mobility, scientific productivity, academic research

Transcription

1 Project Number: 7PM Project Title: Deliverable Type: Deliverable Number: WP: 7 Version: 1 SISOB (An Observatorium for Science in Society based in Social Models) Report 7.2 Date: February 2013 Title of Deliverable: Editor: Authors: Dissemination level: Keywords: Abstract: Report on the effects of researchers mobility in terms of scientific performance Cornelia Lawson (FR) Aldo Geuna (FR), Chiara Franzoni (FR), Cornelia Lawson (FR), Soos Sandor (MTA KSZI), Giuseppe Scellato (FR), Paula Stephan (FR), Beatriz Barros (UMA), Eduardo Guzman (UMA), Daniel López (UMA) PU mobility, scientific productivity, academic research Capacities. Science in Society.Collaborative Project 1

2 Version history Version Date Description /11/2012 Outline /12/2012 First Version /02/2013 Final Version Acknowledgements The authors would like to thank: Ana Fernandez- Zubieta, fellow at the Institute for Advanced Social Studies (Spanish Council for Scientific Research) for her support with the review of the literature, the framework and the analysis in 5.2; Sotaro Shibayama, associate professor at the Research Center for Advanced Science and Technology (RCAST), The University of Tokyo, for his support with the analysis in

3 Executive Summary This deliverable documents the tangible achievements of Task 7.4 In- depth analyses of the effects of researchers mobility in terms of scientific productivity within Work Package 7 (a case study on Researchers Mobility and Social Impact) of the SISOB project. These activities were carried out both as an independent activity of the WP7 Working Group and in cooperation with the project partners. Firstly, the theoretical framework on why we expect a positive effect of mobility on performance is introduced (Section 2), focussing on expected increases in human and social capital as well as on job matching theory. We conclude that mobility should have a positive effect on the development of human and social capital but that this effect is not linear as different skills and network positions may be of importance for different roles. The job- matching approach further debates that performance can only be increased with a better job match. Section 3 then outlines expected results based on the theoretical framework. We hypothesise that only a move to an institution of higher quality/reputation will be associated with a medium term increase in productivity; after an initial period in which adjustment costs may constrain the productivity increase we should expect an increased research performance. Section 3 also presents a literature review of articles in leading social science journals on mobility and its effect on performance. We then discuss different indicators for scientific performance based on previous literature and in conjunction with Deliverable 4.2 (section 4). Further, we introduce a discussion on rankings to measure up- and downward mobility, again in collaboration with WP 4. Section 5 then presents five case studies of the effect of mobility on performance. It shows result of the 16- country Glob- Sci survey and country studies on the UK, US and Japan. The multi- country study based on mobility of academics in science and engineering disciplines and addressing geographic, sector and career mobility, gives a comprehensive picture of the relationship of different types of long- and short- term mobility on different performance indicators. Finally, section 6 summarises some technical details and 7 the main findings. 3

4 Table of contents Introduction Frameworks addressing mobility and its effect on performance Human and Social capital theories Human capital or tacit knowledge and skills Social capital or invisible colleges Search and matching theory Drivers of mobility Linking mobility to performance The effect of mobility on scientific performance Expected effects Previous findings Indicators and econometric challenges Scientific performance University rankings Background UK university Ranking Sample bias and endogeneity Data analysis countries: GlobSci Survey The Mover s Advantage: Scientific Performance of Mobile Academics Mobile Scientists and International Networks Mobility of UK academics Researchers mobility and its impact on scientific productivity Mobility of US academics Mobility Patterns of U.S. Faculty Mobility of Japanese academics Appointment, Promotion and Mobility of Bioscience Researchers in Japan Technical Details Conclusions References and Bibliography

5 Introduction The establishment of research networks and the mobility of researchers across different countries, fields and sectors has been identified as a major policy goal in recent years. In the EU, the commitment to develop a European Research Area (ERA) also implies the promotion of greater mobility of researchers (EC, 2001: 1; EC, 2010: 11, 17). National reports additionally point out the need for greater intra- national mobility and flexibility of researchers for knowledge diffusion between different institutions and sectors (e.g. CST, 2010). These policy papers assume that scientists mobility facilitates knowledge and technology transfer, the creation of networks and the increase in productivity. However, very few systematic studies have been carried out to measure the impact of mobility on individual productivity. Mobile researchers have been shown to create positive spillovers by enabling knowledge flows and the exchange of expertise. Studies on social capital of mobile inventors have shown that links to the original location are maintained and that knowledge flows are deeply embedded in labour mobility (Agrawal et al., 2006; Almeida and Kogut, 1999; Breschi and Lissoni, 2003). Similarly Azoulay et al. (2011) in a recent paper find that researchers moving to a new institution increase their citations from the destination university noticeably, while citation rates from the origin institution are not affected. The evidence shows that researchers can increase their visibility and credibility by moving to a different academic environment. Additionally, a new environment should have a positive effect on the productivity of a researcher. Literature on mobile inventors has indeed found a positive link between inventor mobility and productivity (Hoisl, 2007). There is very little evidence on the effect of mobility on research productivity but only some weak evidence suggesting a negative impact of immobility (Hagens and Farr, 1973) and the special role of post- doctoral research abroad (Zubieta, 2009). Instead there is some evidence suggesting that mobility is a characteristic of productive researchers and does not help enhance productivity (van Heeringen and Dukwel, 1987). Further, researchers tend to adjust their productivity to that of the department (Allison and Long, 1990). In a recent study, Kim et al. (2009) have challenged this view and shown that the effect of the department 5

6 on the researcher has diminished in the 1990s. However, the pull effect of elite institutions still leads to the agglomeration of top scientists in high ranked universities. Looking at the mobility and promotion patterns of a sample of 1,000 top economists, Coupé et al. (2005) suggest that sensitivity of promotion and mobility to production diminishes with experience, indicating the presence of a learning process. Cañibano et al. (2008) support the idea that the qualitative dimension of mobility impact is an important one to consider. Most internationally mobile researchers might very well be those embedded in larger networks, co- operating more with foreign researchers and having access to international funding sources. The positive spillovers of mobile researchers on their new institution have become more evident in recent years. Increased mobility through tactical hiring in the UK following the introduction of the Research Assessment Exercise has been pointed out as a mechanism for UK university departments to increase their credibility. Also in the case of short term research visits, receiving institutions hope to indirectly increase their reputation amongst the visiting scientist s home environment. Based on individual data we therefore aim to identify and describe migration patterns across universities, regions and sectors and investigate whether there is a positive effect on individual performance. 6

7 1. Frameworks addressing mobility and its effect on performance Deliverable 7.1 raised the question of the direction of causality between productivity and mobility (See Figure 1). It argued that it is necessary to address the relationship between mobility and career development to make statements about causality. This chapter will review and advance several economic theories that can be applied to the context of mobility and scientific performance: human capital; social capital; and search and matching theory. Figure 1 - Productivity and mobility relationship 1.1. Human and Social capital theories Human capital and social capital theories have examined job changes and the migration of highly skilled work force in the broad labour market (e.g. Grubel and Scott, 1966; Lin et al., 1981) and more recently in the case of individual scientists (Almeida and Kogut, 1999; Dietz and Bozeman, 2005). The theories suggest that human (what we know) and social capital (who we know) enhance productivity and recognition. They see mobility as an investment in human capital (Becker, 1962) while social capital facilitates mobility (Granovetter, 1973). Moreover, human and social capital theories are closely entwined. Network ties provide access to resources and allow for recombination thus increasing individual human capital (Bourdieu, 1986; Coleman 1988; Burt, 1992). Similarly, social capital is reinforced and maintained through the exchange of knowledge (Bourdieu, 1986). To fully understand the impact of mobility on productivity we first need to understand the role of mobility in human and social capital theories Human capital or tacit knowledge and skills Human capital theory describes the knowledge and skills that provide increases in cognitive ability, leading to higher productivity (Schultz, 1959; Becker, 1962). In other words, researchers with higher human capital will be the ones identifying new research 7

8 areas and making scientific contributions. It is constituted of human capital acquired through education and labour market experience (Becker, 1962), but also tacit knowledge acquired through learning by doing (Polanyi, 1962). In an academic context, researchers seek new skills and knowledge to increase their scientific human capital and thereby improving their potential for scientific breakthrough and innovation (Bozeman and Rogers, 2002). Mobility should benefit the development and utilisation of human capital as it leads to the acquisition of new knowledge and skills, especially that of tacit nature. The cognitive dimension of researchers mobility may then spread and increase human capital leading to better scientific performance. In a broader interpretation this implies that transfer and increase of human capital comes from changes in job positions and through collaborations. The cognitive interpretation of researcher mobility as the transfer of embedded tacit knowledge through mobility, assumes that researchers move or collaborate because they have assets that can be transferred. This is an interpretation followed by Zucker et al. (2002). They assume that star scientists, those with most tacit intellectual human capital, are responsible for knowledge transfer and innovation (Rosen 1981). They focus a researchers probability to move with no assumption about the possible increases in researchers human capital through mobility or their effects on productivity and career development. The second cognitive interpretation of researcher mobility takes the view that mobility increases researchers human capital. Such increases in human capital should have a positive effect on researchers productivity and career development. Dietz & Bozeman (2005) build their S&T Human capital theory around the cognitive intersectorial dimension of researchers mobility and its effect on career development. They assume that job experience in different sectors provides scientific and technical human capital that increases researchers productivity. However, they do not find enough evidence to support this diversity hypothesis. They found that early career publications are the main factor explaining academic productivity. The number of years spent in industry has a positive effect on patent productivity (see also Edler et al, 2011 on knowledge transfer to firms). This suggests that job transitions and/or collaborations do not guarantee an increase of 8

