already developed countries that underinvest in R&D and innovation. Our research has investigated the scientific profiles of doctorate holders in economics, finance, and

High Educ

business management graduating in the period 2008–2010 and moving abroad after graduation compared to those staying in Italy, which is their native country. We also deepen the analysis by distinguishing the destination of international migrants among (1) reputable universities; (2) non-ranked universities; and (3) other non-research-oriented institutions. These explorative findings suggest that research-active PhD graduates who have a higher probability of migrating fall into two classes: the best scientific performers during the PhD program and PhD holders with a relatively lower research performance. We find a curvilinear effect, suggesting that as scientific standing increases, the probability of moving internationally decreases up to a certain level, where the trend inverts. Deepening the analysis by considering the nature of the cross-border destination, we also provide evidence that those having a better scientific profile are more likely to target reputable universities, whereas lower performers usually move to another country to look for an occupation in a non-research-oriented institution. Our findings might be of help in interpreting the continuous outflow of well-educated people to other countries. Mass media and national newspapers8 have commonly reported that Italian Bbrains^ have left their home country to look for better opportunities abroad, preventing Italy from capitalizing on its educated human resources to seize global opportunities. Nevertheless, it is worth noting that within the current educated diaspora, Italy has retained good-quality doctorate holders on average, allowing for the migration of extreme cases, the potential star researchers, and the relatively lower scientific performers during and a year after the PhD. We therefore discourage indiscriminately demonizing the brain drain, and we point out that it is essential, especially for studies considering other countries, to understand who is migrating. Our results open the debate by suggesting some policy implications: (1) countries like Italy might benefit from implementing new policies aiming to increase the retention rate of top performing native doctorate holders and, at the same time, foster its attractiveness to foreign researchers9 (stimulating brain circulation and its benefits) by improving the system of post-doc grants and reducing bureaucratic difficulties.10 Financing these actions at a government level would be essential for ensuring their concrete implementation and success in time. From a non-academic perspective, a concrete measure to exploit the value embedded in doctorate holders is that of increasing the number of suitable jobs and also better broadening the job skills that PhD students acquire during their training in light of the needs of the job market (Di Paolo and Mañé 2016).11 (2) Increase the admission standards of PhD programs to pursue high-quality PhD students in order to avoid the investment of public money in lower quality PhDs that most likely will migrate to foreign Baverage-level^ research institutions. 8 La Repubblica—BQuei 3mila cervelli in fuga ogni anno da un’Italia che non saprebbe cosa farne,^ Salvo Intravaia, 26 febbraio 2016. 9 Santos et al. (2016) highlight the importance of implementing active public policies to both attract and retain PhDs in the Portuguese case. 10 ICREA’s grant proposal initiative is a valid example. 11 This is a major concern also for other Southern European countries, such as Spain, where recent studies detect that, once achieving a job, there are important earnings penalties for doctorate holders who are both overeducated and overskilled (Di Paolo and Mañé 2016). Interestingly, Santos et al. (2016) contradict the Btoo many PhDs^ argument when considering Portugal. They indeed highlight that, although not immediately easy to absorb, the need for attracting doctorates is beyond a mere immediate economic rationale. They indeed represent key resources for innovation and economic recovery.

High Educ

Our study, however, does not come without limitations, which can lead to promising avenues for future research. First, our findings might be extended and further corroborated by considering a more extensive analysis across all different disciplines, from engineering to health (it is recognized that there are significant differences in mobility profiles across disciplines) (Cañibano et al. 2011), and a more extended time span. Indeed, the fact our analyses specifically refer to the first years of the economic recession in Italy would require to examine in-depth whether the mobility patterns for different education levels has changed or not over time in order to better distill the determinants of highly skilled human capital migration. Second, although we do not have at our disposal human capital features (e.g., individual social and cultural experiences and household characteristics), and further information regarding how students addressed their PhD program (e.g., part- or full time), because we use data provided at the Ministerial level, we acknowledge the importance of these features for future contributions aiming to assess the likelihood of individuals to migrate in-depth (Sjaastad 1962; Roh 2015). At the same time, collecting further data to investigate whether the brain drain phenomenon could be also associated with brain circulation dynamics might contribute to a better understating of the international migration pattern of doctorate holders (Gribble 2008). A further caveat associated with data constraints is the impossibility of investigating the role of mentors in the relationship between scientific performance and the probability of migrating abroad after the PhD. Finally, although the hierarchical structure of Italian universities has started to slowly change as a result of the higher competitive funding mechanisms that have been implemented by the government, directly interviewing migrating doctorate holders would contribute to clarify whether the structure of the system has played a key role in absorbing the so-called average students as a factor of continuity and less of change/renewal. We leave this to future research. Acknowledgements We are grateful to the Ministry of University and Research (MIUR) and in particular to the General Director Daniele Livon for having supported us in this research project. The authors are also grateful to the General Director of the Conference of Italian University Rectors, Emanuela Stefani, and Susanna Terracini, a member of the National Agency for the Evaluation of Universities (ANVUR), for extremely helpful comments.