9 human capital, and if they do this increased human capital does not necessarily have a positive effect on researchers performance as it may consist of skills not recognised in academia. This makes it necessary to pay attention to the process of recognition of human capital acquired through mobility. Different research sectors have different incentives and recognition systems and this should be taken into account when assessing human capital effects for different movements and measures of research productivity. In this context one could extend that previous knowledge is critical for scientific performance as it assists the integration and accumulation of new knowledge. This follows the concept of absorptive capacity: even though knowledge is available a researcher may not be have the capacity to absorb and apply it (Cohen and Levinthal, 1990). Therefore, mobility may be more beneficial to researchers with already high levels of human capital. Mobility that does not allow the utilisation of prior accumulated knowledge, e.g. mobility to an unrelated job, may further decrease scientific productivity as new skills have to be learned. Thus careers are bounded by prior experience, promotion opportunities and organisational expectations (Gunz et al., 2000). The occupational and social context is important and human capital therefore cannot be viewed without the network in which it is routed (Granovetter, 1985; Coleman, 1988). Overall we can assume that human capital is positively correlated with scientific productivity. However, the relationship is not linear. Different types of human capital may be important for different roles or along the career path. Also, it has been shown that scientific performance declines over the life- cycle as older faculty get stuck on old ideas unable to create new innovations (Levin and Stephan, 2001). The same may be true for mobile researchers Social capital or invisible colleges Social capital describes the ability to benefit from networks, memberships or social structures to create knowledge, frame research questions and publish (Bourdier, 1986; Coleman, 1988). Several authors pointed out that human capital is closely embedded in social networks and social context (Granovetter, 1985; Bourdier, 1986). Looking at academic scientists several authors have described these networks as invisible colleges (de Solla Price, 1963; Crane, 1969) which may present a source of collaboration but also offer recognition. The social capital of science is thus crucial for 9

10 advancement and for gaining credit and assigning value to scientific work. Networks provide access to knowledge of other scientists research activities and thus help to integrate and shape science (Bozeman and Rogers, 2002). Social capital also provides a structure for human capital which is embedded in networks (Dasgupta and David, 1994). Jones (2009) describes how individual researchers come together in teams to create major innovations and that the combination of individual human capital is crucial to innovative processes. Thus, researchers with higher social capital will be the ones enjoying the support of peers and making scientific contributions. Social capital grows through the accumulation of individuals in organisations that facilitate the exchange of ideas (Burt, 1992; Kogut and Zander, 1992). However, social capital does not automatically increase when a researcher moves to a new organisational context. New links have to be built and old ones maintained for mobility to lead to an increase in social capital. Social capital theory has shown the importance of social links for enabling mobility; however, two opposing views exist on which network structures facilitate mobility. Granovetter (1973) coined the term weak ties to refer to influences outside the core cluster that will encourage mobility. Burt (1992) extends this to structural holes using similar arguments. Following up this argument, dense networks should discourage mobility as a person with strong interpersonal ties within the organisation is unlikely to terminate her contract (Dess and Shaw, 2001). On the other side, dense networks (strong ties) are seen as a source for mobility for some authors (Lin et al., 1981; Coleman, 1988), however, this may mainly apply to intra- organisational mobility. Glaeser et al. (2002) indeed counter their argument in an inter- regional context, predicting a negative relationship between expected mobility and investment in social capital. They even argue that mobility decreases the returns to social capital and that thus mobility should generally be bad for social capital accumulation. However, this may be less true in the context of mobility of scientists. Several papers have shown that researchers in industry and academe are able to maintain links to their old institutions while expanding their network, leading to greater recognition across institutional and geographical boundaries (Almeida and Kogut, 1999; Azoulay et al. 2012; Waldinger, 2010). Mobility could thus benefit the development and utilisation of social capital as it leads to a broader network creating extended networks, 10

11 recombination and possibilities for learning, collaboration and productivity. In the context of international mobility, personal and professional ties often remain between immigrants and their country of origin or prior residence and constitute an invisible backbone enabling reverse knowledge flows (Saxenian 2002). Previous contributions have highlighted the presence of a complex network of links between flows of researches and the generation and transformation of scientific networks (Melin, 2004; Jonkers and Tijssen, 2008). As argued above, mobility, especially intersectorial mobility, may not always result in an increase in scientific research. Links and networks may not be useful for advancing scientific research or cannot be carried into the new environment. This makes it necessary to pay attention to the process of recognition of social capital acquired through mobility, the reward associated with social capital and the degree to which contacts can be maintained. Different research sectors have different incentives and recognition systems and this should be taken into account when assessing social capital effects for different movements and measures of research productivity. Overall, however, we can assume that social capital is positively correlated with scientific productivity. However, the relationship is not linear. Different network positions and network sizes may be important for different roles or along the career path. Also, it has been shown that scientific performance declines over the life- cycle as older faculty get stuck in old networks unable to create new collaborations (Levin and Stephan, 2001). The same may be true for mobile researchers Search and matching theory Labour market analysis based on job matching and the search theory model (Jovanovic, 1979; Mortensen, 1986) has examined job changes in general and, more recently, in the case of scientists (Zucker, et. al, 2002) emphasizing, in the latter case, the role of productivity in explaining mobility. However, it has ignored the impact of mobility on productivity. The theory suggests that mobility asserts a positive impact on productivity only if researchers find better conditions to pursue their research, hence they move to a new job in order to increase their research performance. However, there are other reasons leading to mobility that are not related to research performance, for example, wage, family concerns, etc. To fully understand the impact of mobility on productivity 11

12 we first need to understand the drivers of researchers mobility in search and matching theory, and then model the impact of mobility on productivity controlling for those factors that could have a confounding effect Drivers of mobility Depending on different institutional set ups such as the public servant role of academics in some European countries the academic labour market is driven by traditional labour market factors, such as wage and search costs, contextualized to the academic market, and a set of academic specific factors that are related to research and reputation. Among the former the most important are: (1) wage related the difference between current compensation and new wage offer (particularly relevant for a move to a business job, usually associated with a much higher salary); (2) career related a promotion to associate or full professor usually associated to the access to more resources (possibility of hiring and directing doctoral and post- doctoral fellows) and with a higher salary; 1 (3) employment opportunity related non- permanent academic jobs are getting more common in all countries, these are associated with termination and non- renewal resulting in involuntary mobility; (4) market related the fluidity of the job market differs across countries and disciplinary fields and the thickness of the market varies depending on the time period; 2 (5) mobility cost related the relevance of costs associated with mobility is not fixed and depends on previous mobility experience; 3 (6) family related reasons partners moving, ageing parents and children s education related considerations can be common reasons for involuntary mobility and may reduce the propensity to move introducing a gender and age bias. Academic labour market can further be explained by academic specific factors, which are the focus of this paper. Setting aside redundancy, the wage received is the single most important determinant of the choice of accepting/leaving a business job. This is 1 In some countries, for example Germany, one is usually required to move to a different university to gain full professorship. 2 See the discussion of transfer markets for top scientists as a feature of the Research Assessment Exercise in the UK (Elton, 2000). 3 First time mobility is the most costly (leaving home effect), multiple job changes are associated with learning by doing, which decreases mobility costs (for example, foreigners or national with foreign PhD will have lower mobility costs). 12

13 not the case in the academic labour market, where other research and reputational factors are crucial. For academics, research is the most important aspect of their job satisfaction entering positively in their utility function while at the same time being a work activity that produces outputs. The time spent doing research is partially perceived by academics as leisure (consumption) time, resulting in their willingness to forego higher wages available in industry where independent research is pursued less. Hence, academics are willing to earn less, everything else being equal, to be able to remain in academic research (Stern, 2004; Sauermann and Roach, 2011). Another important factor in the utility function of a researcher is her reputation, which is affected by institutional reputation. A researcher values working in a highly prestigious institution because of direct benefits, such as fewer teaching obligations, more research time, higher financial endowments, etc., but also for positive externalities attached to these positions. These are important in the market for science in which individual quality assessment is not easy, especially in the early phase of the career, and prices publications - are not perfect carriers of information. An academic will move to a better- ranked institution (expecting benefits higher than the mobility costs), as research and reputation enters positively in her utility function. Second, especially in new and fast changing disciplines, mobility is driven by the prospect of accessing tacit knowledge and new equipment. In an early phase of development of a new discipline, knowledge is located in a small number of laboratories, where the original discoveries happened. Through publications this knowledge percolates through the university system but, especially due to the invention of new equipment (see for example the case of the production of the onco- mouse, Murray, 2011), some knowledge remains sticky to a laboratory and can only be passed on through training and equipment use. Researchers are willing to bear the costs of a move to such centres to acquire the tacit knowledge held there. This can happen through short stays (such as during a sabbatical leave) or with a job change. Third, academic mobility is strongly affected by the relative opportunity advantage. In a market with a clear reputation/quality ranking, researchers working in high rank institutions have a much lower probability of moving. 13

14 2.3.2 Linking mobility to performance The relationship between mobility and researcher s productivity is bidirectional. To model it we need to look at the probability of a job change as depending on the probability of receiving a job offer, and the probability of accepting that job offer. In the typical search theory model, the probability of receiving an offer depends decreasingly on search efforts. The academic profession being an intrinsically networked profession, the more connected a researcher is to a densely populated network of public and private organisations the lower will be her search costs as she will be well informed about available positions. The extent of one s social network (or social capital), hence, increases the researcher s probability of receiving an offer. The probability of receiving a job offer also depends on environmental labour market characteristics such as the existence of a strong potential demand. Potential demand in terms of flexibility and thickness of the academic market is scientific field, country and time dependent. The researcher s personal characteristics (or human capital), which could be interpreted as signalling high individual performance, positively affect the probability of receiving a job offer. In traditional job changes models, the probability of accepting an offer depends on the salary offered and the retention strategy of the company that can offer an increase in the salary and these factors are affected by personal characteristics, e.g. performance. In academia, the higher the academic s position and academic experience in that position, the higher will be the salary in her current university. However, academic salaries tend to vary within a defined national ladder based on experience with some limited flexibility at the top depending on the country. In the US, and much less so in the UK, professorial salaries can vary significantly, however, in most other countries public employee contracts or tradition give little space for salary increases. This leads, in the academic labour market, to a reduced effect of the salary on the probability of moving; in Europe wage offer relative to the individual s current compensation paly a very small role in explaining mobility. Among personal characteristics a key determinant of the probability of accepting a job offer is the academic position of the researcher. Non- tenured researchers are more likely to accept an offer than tenured university staff as they do have non- zero probability of having a non- renewal of contract (all non- tenured positions are based on soft money that is time limited). 14

15 The probability of accepting an offer depends negatively on mobility costs. Mobility costs include direct personal costs of moving to another city or country and skill adjustment costs that are particularly important for high skilled jobs. If the researcher s skills are university specific (i.e. not all the routines of the academic teaching and research work will be transferable to the work in the new university and even more so for a move to a firm), she must learn new practices, protocols, routines and adjust to different management and administration procedures. Even if these skill adjustments are minor, they can be considered as sunk costs and could deter some researchers from moving. 4 This is especially true for mature academic researchers, who have invested a lot of time in accumulating the skills and reputation needed to succeed in a specific university environment. Individual personal characteristics such as age and sex can also affect mobility costs due to family related consideration. Both direct and skill adjustment mobility costs are decreasing in the number of times a researcher has moved due to learning effects. At the same time the probability of accepting an offer also depends on the expected higher research performance that the researcher can achieve in the new job at a higher rank institution.. This can happen because of traditional job search related factors discussed above and/or because of an expected better research and reputation environment. Only if the job change is driven by research and reputation related motives we could expect a positive impact on the performance of the researcher. Hence, not all types of mobility are associated with increased research productivity. In the basic job search model the difference between expected and current utility needs to be higher than mobility costs for a job change to happen. Mobility costs are assumed to be instantaneous. However, mobility can be associated with significant deferred adjustment costs that can have a negative impact on the post- mobility productivity as the researcher will need to spend more time on learning tasks that could have been done more efficiently in her previous job due to knowledge of practices, protocols and routines (Shaw, 1987; Groysberg, 2008). Following the job change the researcher therefore witnesses a period of decreased productivity also associated with the setting up of the new laboratory in lab- based sciences. The length of the period and depth of the reduced productivity depends on the relevance of the adjustment costs, which in turn 4 A related interpretation of mobility costs can be found in Shaw (1987). 15

16 depend on the learning required to adjust to the new job. Job changes can therefore be associated with no change in the short to medium term scientific productivity if the reasons for moving are exclusively related to traditional job search factors and to a positive increase if mobility is driven by research and reputation reasons. In both cases we can expect a decrease in productivity due to adjustment costs in the short run. 16

17 2. The effect of mobility on scientific performance 2.1. Expected effects Two approaches were framed here to explain the positive impact of a move to a better research and reputation environment based on the matching model idea and on the increase in human (more diverse opportunities of learning) or social (better network connections) capital through mobility. The matching model predicts that researchers with high potential productivity unexploited in a lower quality department move to a higher quality department, where they can find better endowed laboratories (better equipment and more junior research staff) and hence increase their productivity. Second, a move to a better department means a move to a better research group with positive peer and network effects that increase the productivity of the researcher. Research group composition and local peer effects have been identified as important predictors of individual performance (Weinberg, 2007), and researchers are more productive if they co- locate with productive scientists. Working in a department with high quality peers does not only enhance productivity through direct interactions, but also through privileged access to their social network. Moreover, mobile researchers benefit from their existing network, which they carry into the new environment (Azoulay et al., 2012; Waldinger, 2010) creating new extended networks with the potential for new combinations. It is very difficult to disentangle the matching effect from the social and human capital model as in a highly reputable department the two are present (funding for good labs and high rank peers that allow the access to better social network and better learning). Moreover, highly reputable researchers tend to concentrate in highly ranked departments (Oyer, 2007). However, Kim et al. (2009) find that peer- effects have diminished since the 1990s perhaps due to better communication technology (see also Ding et al. (2010)). Within this framework, we hypothesise that only a move to an institution of higher quality/reputation will be associated with a medium term increase in productivity; after an initial period in which adjustment costs may constrain the productivity increase we should expect an increased research performance. If we think that scientific production is strongly affected by cumulativeness and self- reinforcement phenomena (Dasgupta and David, 1994), we could expect that an improvement in medium term productivity 17

18 will be persistent and thus will affect the long term performance of researchers. Conversely, mobility to an institution of the same or lower quality/reputation level should be associated with a short term productivity decrease due to adjustment costs that can only be slightly mitigated and at best stabilized to pre- mobility (for same rank change) or lower levels of productivity in the medium to long term due to a deficiency in financial and human resources assuming that the move is associated with a similar work profile (e.g. similar teaching and administration load). The impact of job changes on scientific productivity is mediated by the interaction between mobility and career development. It is possible that the relationship between mobility and researchers productivity not only changes across different types of mobility (to higher or lower quality institution) but also across the career cycle. Moreover, there may be a trade- off between career progression and quality/reputation of the institution. These relationships are important to qualify the impact of mobility on researchers performance. Mobility at different levels of the career (UK: lecturer, senior lecturer/reader, full professor) could result in differences in productivity growth. Mobility could have a higher impact on research performance in the early phase of a researcher s career when highest quality research is realized (Zuckerman & Merton, 1973). Mobility in early career stages would enable researchers with unexploited productivity potential to perform at their best in a higher ranked institution having a major impact on their future performance due to the cumulative characteristic of knowledge production. Oyer (2008) extends this by arguing that hiring departments have little information about the quality of young academics and a learning process will lead to corrections where misallocated researchers move up or down. Foreign- born and foreign- educated scientists have been shown to be disproportionately distributed among those that make exceptional contributions in science (Stephan and Levin, 2001). Hunter et al. (2009), for example, found that 50 percent of all highly- cited PhD physicists in the world work in a different country than that in which they were born. Stephan and Levin (2001) found exceptionally productive scientists and engineers working in the United States to have a higher probability of being foreign born and foreign educated than the underlying population of U.S. scientists. Though the papers do not directly address the question of causality, they hint at a possible strategic move of able young scholars to better institutions abroad to fully exploit their potential. 18

19 Career progression may also have an impact on researchers productivity as, especially in the case of promotion to full professor, career steps are associated with access to more resources and larger labs resulting in higher productivity (Long & McGinnis, 1981). A researcher may choose to move to an institution of the lower quality/reputation but in a higher career position and thereby counterbalancing the null or negative effect of downward mobility with the positive bust given through career progression. This is particularly true in the case of strategic hiring of highly qualified researcher from higher ranked institutions by lower ranked institutions that might offer attractive research and teaching packages. We further hypothesise, that a move to a higher ranked institution in a higher career position is unlikely to be achieved. Oyer (2007) confirmed for a sample of US tenured economists that the chance of an external researcher to be offered a position in a different department is lower than for a local, immobile researcher to received tenure. Similarly, Chan et al. (2002) find that very few researchers are able to move to a higher ranked institution and that these few exceptional scientists are two times more productive than the average academic at the destination university Previous findings We review the existing literature focusing on empirical studies that analyse the effect of researcher mobility and limit it to studies published in sociology, economics, and management literature between 1970 and We select two journals in each discipline by identifying top- ranked journals in the relevant field. Management journals include Academy of Management Journal and Management Science; economics journals include the American Economic Review and the Journal of Labour Economics and sociology journals include American Journal of Sociology and American Sociological Review. Additionally we consider articles published in Research Policy, a multidisciplinary journal of particular relevance to scholars in the field of science studies. We then searched the titles and abstracts of these journals for keywords related to mobility and academic researchers. The search returned just 90 articles indicating how little research has been done on the topic. We then reviewed the abstracts for relevance as we were only interested in empirical papers that investigate the effect of mobility on performance and excluded 75 articles. We then categorised and analysed the remaining 15 articles in detail (Table 1). The earliest articles on the topic were 19

20 published in the 1970s and 1980s in Sociology but were primarily based on data gathered in the 1950s and 1960s. In fact all the early papers are works by Long and Allison who both submitted a PhD thesis on the topic in 1976 and By the 1990s sociologists had largely lost interest in the topic and while there were few papers in the 2000s, Research Policy has started to embrace the topic starting in The American Journal of Sociology and the Academy of Management Journal have no article on mobility and performance in the 40 year period, while the American Economic Review only published papers that relate to the mobility of Economists. The review will focus on the four mobility dimensions defined in Deliverable 7.1, Section 2.2. Geographic mobility Cross- border, international mobility has been analysed extensively in the brain- drain- gain literature, but rarely touched on the effects for individual researchers. In our review two main measures for geographic mobility could be identified, firstly the foreign born and secondly returnees. Haeck and Verboven (2012) and Sabharwal (2011) compare local and foreign faculty in the Belgian and the US context. Haeck and Verboven mostly find no difference in promotion hazard between foreign and Belgian faculty, but they observe a negative effect for promotion to professor. Similarly, Sabharwal finds that foreign born faculty in the US are less likely to hold a chair and receive a lower salary, despite better performance in terms of publications, funding and patenting. Foreign born faculty therefore may report lower job satisfaction. Baruffaldi and Landoni (2012) examine publication outcomes of foreign born researchers in Italy and Portugal in more detail and find that only those with a strong Diaspora network perform better and that performance is lower for those that have spent a longer time abroad. These findings indicate that foreign born scientists may outperform their colleagues in terms of publications but that this effect may become weaker with time and largely depends on a strong network with other researchers from one s home country. However, there is discrimination in terms of salary and promotion to professorship. Reasons for this may lie in other academic duties like teaching and administration. Borjas (2000), for example, reports that US students taught by foreign born teachers receive lower marks. Secondly, geographic mobility has been measured as return mobility. Cruz and Sanz (2010) analyse the effect of Spanish researcher s participation in international postdocs 20

21 on being hired in a first permanent position and pre- hiring publication rates. They find a positive effect for both and argue that such stays increase social capital and thus promote performance. Haeck and Verboven (2012) look at promotion hazard of researchers at KU Leuven that received their PhD outside Belgium. They find a positive effect for promotion from assistant to associate professor but not for promotion to professor. They thus reason that the human capital acquired abroad may only play a short- term role. Edler et al. (2011) instead look at the knowledge transfer activities of German researchers that spent some time abroad. They find that researchers that regularly travel abroad or spent more than 12 months abroad are more likely to be involved in knowledge transfer in Germany and abroad than researchers that spent a shorter period abroad. Similarly, German researchers that are indefinitely stationed abroad have a higher involvement in KT but mainly to firms in their host country. Thus, returnees can only benefit from increased social and human capital for a short period of time unless they frequently return to their host country. Similarly, those indefinitely placed abroad are likely to lose contact to their home country. Thus, research contacts have to be actively maintained through continued mobility abroad to sustain higher levels of social and human capital. Sector mobility Dietz and Bozeman (2005) first investigated the effect of mobility from industry to academe on researcher s performance. They found that while industry experience of US science faculty has no effect on publication numbers, it positively impacts on the number of patents. Cruz and Sanz (2010) also find a positive career effect in terms of early tenure for Spanish researchers that had their first job in industry. However, both papers look at researchers that have spent just a few years outside academe and chose to return. Toole and Czarnitzki (2010) on the other hand focus on researchers that leave academe to join a firm. They show that these entrepreneurial researchers outperform their purely academic colleagues in terms of publications, funding and patents. However, they also show that publication and funding decrease following the switch to a firm. Patent numbers on the other hand are unaffected. Thus, intersector mobility can bring some benefits to researchers in terms of career prospects and commercialisation efforts but may reduce the number of scientific publications. 21