Appendix

A1 Fresh doctorate holders in economics, finance and business management per university University Milano - Università commerciale BLuigi Bocconi^ Roma - Università degli studi BLa Sapienza^ Bari - Università degli studi Bologna - Università degli studi Ancona - Università Politecnica delle Marche Roma - Università degli studi di BTor Vergata^ Palermo - Università degli studi Trieste - Università degli studi Roma - III Università degli studi Siena - Università degli studi Firenze - Università degli studi Torino - Università degli studi

2008

2009

2010

25 21 18 18 15 15 16 25 27 15 17 18

37 34 31 22 30 17 19 16 19 17 9 10

21 20 19 24 17 29 25 19 8 19 22 20

Tot 83 75 68 64 62 61 60 60 54 51 48 48

High Educ A1 (continued) University

2008

2009

2010

Tot

Pisa - Università degli studi Catania - Università degli studi Napoli - Università degli studi BFederico II^ Milano - Università Cattolica del BSacro Cuore^ Foggia - Università degli studi Pavia - Università degli studi Roma - Libera Università internazionale degli studi sociali Guido Carli (LUISS) Udine - Università degli studi Chieti e Pescara - Università degli studi Gabriele D’Annunzio Padova - Università degli studi Venezia - Università degli studi BCà Foscari^ Verona - Università degli studi Salerno - Università degli studi Lecce - Università del Salento Cagliari - Università degli studi Sassari - Università degli studi Messina - Università degli studi Parma - Università degli studi Genova - Università degli studi Ferrara - Università degli studi Campobasso - Università degli studi del Molise Bergamo - Università degli studi Milano-Bicocca - Università degli studi Macerata - Università degli studi Napoli - Università degli studi BParthenope^ Milano - Università degli studi Milano - Libera Università di Lingue e Comunicazione (IULM) Catanzaro - Università degli studi BMagna Grecia^ Reggio Calabria - Università degli studi Mediterranea Cassino - Università degli studi Perugia - Università degli studi Pisa - Scuola superiore studi universitari e perfezionamento BS. Anna^ Napoli - Seconda Università degli studi Reggio Calabria - Università degli studi Mediterranea Teramo - Università degli studi Modena e Reggio Emilia - Università degli studi Varese - Università dell’ Insubria Urbino - Università degli studi BCarlo Bo^ Potenza - Università degli studi della Basilicata Brescia - Università degli studi UNINT - Università degli studi Internazionali di Roma Napoli - Università degli studi BL’ Orientale^ Viterbo - Università della Tuscia Pavia - Istituto universitario di studi superiori Tot.

11 20 17 11 17 13 13 9 10 15 10 6 7 12 8 8 8 11 7 5 6 5 5 5 4 4 4 5 7 6 2 7 7 3 4 0 2 1 0 1 3 0 0 0 529

14 18 18 24 11 13 5 9 10 9 3 11 7 0 7 8 8 4 7 7 3 5 6 4 3 7 5 0 1 2 2 1 0 4 3 0 1 0 2 0 0 0 0 0 503

23 9 10 2 9 9 15 15 12 8 13 9 10 9 6 5 4 5 4 4 7 5 4 5 6 2 3 6 2 1 5 0 0 0 0 5 2 3 2 2 0 3 2 2 491

48 47 45 37 37 35 33 33 32 32 26 26 24 21 21 21 20 20 18 16 16 15 15 14 13 13 12 11 10 9 9 8 7 7 7 5 5 4 4 3 3 3 2 2 1523

References Abbott, A. (2001). Forza scienza! Nature, 412, 264–265. Abramo, G., D’Angelo, C. A., & Caprasecca, A. (2009). Allocative efficiency in public research funding: can bibliometrics help? Research Policy, 38, 206–215. Agrawal, A., Kapur, D., McHale, J., & Oettl, A. (2011). Brain drain or brain bank? The impact of skilled emigration on poor-country innovation. Journal of Urban Economics, 69, 43–55.