22 Career mobility The earliest papers investigating the effect of mobility looked at the importance of department prestige for explaining differences in publication outcomes following the move. Long (1978) and Allison and Long (1990) found a strong department effect for US scientists, showing that publication numbers and changes in publication and citation counts are due to mobility to a department of different rank. Specifically, mobility downwards decreased publication outcomes while upward mobility increased performance. At the same time Allison and Long (1987) also find that those researchers moving down to an unrated department are promoted quicker than immobile peers. Particular attention amongst immobile researchers has been paid to inbred faculty, those hired by their PhD institution. Several papers have looked at the effect of inbreeding vs. mobility on performance. Long et al (1993) using the same data as in the 1978 study find no effect of inbreeding on promotion and even a negative effect of mobility on promotion to associate professor. Oyer (2007) looks at inbreeding effects for US economists and finds that mobility is beneficial for researchers outside the top 10 institutions. While the top researchers will always work in the top departments, mobility in the middle allows for a better match and thus increases productivity. Cruz and Sanz (2010) also find that inbred faculty are less likely to be tenured early and that mobility is in fact beneficial for the award of a permanent position. Horta et al. (2010) also interpret their results as showing that inbred faculty are promoted less frequently. Using a sample of Mexican academics they show that these produce fewer publications, have a smaller network, do more teaching and local consultancy and produce more patents. Thus, they provide a different set of services and are less involved in research than their externally hired peers. Hilmer and Hilmer (2010, 2011) look at US economists and the timing of a move and find that multiple moves increase a researcher s salary as does late mobility (as associate professor). This is slightly contradicted by Haeck and Verboven s (2012) findings that show that external hires at KU Leuven are only promoted to professor sooner if they were hired early in their career (as assistant professors). They also show that PhD graduates from the KU Leuven are most likely to be promoted to professorship, giving some evidence for inbreeding at higher levels. Overall the career mobility findings show that mobility is heavily influenced by the 22

23 specific science system. Civil servant status in Europe or Mexico may promote immobility and faculty s focus on academic tasks outside research. Language barriers in some academic contexts may further promote career prospects of internal candidates. Overall, the timing of the move seems relevant and early mobility is of benefit in some contexts while late mobility could be career advancing in other contexts. Mobility further may only increase performance if it constitutes a better match. If the best match is already found, than any further mobility would no longer be of benefit. Task mobility None of the papers reviewed in this analysis looked at task mobility specifically. The papers on sector mobility could be interpreted as changes in tasks which may explain the decrease in scientific productivity amongst researchers that join firms in the sample of US scientists analysed by Toole and Czarnitzki (2010). Overall, the results show that there is still very little research on the effects of mobility on scientific performance. Many samples are limited to single institutions, countries or top- researchers. There is thus a requirement to analyse careers for larger samples and for longer time periods across different strata of the academic profession to find more robust evidence on the circumstances under which mobility can be beneficial. 23

24 Table 1: Mobility and scientific performance Paper Sample Observations Time- scale Data Method Performance Variable Mobility Variable Effect Geographic Mobility Haeck and KU Leuven, Belgium; all scientific personnel Hazard rate promotion foreign- born 0 / - (promotion rank3) Verboven, 2012, JLE fields; whole population; staff registers professors data, ISI publications models PhD abroad 0 / + (promotion rank2) Baruffaldi and Landoni, 2012, RP Edler et al., 2011, RP Sabharwal, 2011, RP Cruz and Sanz, 2010, RP Borjas, 2000, AER Sector Mobility Toole and Czarnitzki, 2010, MS Cruz and Sanz, 2010, RP Dietz and Bozeman, 2005, RP foreign researcher in Italy and Portugal; survey; response rate: Portugal 19%, Italy 18% scientists at German universities; survey; response rate 15.8%; subjects: science and engineering 2003 Survey of Doctorate Recipients (US); response rate 79%; limited to science and engineering faculty Survey of scientists with permanent positions in universities or CSIC in Spain; response rate 50%; limited to science and engineering faculty Survey of undergraduate students in economics at one US university NIH awardees ( ) in biology, chemistry and health science Survey of scientists with permanent positions in universities or CSIC in Spain; response rate 50%; limited to science and engineering faculty US Department of Energy, Defence and NSF funded researchers 497 professors (238 Portugal; 259 Italy) 2007 survey, ISI publications 958 academics 2006 survey Heckman probit 6375 faculty members (141,675 weighted) 2003 survey mean differences 1583 academics 2005 survey, ISI publications ZINB publications years abroad - home link + (Diaspora network) reasons for move 0 KTT share of time spent abroad + (10%, KTT in Germany) medium stay 0 long stay + OLS/NBREG Logit undetermined length + (KTT in host country) satisfaction foreign- born - publications + conference attendance + inventor + grant receipt + chair - salary - early publications international postdoc + early tenure international postdoc students 1995 survey OLS student marks foreign- born teacher - (for US born students) 89 SBIR researchers (entrepreneurs) and control group of Grant data, PubMed publications, NBER patents 1583 academics 2005 survey, ISI publications 1200 academics 2005 CV, USPTO patents 2- step Poisson publications entrepreneur/firm move + pub impact + Tobit grants + patents + pub/grant switch to firm - patents 0 Logit early tenure first job in industry + Tobit avg publications first job in industry 0 years in industry 0 Tobit / avg patents first job in industry 0 Poisson years in industry + 24

25 Table 1 continued Paper Sample Observations Time- scale Data Method Performance Variable Mobility Variable Effect Career Mobility Department Prestige Allison and US faculty members in science 179 job CV; ISI Logit publication difference upwards mobility + Long, 1990, (from Hagstrom, 1974) changes publications downwards mobility - ASR citation difference upwards mobility + downwards mobility - Allison and Long, 1987, ASR US faculty members in science (from Hagstrom, 1974) 274 job changes CV; ISI publications Logit promotion move to unrated department + Long, 1978, ASR US researchers with PhDs in Biochemistry from US universities; CVs of 80% of original sample Career Mobility Job Stability vs. Job Match Haeck and KU Leuven, Belgium; all scientific Verboven, fields; whole population; staff 2012, JLE registers Hilmer and Hilmer, 2011, AER Hilmer and Hilmer, 2010, AER Horta et al., 2010, MS Cruz and Sanz, 2010, RP Oyer, 2007, AER Long et al., 1993, ASR Economists at US universities; sample based on 2007 salary data information Economists at US universities; sample based on 2007 salary data information Survey of faculty members in Mexico; response rate 79%; limited to PhD granting institutions and PhD holders Survey of scientists with permanent positions in universities or CSIC in Spain; response rate 50%; limited to science and engineering faculty Economists at US universities; sample based on job market PhDs between 1979 and 1994 US researchers with PhDs in Biochemistry from US universities; CVs of 80% of original sample 134 non- movers; 47 movers 2691 professors 1009 academics 73 women, 680 men CV; ISI publications personnel data, ISI publications 2006 Salary data; EconLit publications, CVs 2006 Salary data; EconLit publications, CVs OLS publication difference department prestige new + department prestige old 0 Hazard rate models promotion External entrant 0 / + (promotion rank3) OLS salary multiple moves + (low rank institutions) overlap salary multiple moves highest number analysis top 5 publications multiple moves highest number OLS salary number moves + (for men) late move 414 academics 2004 survey OLS external openness inbreeding - NBREG publications - supervision 0 consultancy + patents academics 2005 survey, ISI publications 1263 economists 450 women, 556 men CV; EconLit publications CV; ISI publications NBREG Logit early publications non- inbred 0 early tenure non- inbred + mobility + OLS publications mobility (with top 10) 0 mobility (no top 10) + Hazard promotion to inbred 0 rate associate mobility - promotion to full inbred 0 prof mobility 0 + (for women) 25

26 3. Indicators and econometric challenges 3.1. Scientific performance The papers reviewed above have used a variety of measures for scientific performance. Table 2 gives an overview of previously used measures and measures used in the context of this study. Table 2: Scientific performance indicators Variable Previously used Use in current report Publications Baruffaldi and Landoni, 2012; Toole and 5.2 Czarnitzki, 2010; Horta et al., 2010; Cruz and Sanz, 2010; Oyer, 2007; Dietz and Bozeman, 2005; Allison and Long, 1990; Long, 1978 Quality weighted publications Hilmer and Hilmer, 2011; Toole and 5.1; 5.2 Czarnitzki, 2010; Allison and Long, 1990 Patents; KTT; consulting Sabharwal, 2011; Edler et al., 2011; Toole and Not used Czarnitzki, 2010; Horta et al., 2010; Dietz and Bozeman, 2005 Participation in international Sabharwal, 2011 Not used professional events (conferences, workshops etc) Networks (co- authorships, 5.1 citation flows) Awards, grants, scholarships, memberships Toole and Czarnitzki, 2010; Sabharwal, 2011 Not used Supervisor roles, undergraduate Horta et al., 2010; Borjas, 2000 Not used students, PhD students, (capita/year) Promotion Haeck and Verboven, 2012; Cruz and Sanz, ; Long et al. 1993; Allison and Long, 1987 Salary Hilmer and Hilmer, 2010, 2011; Sabharwal, 2011 Not used The general performance of researchers is conceptualized in terms of publication productivity. Deliverable 4.2 describes a productivity measure that counts the number of published items, and a quality measure based on citations. In the literature of scientometrics the two categories are frequently referred to as quantitative ("extensive") and qualitative ("intensive") indicators. As already pointed out in Deliverable 4.2, the significance of the distinction between the two types is not only theoretical, but gains practical role whenever indicators are used in evaluative rankings. Ranking by extensive indicators obviously leads to "bigness bias", such as the notorious leading positions of the US and the superpowers in national rankings. On the other hand, intensive indicators (particularly, the usual "specific" indicators) may highly rank items (e.g., countries) so small, that the quotient of the two small extensive indicators 26