High Educ Altbach, P. (1991). Impact and adjustment: foreign students in comparative perspective. Higher Education, 21, 305–323. Altbach, P. (2015). Brain drain or brain exchange: developing country implications. International Higher Education, 2–4. Arah, O. A., Ogbu, U. C., & Okeke, C. E. (2008). Too poor to leave, too rich to stay: developmental and global health correlates of physician migration to the United States, Canada, Australia, and the United Kingdom. American Journal of Public Health, 98, 148–154. Bäker, A. (2015). Non-tenured post-doctoral researchers’ job mobility and research output: AN analysis of the role of research discipline, department size, and coauthors. Research Policy, 44, 634–650. Baruffaldi, S. H., & Landoni, P. (2012). Return mobility and scientific productivity of researchers working abroad: the role of home country linkages. Research Policy, 41, 1655–1665. Beine, M., Docquier, F., & Oden-Defoort, C. (2011). A panel data analysis of the brain gain. World Development, 39, 523–532. Cañibano, C., Otamendi, F. J., & Solís, F. (2011). International temporary mobility of researchers: a crossdiscipline study. Scientometrics, 89, 653–675. Carr, D. L., Markusen, J. R., & Maskus, K. E. (2001). Estimating the knowledge-capital model of the multinational enterprise. The American Economic Review, 91, 693–708. Carrington, W. J., & Detragiache, E. (1999). How extensive is the brain drain? Finance and Development, 36, 46. Checchi, D., & Peragine, V. (2010). Inequality of opportunity in Italy. The Journal of Economic Inequality, 8, 429–450. Clemens, M. A. (2009). Skill flow: a fundamental reconsideration of skilled-worker mobility and development. Center for Global Development Working Paper 180. Cruz-Castro, L., & Sanz-Menéndez, L. (2010). Mobility versus job stability: assessing tenure and productivity outcomes. Research Policy, 39, 27–38. Davenport, S. (2004). Panic and panacea: brain drain and science and technology human capital policy. Research Policy, 33, 617–630. De Filippo, D., Casado, E. S., & Gómez, I. (2009). Quantitative and qualitative approaches to the study of mobility and scientific performance: a case study of a Spanish university. Research Evaluation, 18, 191–200. De Haas, H. (2010). Migration and development: a theoretical perspective1. International Migration Review, 44, 227–264. Docquier, F., Lohest, O., & Marfouk, A. (2007). Brain drain in developing countries. The World Bank Economic Review, 21, 193–218. Docquier, F., & Rapoport, H. (2012). Globalization, brain drain, and development. Journal of Economic Literature, 50, 681–730. Dodani, S., & LaPorte, R. E. (2005). Brain drain from developing countries: how can brain drain be converted into wisdom gain? Journal of the Royal Society of Medicine, 98, 487–491. Ermini, B., Papi, L., & Scaturro, F. (2017). An analysis of the determinants of over-education among italian Ph. D graduates. Italian Economic Journal, 3, 167–207. Finn, M. (2007). Stay rates of foreign doctorate recipients from US universities, 2005. Oak Ridge, TN: Oak Ridge Institute for Science and Education. Florida, R., Mellander, C., & Stolarick, K. (2008). Inside the black box of regional development—human capital, the creative class and tolerance. Journal of Economic Geography, 8, 615–649. Franzoni, C., Scellato, G., & Stephan, P. (2014). The mover’s advantage: the superior performance of migrant scientists. Economics Letters, 122, 89–93. Gaulé, P., & Piacentini, M. (2013). Chinese graduate students and U.S. scientific productivity. Review of Economics and Statistics, 95, 698–701. Gennaioli, N., La Porta, R., Lopez-de-Silanes, F., & Shleifer, A. (2013). Human capital and regional development. The Quarterly Journal of Economics, 128, 105–164. Globermaaa, S., & Shapiro, D. (2008). The international mobility of highly educated workers among OECD countries. Transnational Corporations, 17, 1–36. Glytsos, N. P. (2010). Theoretical considerations and empirical evidence on brain drain grounding the review of Albania’s and Bulgaria’s experience1. International Migration, 48, 107–130. Gonçalves, R. R., Kieling, C., Bressan, R. A., Mari, J. J., & Rohde, L. A. (2009). The evaluation of scientific productivity in Brazil: an assessment of the mental health field. Scientometrics, 80, 529–537. Gonzales, A. (1992). Higher education, brain drain and overseas employment in the Philippines: towards a differentiated set of solutions. Higher Education, 23, 21–31. Greene, W. H. (2012). Econometric analysis (7th ed.). NJ: Upper Saddle River. Gribble, C. (2008). Policy options for managing international student migration: the sending country’s perspective. Journal of Higher Education Policy and Management, 30, 25–39.