27 are rather unreliable and arbitrary. Deliverable 4.2 pointed out three types of measures to tackle these discrepancies: extensity, intensity and mixtensity measures. In this report we will therefore be using several types of publication measures: the number of publications, the impact factor of the publishing journal and the number of total cites received by these papers in the first two years (2011 survey) or first five years after publication (panel data). As a second type of performance measure we look at the effect that mobility has on international network building. As pointed out in Deliverable 4.2, co- author networks are traditional social (collaboration) networks representing the organisation of a scholarly community. Mobility could enhance the size and quality of these networks and one of our case studies therefore looks at researcher s international linkages through co- authorship and surveys. Again, we are confronted with two dimensions: a quantitative ("extensive") and a qualitative ("intensive") dimension. In our case study we only look at the existence of international linkages and thus consider neither their extensity nor their intensity dimension. Examples (edited from D4.2): Publication and Citation distribution Existence Publications or citations received Not used Extensity number of items (sample size) Number of Publications (5.2) Intensity mean number of citations Mean number of Citations, mean impact factor (5.1/5.2) Mixtensity h- index (the greatest number that h papers received at least h citations) Not used Network and Partnership distribution Existence network yes/no Existence of international partners (5.1) Extensity number of partners Not used Intensity mean number of joint actions per partners Not used Mixtensity PHI (the greatest number that PHI partners acted as co- authors of at least PHI papers) Not used 27

28 As a third and last performance measure we concentrate on promotion in terms of career advancement (advancement in academic rank) and we hypothesised that mobility affects the speed of promotion. Career advancement will be measured directly through promotion from one rank to the next. Promotion requirements differ between rank levels and we thus also need to differentiate advancement from assistant to associate from advancement from associate to professor University rankings Background Deliverable 7.1 has pointed out the importance of vertical mobility as mobility between universities of different reputation and prestige. The literature review showed some important differences in performance development for researchers that moved down as opposed to those that move up the prestige ladder. We thus need to consider mobility along ranked systems or hierarchies of positions and locations, e.g. to a different level institution or position (upward, downward and lateral mobility). The main problem with regard to vertical mobility is the non- existence of comparable rankings across space and time. In the Mobility Case Study it is of interest to rank institutions according to capital availability (resources and peers). Quality weighted publications per institution and field could provide such a measure that is sufficient for the specific purpose of this case study but would not provide a ranking overall. Several case studies take into account vertical mobility. Case study 5.1 looks at the role of mobility from countries of higher or lower scientific activity and their effect on individual performance using country hindex measures. Case study 5.4 takes into account the funding awarded to specific scientific departments to measure quality and how it influences the effect of mobility. The case study in 5.2 will take into consideration the rank of institutions based on quality weighted publications. The process of developing this ranking is described below and was done in close collaboration with WP UK university Ranking Goal definition. Based on the data compiled by Thomson Evidence on UK Higher Education Institutions (HEIs data), a (system of) time- variant ranking(s) is to be constructed with the following minimal set of features: 28

29 A separate ranking is required (1) for each year and (2) for both disciplinary categories provided (natural sciences, engineering). The ranking system should provide means for registering significant career steps (through e.g. derived threshold values or scales for each ranking). Available indicators and measures. In the HEIs data two bibliometric indicators/measures are provided, accounting for the productivity and impact dimensions, respectively: Raw number of publications (for each HEI, for each year and for both categories), The relative impact within the discipline (per year) for each HEI given as the ratio of its mean citation rate to the world average: RI ( HEI year) C( HEI, year) / P( HEI, year) ( Total, year) / P( Total, year), : =. C Where C (HEI, year) is the number of citations received so far by the HEI for items published in year, and C(Total, year) is the total amount of accumulated citations within the discipline for items published in the same year. P(.,.) analogously stands for the number of publications in both cases. Preliminary evaluation of indicators. As of the usability of these two indicators for the ranking task, two issues should be addressed: (1) The readily provided data is aggregated and not decomposable. A major issue concerns the general aggregation level at which the HEI data has been provided. Instead of the original publication record, indicators are only available in pre- defined aggregations (sums and derived averages). As a consequence, there is no possibility of forming further measures based on the original data. In particular, since the base citation distributions are unknown, measures based on these, such as H- type indices or percentile- based excellence indicators, cannot be calculated. The averages are also not transparent: the Relative Impact indicator exposed above does not allow for the extraction of (1) the raw sum of citations, (2) the average citation count cit per paper as the absolute impact or (3) the 29

30 field - specific average as the reference ratio. In sum, the ranking should fundamentally be based on the two indicators included in the dataset. (2) Indicators are fieldwise overaggregated. An even more problematic issue is the aggregation scheme used for distinguishing between scientific fields. As it can be understood from the specification, sorting publications into natural sciences and engineering has been based on a mapping between OECD categories and ISI Subject Categories (SCs). Further issues are discussed in Deliverable 4.3. Proposal for an indicator set. The following choices can be considered as measures supporting the rankings. Share of the total HEI output/year. The contribution of the particular HEI to the yearly production of the UK sector (measured as the percentage of the total output). Cumulative share of the total HEI output/year. As an adjustment for possible annual fluctuations, bursts or sudden decreases, the share of the particular HEI from the total output, both aggregated up to the target year. Impact weighted productivity/year (IWP). Despite of the serious concerns outlined above, the impact indicator should still be considered for application. Assuming some robustness, and appealing to a potential trade- off between an unbiased and a multidimensional ranking, it is worth experimenting with as the sole quality- related indicator available in the dataset. To avoid multiple rankings implying a further method for producing a consensus, a combined measure integrating both the productivity/impact dimensions might be suggested. The annual Impact Weighted Productivity (IWP) of the HEI is a straightforward choice, and it can be assigned with a well- conceivable interpretation, as shown below: 30

31 ( HEI, year) P( HEI, year) C C( HEI, year) / P( HEI, year) ( Total, year) / P( Total, year) C( HEI, year) ( Total, year) / P( Total, year). ( HEI, year) IWP = def RI = P = C In verbal terms, the IWP yields the ratio of (1) the number of citations of the HEI, and (2) the world average of citations in the field for the same year. This can be conceptualized as the number of documents of the HEI (in year) equivalent to an average cited document in the field, each. Still another way, the measure can be conceived as the field- standard- equivalent output, which already takes into account the quality of papers as well. Cumulative version of IWP. As in the case of the simple output, it is also worth to formulate the cumulative version of IWP: Ranking schemes ( HEI, year) = RI( HEI, i) P( HEI i) CIWP def, i year Based on these measures obtained from the dataset, a set of ranking experiments are planned in order to (1) obtain a realistic result with reasonable scales and (2) to simulataneously test the rank shift indicator introduced in WP4.2. The following ranking schemes have been considered: (1) Percentile ranks (PR). Instead of an ordinal scale (given the skewed distribution of most indicators), HEIs are provided a percentile rank based on their underlying distribution, reflecting size differences between them. Usability in mobility models: Vertical mobility can be grasped with a real- valued PR- variable, whereby the difference between percentile ranks conveys the amount of mobility involved. (No thresholds needed.) It implies a weighted version of the monotonicity- indicators. (2) Minimal rank distance (MRD) ranking. The ranking is based on the rank shift indicator introduced in WP4.2. Each HEI is provided a (derived) rank 31

32 based i) on its position with respect to the predefined partitions of the ordinal ranking and ii) the minimal rank distance calculated for these partitions. Usability in mobility models: Vertical mobility can be grasped with a real- valued MRD- variable, whereby the distance (difference) between transformed ranks conveys the amount of mobility involved. (No thresholds needed.) It also implies a weighted version of the monotonicity indicators. Rule of thumb - like threshold values for specifying relevant mobility steps can also be introduced according the proposal in WP4.2. In this case, a discrete/ordinal level of the MRD- variable can also be used in the models. (3) Thresholds based on characteristic scores. The method of characteristic scores and scales is used for obtaining threshold values for relevant mobility episodes. The aim of the procedure, suited to skewed distributions, is to partition the set of items into comparable classes based on the underlying distribution. The procedure, in practice, yields five ranges (in this case: non- performers, poor performers, fairly good performers, remarkable and outstanding performers) Sample bias and endogeneity Econometric research has to deal with the recurrent problem of biases (Heckman, 1979) when estimating causal effects (e.g. omitted variables, reverse causality, measurement error, sample selection). A model that wants to measure mobility and its impact on productivity should include information on the characteristics of the researcher, the job and the match between the two to address selection into mobility (Osberg et al., 1986). Biographical information of the researcher, such as age, gender and household structure are equally as important as information on prior mobility events and research ability as well as the researcher s expectations about mobility outcomes. What regards job characteristics models should include details on teaching hours, capital availability and job satisfaction. Further some measures for the structure of the academic market are necessary. These data are difficult to obtain and thus selection bias difficult to resolve. Further, in a life course perspective we encounter specific problems related to personal and job characteristics that may change over time and effect selection into mobility as well as its effect on productivity. Firstly, personal characteristics change over time. As was mentioned in previous sections, younger 32

33 researchers are more likely to move as they are less likely to have found their optimal match. However, a mismatch, as is likely for early job changes, would not increase productivity but may even result in a decrease of productivity. Job mobility thus should have a stronger positive effect on productivity for older researchers. Additionally, the family situation of researchers changes over time and so does the motivation for job mobility that may be unrelated to job matching concerns during some periods. Similarly, job characteristics change over time. Teaching commitments and capital availability may change with each academic year and thus affect both, the researchers productivity as well as her willingness to move differently in each period. Also the structure of the academic labour market is subject to changes. In the UK the research assessment exercise has introduced an ever changing measure for the assessment of research quality that has impacted on requirements of the academic profession and on competition between universities for the best staff. Thus, when estimating the effect of mobility on productivity, standard regression estimates are biased as unobserved personal attributes are correlated with both the variable of interest and the independent variable. Mobile researchers might appear more productive than members of the comparison group. However, this effect might have been caused by the variable of interest (mobility) or by other variables (e.g. ability, greater motivation) in the mobile group for which data is not available and thus not controlled. If we assume that only the most promising and productive researchers are offered a position and can move, we have to solve two elements of this problem: 1) reverse causality: the probability of moving depends on past productivity (the variable of interest is not exogenous) and 2) omitted variable: ability to do research (and be productive) influences both, the probability to be offered a position and research productivity. Then, the challenge is to isolate the effect of the omitted variable and to solve reverse causality. In an ideal setting economists want to study the effect of mobility in an experiment that randomly assigns researchers into a mobile and immobile group, thus, exploiting an exogenous source of variation in the explanatory variable to analyse the effect of mobility on productivity. A natural experiment sets an exogenous and abrupt change in the group under scrutiny. For example, a university department that is unexpectedly closed forcing all researchers to move, assuming that the closure is not affected by their 33

34 productivity. Then, the possible causal link of productivity on mobility can be controlled by using this external shock as an instrument or by defining an unaffected control group. Moser et al. (2011) use the dismissal of Jewish scientists from Nazi Germany as a natural experiment to address the endogeneity problem in analysing mobility to the US. Similarly, Borjas and Doran (2012) use the collapse of the Soviet Union as an external shock for measuring mobility and productivity. Alternatively, the regression discontinuity (RD) design (Imbens and Lemieux, 2008) can be used when treatment is effectively random. It assumes that the value of a treatment predictor is on either side of a fixed threshold. The design thus can be applied to situations where an administrative body sets transparent rules for treatment and defines a cut- off point due to resource restrictions. RD design was originally implemented by Thistlewaite and Campbell (1960), who analysed the impact of scholarship award, which was based on observed test scores, on future outcomes, comparing the group that just passed with the group that just failed. In the case of mobility RD designs can be particularly useful for student mobility if places in higher education institutions are determined by a placement test or for participation in research visit programmes if these are determined by a stringent set of criteria. In the case of job- job mobility RD design is difficult to implement as assignment to treatment is not based on program participation or observable selection criteria. Thus, natural experiments and quasi- randomised assignment represent rare events that allow for advanced econometric exercises but are of little policy relevance when we want to investigate the importance of job- job mobility. Therefore if we are interested in analysing common patterns of mobility in science we need to develop econometric models that address intrinsic problems of heterogeneity, endogeneity and selectivity differently. As mobility cannot be randomised in practice the most common solution in econometric analysis has been to control for confounding factors, for example gender, age and past productivity. However, standard models fail to adjust for confounding if the treatment, mobility in our case, is time- variant (Robins, 1999). Thus, controlling for, for instance, past values of productivity, which are affecting later mobility, but may also be subject to earlier mobility, can lead to biased estimates because we control for the very pathway that is hypothesised to lead to higher productivity. A second solution is the use of fixed- effects estimators that remove individual unobservable time- invariant 34

35 differences between mobile and immobile researchers. However, they do not adjust for unobserved time- varying confounding and might thus still result in spurious correlation between mobility and publications (Robins, 1999). We therefore need to discuss other research designs that can deal with reverse causality and selection processes: instrumental variable approach (cf. Wooldridge, 2010) and treatment effects (cf. Wooldridge, 2006). Instrumental variables are variables that affect the regressor causing the bias but do not affect the variable under scrutiny. When analysing the effect of mobility on productivity, instrumental variables for mobility are variables that affect mobility but not productivity. However, finding plausible instruments is very difficult, especially in the case of mobility and productivity, where one researcher s instrument may be another researcher s hypothesised cause of publications. Addressing the causality between inventor mobility and productivity, Hoisl (2007) proposes a simultaneous relationship and considers city size as instrument for mobility in the productivity equation and external sources of knowledge as an instrument for productivity in the mobility equation. Toole and Czarnitzki (2010) use lagged regional variables as an instrument for joining or founding a firm in their productivity equations. While these instruments might be able to explain mobility opportunities to business firms without effecting productivity, they are not very convincing in the academic context. Researchers in larger or more dynamic cities may not only have more employment opportunities, but proximity to more peers may also affect their productivity. A positive effect of mobility on publications may then simply be a spurious relation caused by access to larger networks. However, instruments beyond regional indicators are difficult to identify and measure. We propose four instruments for voluntary mobility of permanent academic staff that we deem feasible albeit they are likely to involve extensive data compilation. Dahl and Sorenson (2010) showed for a sample of Danish scientists and engineers that also the highly skilled value proximity to family and friends and are willing to forgo parts of their income to live closer to home. We thus propose the distance to once place of birth and the distance to the city of undergraduate education as first set of instruments. Researchers that live further away from home are more likely to move as they have less social costs associated to the move. Distance from home should not affect productivity 35

36 though of course this cannot be ruled out as close family may provide help with child care which in turn could affect productivity. As a second and directly related instrument we propose homeownership. DiPasquale and Glaeser (1999) observe the switch between renters and homeowners and show that homeowners are less mobile. This may be due to different values and changes in interests that come with homeownership. It could effectively describe the difference between a person that values travel to one that values neighbourhood relationships and DIY. Glaeser et al. (2002) indeed show that homeowners participate in more local activities and are more likely to be members in church groups or hobby and sports club. Thus, researchers that own their home are hypothesised to be less mobile, while there should be no effect on productivity. Of course such an effect cannot be ruled out as homeownership may be correlated with family factors that also effect productivity. Further, homeowners may choose to locate outside the dense city centres and spatial distance to work may negatively affect social capital as less time is spent with colleagues in after work activities, thus possibly also reducing productivity. As an alternative instrument similarly measuring an academic s attitude we suggest using the foreign language proficiency of researchers. Language proficiency is assumed to present an interest in other cultures and thus to indicate a higher propensity to be mobile without effecting productivity. As a last instrument we propose the number of available job openings. Job openings directly affect the opportunities for receiving a job offer and in turns for accepting an offer. Detailed information on available positions in a given year, field and rank could thus serve as instrument for mobility and should not affect productivity unless they represent an overall increase in capital availability for universities. In the absence of natural experiments and instruments we can address the problem of reverse causality by modelling the likelihood of being treated. Thus, one can account for the selection into mobility based on pre- mobility observable characteristics. Amongst the important selection criteria that should be considered in matching of academics are age, experience, contract type, past productivity, prior mobility events, family related factors and reputation factors. To do so, one divides the sample into a treated group and an untreated control group. Treatment effect methods then assume that each individual 36

37 has potentially two outcomes dependent on the treatment. Thus, by assuming that treatment and control group are alike we can estimate the causal effect. Selection on observables can be done through propensity score (PSM) matching (Rosenbaum and Rubin, 1983). PSM allows modelling the propensity for each individual to be mobile and to create a control group based on propensity scores. If treatment, confounding variables and outcome variable vary over time, estimates based on PSM may be biased (Robins, 1999). Robins and colleagues (e.g. Robins, 1999; Robins et al., 2000; Hernan et al., 2001) propose inverse probability of treatment weights (IPTW) to deal with this problem. IPTW allows the estimation of average treatment effects even if confounding variables predict publications, mobility and are themselves predicted by past- mobility. Thus, they allow us to consider past publication and past mobility events for estimating selection into mobility. The problem of matching based on observables is that the likelihood of being mobile also depends on unobservable characteristics. Thus matching can only reduce endogeneity concerns but not eliminate them (Heckman and Navarro- Lozano, 2004). A life course perspective of researcher mobility calls for advanced statistical methods and complex data on individual, job and matching level to reduce selection bias and endogeneity. The main goal of economists in the field will be the collection of more and more reliable longitudinal data. 37

38 4. Data analysis This chapter shows the achievements of the Mobility Work Package regarding the analysis of the effect of mobility on scientific performance. Table 3 summarises the findings of the different sub- chapters. The summary will again focus on the four mobility dimensions defined in Deliverable 7.1. Geographic mobility Most literature on geographic mobility has focussed on cross- border mobility. In chapter 5.1 the authors concentrate on internationally mobile scientists, defining them either as non- mobile, foreign- born or returnees. Non- mobile researchers never leave their home country except for short visiting spells, foreign- born live and work outside their native country while returnees have returned to their native country after longer periods abroad. The analysis shows that internationally mobile researchers are authors on papers of higher quality. Especially the foreign born outperform native researchers as already seen in the paper by Sabharwal (2011) for the US. Amongst returnees, a positive effect can only be shown for researchers that went abroad for their post- doctoral research. Return mobility, however, never has a negative effect. Chapter 5.4 also considers international postdoc mobility of Japanese researchers and find it to have a positive effect on promotion. This effect, however, dwindles with time once we control for pre- mobility characteristics in a matched sample of researchers. Just like shown in previous papers, the post- doc effect may only be short- lived. Additionally, chapter 5.1 looks at the effect of geographic mobility on network building. It finds that foreign- born and returnees are more likely to have international co- authors and have a more diverse network. However, those researchers that go abroad for their PhD do not show higher levels of connectivity, but only those arriving later. Chapter 5.4 also looks at the effect of international visits on promotion. While chapter 5.1 does not find that researchers that report visiting periods as authors on higher impact publications, chapter 5.4 shows that Japanese researchers participating in such visits of 12 to 36 months to be promoted sooner than their peers. In a matched sample we find that they are promoted up to one year earlier and that this effect is lasting and stronger than for post- doctoral mobility. 38

39 Thus, international mobility is positive for several different performance indicators and mobility measures. Sector mobility Chapter 5.2 looks at sector mobility of academic scientists as mobility from industry to academe in the UK context. It shows that mobility is followed by a decline in productivity in the first few years following the move but that these researchers publish more in the long run. Thus, intersector mobility enhances personal performance but may also contribute to the science system as a whole by providing new impulses. Career mobility As pointed out previously, mobility cannot be separated from potential prestige gains associated to the move as previous papers have argued that performance is largely department prestige driven. All case studies consider some prestige measure in their analysis of mobility. Chapter 5.1 qualifies mobility in terms of moves between countries with different h- index. Thus, a move from a country with low scientific performance to a country with high scientific performance is considered an upward move. The analysis finds a positive effect of the h- index of the country of origin on the quality of the focal paper of a foreign born scientist. It also shows a positive effect for upward mobile researchers (from a low to a high performing country). At the same time, however, we also find a positive effect of downward mobility. Thus, mobile researchers are always outperforming their immobile peers. Chapter 5.4 also finds a positive effect for up- and downward mobility on the hazard of experiencing promotion in a sample of Japanese biologists. The effect of downward mobility is slightly higher, however, indicating that researchers move strategically to gain promotion. Chapter 5.2 on the other hand looks at post- mobility development of publication numbers and shows that downward mobile researchers indeed experience e a decrease in publication numbers, as argued by Allison and Long in their papers (1987, 1990). Task mobility None of our case studies has been able to look at task mobility specifically. Deliverable 7.3 will address this topic. 39