High Educ Güngör, N. D., & Tansel, A. (2014). Brain drain from Turkey: return intentions of skilled migrants. International Migration, 52, 208–226. Hanushek, E. A., & Kim, D. (1995). Schooling, labor force quality, and economic growth (No. w5399). National bureau of economic research. Hellemans, A. (2001). Beating the European brain drain. Nature, 414, 4–5. Horta, H., Cattaneo, M., & Meoli, M. (2016). PhD funding as a determinant of PhD and career research performance. Studies in Higher Education, 43, 542–570. ISTAT (2014). L’inserimento professionale dei dottori di ricerca: informazioni sulla rilevazione. Keith, K., & McWilliams, A. (1997). Job mobility and gender-based wage growth differentials. Economic Inquiry, 35, 320–333. Kim, D., Bankart, C., A., S., & Isdell, L. (2011). International doctorates: trends analysis on their decision to stay in US. Higher Education, 62, 141–161. Krishna, V. V., & Khadria, B. (1997). Phasing scientific migration in the context of brain gain and brain drain in India. Science, Technology and Society, 2, 347–385. Kwok, V., & Leland, H. (1982). An economic model of the brain drain. The American Economic Review, 72, 91–100. Hakim, C. (2006). Women, careers, and work-life preferences. British Journal of Guidance and Counselling, 34, 279–294. Horta, H. (2013). Deepening our understanding of academic inbreeding effects on research information exchange and scientific output: new insights for academic based research. Higher Education, 65, 487–510. Lutter, M., & Schröder, M. (2016). Who becomes a tenured professor, and why? Panel data evidence from German sociology, 1980–2013. Research Policy, 45, 999–1013. Maddala, G. S. (2001): Introduction to econometrics, 3rd edition. Wiley Europe. Manuelli, R. E., & Seshadri, A. (2014). Human capital and the wealth of nations. The American Economic Review, 104, 2736–2762. Marginson, S. (2006). Dynamics of national and global competition in higher education. Higher Education, 52, 1–39. Mashelkar, R. A. (1984). Causes, consequences and remedies. New Delhi: Indian National Science Academy. Massey, D. S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., & Taylor, J. E. (1993). Theories of international migration: a review and appraisal. Population and Development Review, 19, 431–466. Nivalainen, S. (2005). Interregional migration and post-move employment in two-earner families: evidence from Finland. Regional Studies, 39, 891–907. Pang, T., Lansang, M. A., & Haines, A. (2002). Brain drain and health professionals: a global problem needs global solutions. British Medical Journal, 324, 499–500. Paolo, A. D., & Mañé, F. (2016). Misusing our talent? Overeducation, overskilling and skill underutilisation among Spanish PhD graduates. The Economic and Labour Relations Review, 27, 432–452. Pezzoni, M., Sterzi, V., & Lissoni, F. (2012). Career progress in centralized academic systems: social capital and institutions in France and Italy. Research Policy, 41, 704–719. Rauhvargers, A. (2014). Where are the global rankings leading us? An analysis of recent methodological changes and new developments. European Journal of Education, 49, 29–44. Rhode, B. (1991). East-west migration/brain drain. European Co Operation in the Field of Scientific and Technical Research COST, Kommission der Europ aischen Gemeinschaften, Brussel. Roh, J.-Y. (2015). What predicts whether foreign doctorate recipients from U.S. institutions stay in the United States: foreign doctorate recipients in science and engineering fields from 2000 to 2010. Higher Education, 70, 105–126. Santos, J. M., Horta, H., & Heitor, M. (2016). Too many PhDs? An invalid argument for countries developing their scientific and academic systems: the case of Portugal. Technological Forecasting and Social Change, 113, 352–362. Schuster, J., H (1994). Emigration, internationalization, and Bbrain drain^: propensities among British academics. Higher Education, 28, 437–452. Seeber, M., Cattaneo, M., Huisman, J., & Paleari, S. (2016). Why do higher education institutions internationalize? An investigation of the multilevel determinants of internationalization rationales. Higher Education, 72, 685–702. Seeber, M., Cattaneo, M., Meoli, M., & Malighetti, P. (2017). Self-citations as strategic response to the use of metrics for career decisions. Research Policy (In press). https://doi.org/10.1016/j.respol.2017.12.004. Sjaastad, L. A. (1962). The costs and returns of human migration. Journal of Political Economy, 70, 80–93. Stephan, P. (2012). How economics shapes science. Cambridge, MA: Harvard University Press. Van Bouwel, L., & Veugelers, R. (2012). An Belite brain drain^: are foreign top PhDs more likely to stay in the US? Working paper.

High Educ Wang, X., Mao, W., Wang, C., Peng, L., & Hou, H. (2013). Chinese elite brain drain to USA: an investigation of 100 United States national universities. Scientometrics, 97, 37–46. Welch, A. R., & Zhen, Z. (2008). Higher education and global talent flows: brain drain, overseas Chinese intellectuals, and diasporic knowledge networks. Higher Education Policy, 2, 519–537.