40 Table 3: Mobility and scientific performance: findings in chapter 5 Analysis Sample Observations Time- scale Data Method Performance Variable Mobility Variable Effect 5.1 GlobSci survey of Survey; ISI OLS Impact factor returnee + The Mover's authors of papers in academics in publications foreign- born + Advantage four scientific fields: 16 countries Visit abroad - biology, chemistry, nbreg citations 2 yrs returnee + earth and foreign- born + environmental Visit abroad 0 sciences, and OLS/nbreg Impact factor/citations hindex origin country (foreign- born) + materials science; hindex host country lower than origin + response rate 35.6% hindex host country higher than origin + postdoc abroad (returnees) + PhD abroad (returnees) 0 Job abroad (returnees) Mobile Scientists and International Networks 5.2 Researchers mobility and its impact on scientific productivity GlobSci survey of authors of papers in four scientific fields: biology, chemistry, earth and environmental sciences, and materials science; response rate 35.6% Survey of researchers that received at least one EPSRC grant, response rate 30% academics in 16 countries 170 academics 2009 Survey; ISI publications CV; ISI publications Probit international co- author returnee + foreign- born + PhD incoming - network >4 countries returnee + foreign- born + PhD incoming - OLS Impact factor returnee + (international papers) foreign- born + PhD incoming - nbreg publication mobility UK + upward mobility + downward mobility - citations 5 yrs mobility UK + upward mobility + downward mobility - nbreg publication mobility from industry + citations 5 yrs mobility from industry + 40

41 5.4 Appointment, Promotion and Mobility of Bioscience Researchers in Japan Survey of full professors in biology that received at least one Grant in Aid; response rate: 44% 400 academics CV; ISI publications; grant in aid stcox promotion assoc in t past mobility 0 mobility in t + inbred 0 visit abroad + postdoc abroad + promotion full prof in t past mobility 0 mobility in t + inbred 0 visit abroad 0 postdoc abroad + poisson years to promotion ass. visit abroad - (Diff- in- Diff visit US - PSM) visit other country 0 visit (top uni researchers) - postdoc abroad - years to promotion full visit abroad - visit US - visit other country 0 visit (top uni researchers) 0 postdoc abroad 0 41

42 countries: GlobSci Survey The Mover s Advantage: Scientific Performance of Mobile Academics Chiara Franzoni, Giuseppe Scellato, Paula Stephan 5 NBER Working Paper Series Working Paper Introduction The circulation of the highly skilled workforce is a global phenomenon, especially characteristic of exceptionally talented individuals for whom productivity differentials and respective differentials in wages- - are higher (Gibson & McKenzie, 2012). Moreover, immigrant workers, especially high- skilled ones, have a higher propensity for mobility than natives, both because they are less tied to the latest location but also because they respond more rapidly to new windows of opportunities in a different location (Kerr, 2009a; Kerr, 2009b). It is thus not surprising that national science and innovation systems compete not only to attract the best and brightest, but also to retain national talent and to attract back those who have left to study or work abroad (Hunter, Oswald, & Charlton, 2009). Empirical evidence on the correlation of mobility and performance in science, however, is inconclusive and often limited to the foreign- born in the US. Levin and Stephan (1999), for example, show that authors of exceptional contributions are disproportionately distributed among the foreign born and foreign educated in the United States, but they do not investigate more broadly representative samples nor do they investigate foreign- born differentials in countries other than the United States. Weinberg (2011) studies highly- cited scientists and finds that, although developing 5 The authors acknowledge support from Regione Piemonte for the GlobSci project and from the IPE Program, National Bureau of Economic Research. Stephan acknowledges support from the European Commission (FP7) Project "An Observatorium for Science in Society Based in Social Models - SISOB" Contract no. FP and Collegio Carlo Alberto Project "Researcher Mobility and Scientific Performance." The authors wish to thank Massimo G. Colombo and Paola Garrone for helpful comments. Alessandro Fornari, Antonio De Marco and Ali Mohammadi provided valuable research assistantship. 42

43 countries produce a considerable number of exceptional scientists, they are disproportionately located in high- income countries. Hunter, Oswald and Charlton (2009) study 138 highly- cited physicists. They show that migrants to the US perform similarly to native- US physicists; their performance is also no different from those who stay. Kahn and MacGarvie (2011) compare the performance of individuals who received Fulbright scholarships for study in the US to a control group and find no substantial difference in productivity in the top half of the Impact Factor distribution for those from richer countries. Finally, Borjas and Doran (2012) find evidence of displacement of US mathematicians after the wave of immigration of Russian mathematicians following the collapse of the Soviet Union and show that US mathematicians, particularly working in areas of research that overlapped with those of the Russian émigrés, experienced a decrease in productivity and a lower probability of producing exceptional contributions. In sum, evidence concerning productivity differentials between immigrants and non- immigrants usually focuses on the high- end of the performance distribution and findings often diverge. Studies are usually limited to comparing immigrants with non- immigrants. Only the Kahn and MacGarvie paper investigates productivity differences between those who return and those who do not. Moreover, the foreign born are usually compared to natives, without making the distinction between natives who have experienced mobility and returned vs. natives who have never experienced professional mobility. Here we make use of a new large dataset on the mobility of academics (Franzoni, Scellato, & Stephan, 2012). We investigate the correlation of mobility to performance within a broad spectrum of destination countries, of which 14 (Australia, Belgium, Canada, Denmark, France, Germany, Italy, Japan, Netherlands, Spain, Sweden, Switzerland, UK and the US) are developed countries with varying levels of scientific and research excellence, and two (India and Brazil) are fast- growing economies. The richness of the data allow us to control for a large number of characteristics such as age, gender, and job position, as well as self- assessed characteristics of the sample paper. Our sample spans the four scientific fields of biology, chemistry, earth and environment and materials and is for research- active scientists publishing a survey article in one of the four quartiles of the Impact Factor distribution of journals for each of the four fields. 43

44 Information about country of origin and mobility is obtained through a webmail survey fielded in the 16 countries and available in seven languages in We assess performance by looking at two different measures related to the survey article: the Impact Factor (IF) of the journal in which it was published and Total Cites (TC) received in the first two years after publication. Following the format and focus of the work of Hunter and colleagues (2009), our research attempts to answer questions such as the following (with answers provided in parentheses). Do foreign- born scientists perform at a higher level than scientists who have never experienced mobility? (Yes) Do returnees, i.e. nationals who have studied and/or worked abroad and subsequently returned home, perform at a higher level than home- grown nationals who have not been abroad? (Yes) Do returnees perform at a higher level than foreign born scientists working in the country? (The answer depends upon measure of performance). Do scientists who return after a period of mobility to their country of origin perform at a lower level than those who remain outside the country of origin? (No, although exceptions do exist). Do these patterns hold for the U.S? (Not always). Given the cross- sectional nature of data, we are not able to infer causality; that is, we cannot tell whether our results imply that a better scientific base attracts and retains the best and brightest internationally, or whether scientists perform better after they are exposed to a stronger scientific base, or both. But the results suggest that mobility is a plus for destination countries and that promoting international experiences can have positive returns for a country. Before presenting the results in section 5, we turn first to a discussion of the mobility of scientists in section 2, then present reasons as to why a relationship may exist between mobility and performance in section 3 and follow that with a discussion of the GlobSci survey and data in section The mobility of scientists Scientists are a highly mobile class of skilled workers. Many scientists move for the prospect of better working opportunities and larger compensation to their skills (Hunter, Oswald, & Charlton, 2009). But scientists are not only motivated by salary. 44

45 Scientists are known to respond to incentives in the form of recognition, intellectual curiosity and the freedom to perform research (Stephan & Levin, 1992; Roach & Sauermann, 2010). Selection in science is based in part on going to countries where one has the opportunity to be productive in terms of the resources and support for research. Many Argentinean scientists, for example, left the country after the financial crisis of 2002, when the devaluation of the currency made it impossible for them to even afford to attend international conferences (Dalton, 2008). Areas of expertise in science are also highly specialized, or require dedicated laboratories and special equipment that exists in very few settings (Stephan P., 2012). Science is also more often than not a team effort. Mobility of scientists often occurs in order to begin or continue working with a network of scientists or pre- existing collaborators (Harvey, 2008). Because of either the specialties or the talents of others, a scientist may perform at a higher level in one setting rather than in another and this can influence migration decision (Gangiuli, 2012). Among other reasons given for moving is the pursuit of an international lifestyle (Richardson & McKenna, 2003). Although family ties are known to be less important in driving the mobility of skilled as opposed to unskilled workers (Beine, Docquier, & Ozden, 2011) within the population of scientists family ties play a role in motivating return (Franzoni, Scellato, & Stephan, 2012). Migration waves have also been driven by political or social reasons. Consider, for example, the migration of German and Austrian scientists of Jewish origin in the first half of the last Century (Waldinger, 2010) or at the massive immigration of Soviet mathematicians to the US after 1992 (Borjas & Doran, 2012; Gangiuli, 2012). Demand also plays a large role in migration decisions. In many countries academic job markets are strongly regulated, recruiting and promotion systems change only rarely and by acts of law (Franzoni, Scellato, & Stephan, 2011). Some destination countries are also more or less willing to accept inward migration. For example, immigration policies are extremely restrictive in Japan for virtually any type of migrant. By way of contrast, several countries have policies for attracting high- skilled migrants that include easier visa procedures, fiscal benefits and special recruiting packages. Canada, for example, has an immigration system that assigns permits based on a point system. Skilled 45

46 workers have a special visa program (the Federal Skilled Workers Visa), under which a PhD entitles the applicant to 25 points out of a total of 100, making it extremely likely that a visa will be granted. A similar system exists in Australia, where a PhD degree currently entitles the recipient to 20 points. Other countries are less focused on attracting immigrants and more focused on convincing nationals who have migrated abroad to return to the motherland. India is discussing the creation of foreign- based contact points to encourage the recruitment of Indian scientists working abroad. Spain has an especially strong and large return policy for academics called Ramón y Cajal. Between 2001 to 2010 the program supported 2500 postdoc positions for Spanish researchers who had worked abroad for at least two years (Adjunar, 2012). In a country of the size of Spain these numbers amount to approximately 4% of the entire population of academics. 3. Performance of the mobile scientists In a context of large differentials in inward and outward migration, a question of considerable importance is whether mobile scientists are disproportionately distributed among the most talented. The standard theory of migration is based on Roy s selection model that posits workers make migration decisions based on the salary premium they would receive if they move, and on the cost of relocation (Roy, 1951). Training decisions, too, are sustained by the prospect of a skill- enhancement premium (Sjaastad, 1962), causing international students flows. With premium salaries that vary by level of skills and positive costs of relocation, provided skills are portable across countries (a condition likely satisfied in the academic workforce) sorting in out- migration will occur (Borjas, 1994; Grogger & Hanson, 2011). When the source and destination country have dissimilar levels of earnings for a certain level of skill (small correlation of earning/skill level), high- skill people will migrate from the low to the high remuneration country. For example, wealthier countries, which offer the highest salary premium, would attract a greater proportion of highly talented individuals from lower- paying countries (Gibson & McKenzie, 2012). When the source and destination country have similar levels of return to skills, which may occur in some cases of mobility between wealthy countries, positive skill- selection will still occur 46

47 toward those destination countries with a greater dispersion of return to skill (i.e. inequality ensures less redistribution, hence relatively higher earnings to the higher- skilled). Conversely, negative skill- selection will occur toward destination countries that have less dispersion of return to skills (i.e. less inequality or more redistribution, means relatively higher earnings to the lower- skilled) (Grogger & Hanson, 2011). Return migration is seen as further accentuating the initial selection. Thus, if the best were those to leave in the first place, returnees will be the worst of the best, or conversely would be the best of the worst, in case the out- migrating were adversely selected in the first place. Amid large salary differentials across countries, positive selection should result in observing a correlation of migration and performance, with the best- performing being disproportionately distributed among the internationally- mobile. It is less clear if a correlation should exist between performance and return and which direction it should go. There are other reasons in addition to selection as to why mobility may be associated with performance. For example, there are private and collective gains in the form of spillovers from brain circulation (Saxenian, 2005). Knowledge that is highly tacit or otherwise difficult to circulate travels fasters when scientists are geographically mobile. Mobility of people facilitates mobility of knowledge and more knowledge from distant sources is associated with greater idea generation and creative attainments (Hargadon & Sutton, 1997; Fleming, 2001). It is possible that the advantages from richer knowledge sets accrue primarily to migrants who sit in positions of arbitrage. Physical mobility is also helpful in establishing effective networks. In a prior paper, we show that immigrants have a higher propensity to establish international links, collaborate with coauthors in a larger number of countries and perform at a higher level than international teams comprised of non- mobile scientists (Scellato, Franzoni, & Stephan, 2012). But an academic environment that exposes scientists to richer knowledge sets may benefit the non- mobile host community. Likewise, it may benefit the native communities from which the mobile scientists have out- migrated. For example, long- distance collaborations between Indians abroad and their native communities in the 47

48 motherland have been found to promote knowledge transfer from the host country to the country of origin (Agrawal, Kapur, McHale, & Oettl, 2011). They have also been found to enhance formal collaboration networks (Gangiuli, 2012), functioning as a balancing force that tends to level- off the performance of movers and stayers. 4. Sample and data 4.1 The GlobSci survey We surveyed active researchers in the four scientific disciplines of biology, chemistry, earth and environmental sciences, and materials science during the period February- June In order to construct the sample, we selected all journals classified by ISI as belonging to one of the four disciplinary fields and sorted them by Impact Factor (IF) for all subfields in each of the four disciplines. 6 We then randomly picked a selection of four journals from each quartile of the Impact Factor distribution in each subfield of the four disciplines, thus obtaining four samples of journals by field stratified by Impact Factor. In aggregate, this process identified approximately 30% of all journals published in the four fields. Starting from these four lists of journals, we then downloaded the full record of all scientific articles published therein in From the affiliation information of the articles, we retrieved the address of the corresponding authors. 7 In case of multiple corresponding authors for a single article, we picked the first name in the list. We randomly selected one record in the case of corresponding authors appearing repeatedly in the corresponding author list. In order to build country panels, we coded these records, based on the final digits of the domain of the address (e.g..au for Australia;.be for Belgium, etc.). We identified US authors by those having.edu in the address, thereby restricting the US sample to academic researchers. 6 IF was taken from the latest available release of the Journal Citation Report of Thomson- Web of Science. 7 The four fields were chosen in part because 95 percent or more of all articles in these disciplines contain an address for the corresponding author. More specifically, in 2009 the estimated number of records that did not report address for corresponding author was 0.9% in biology, 3.6% in chemistry, 2.9% in earth and environmental sciences and 4.5% in materials science. 48

49 We prepared 16 country panels. Surveyed countries are: Australia, Belgium, Brazil, Canada, Denmark, France, Germany, India, Italy, Japan, Netherlands, Spain, Sweden, Switzerland, United Kingdom, United States. This procedure produced a sample of 47,304 unique addresses of scientists divided in 16 country panels (Table 1). Country panel sizes vary considerably, reflecting by construction the size of the country research- active population. The largest panel was the US, with observations; the smallest was Denmark with 513. China was initially included in the survey. However, a low response rate of less than 5 percent for a test sample of Chinese addresses suggested that respondents were either not receiving the invitation or had problems responding to the invitation. We thus decided not to survey researchers based in China. Panelists were invited to answer by . Invitations were sent, one country at a time, during the spring and early summer of 2011 and each panelist was invited a maximum of three times. The survey was initially developed in English and then translated into six other languages: French, German, Italian, Japanese, Portuguese and Spanish. The online questionnaire was developed through the platform Qualtrics that supports multiple languages. Each country survey and the respective invitation was administered in its primary language (two languages in the case of Canada). The platform automatically deployed the language in which the recipient had set her browser, and let the respondent switch from one language to another at any point while filling- out the questionnaire. Table 1 reports a summary of the answers by country. Country responses reflect both the size of the underlying research- active population of scientists as well as variations in response rates across countries. The largest number of responses is for the US (5165 answers) and the smallest is Denmark (227). The overall response rate is 40.6 percent; the high is 69.0 percent for Italy, the low is 30.3 percent for Germany; 11 countries have a response rate of between 35.0 percent and 45.0 percent. Answers are further divided into complete answers and partial (usable) answers (answers from respondents who began the survey, but dropped- out before reaching the last question). The total dropout rate was 5 percent. The response rate, conditional on the respondent completing the survey, is 35.6 percent. Reported response rates do not take into 49

50 account undelivered invitations due to such things as incorrect address, retirement or death and consequently underestimate the response rate. 8 Table 1 Response rate by country 9 PANELS TOTAL ANSWERS OF WHICH COMPLETE 50 OF WHICH DROPOUT TOTAL RESPONSE RATE COMPLETE RESPONSE RATE Australia 1, % 38.8% Belgium % 34.6% Brazil 1, % 45.0% Canada 2,455 1, % 36.5% Denmark % 40.5% France 3,839 1,618 1, % 35.6% Germany 4,380 1,326 1, % 26.2% India 1, % 35.1% Italy 2,779 1,917 1, % 63.3% Japan 5,250 1,860 1, % 32.0% Netherlands 1, % 33.3% Spain 2,303 1,228 1, % 46.9% Sweden % 34.1% Switzerland % 34.8% UK 3,695 1,355 1, % 32.0% U.S. 14,059 5,165 4, % 32.1% Total 47,304 19,183 16,827 2, % 35.6% 4.2 Sample For the purpose of this analysis, we selected the respondents who were working at universities, medical schools and government research agencies and residing in one of the 16 core countries at the time the survey was administered 10. From these, we further drop 56 observations because the country of residency at 18 years of age was not known, 32 observations because foreign international experience was not known or incomplete. We dropped an additional 163 observations of people who stated that they were in a country different from that of origin at the time the survey was administered but in a different question stated that they had no international experience. These answers could either be from people on temporary leave at the time the survey was administered or from people who gave inconsistent replies. The final sample is thus composed of observations, unevenly split among the 16 countries. Recall that during panel construction we randomly drew an equal proportion of source journals from each quartile of Impact Factor of the subfield category of the journals. Our 8 Walsh, Cohen, and Cho (2007) find in a sample of US scientists that undelivered s accounted for approximately 3.2 percent. Roach & Sauermann (2010) find that undelivered s accounted for 6.3 percent in a sample of junior US scientists. 9 Respondents were both academics and non- academics. In this paper we analyze the answers from academics. 10 Government research agencies were not sampled in the US.

51 panelists were therefore drawn from authors of articles in all levels of journal Impact Factor in equal proportions, short of any differences that might exist in the average size of the journals belonging to each group. In other words, the random pick of journals should mean that authors had equal probability of being included in our panel had they published an article in a journal with a very good, medium- high, medium- low or low Impact Factor. Similarly, if they published in more than one quartile, the probability that they would be chosen in one or another quartile would be random and in general not correlated to their prior international experience. Random pick means that panelists would be a fair representation of the population, short of a random noise. Concerning the degree to which the sample of respondents is representative of the panel and consequently of the underlying populations, we perform a number of tests. These are reported in Section A1 of the Appendix and show, at worst, very modest evidence of bias. Moreover, to the extent that bias exists, the final data set may slightly over- represent corresponding authors of papers appearing in higher Impact Factor journals. Table 2 Status of the academic workforce by country and by field NATIVES NOT MOBILE RETURNEES FOREIGN- BORN obs. % obs. % obs. % Italy (1664) Japan (1495) USA (4107) Brazil (657) Belgium (226) Netherlands (289) Denmark (181) Germany (1017) India (452) UK (1065) France (1242) Spain (1112) Sweden (271) Australia (550) Canada (810) Switzerland (283) BIOLOGY (4757) CHEMISTRY (5124) EARTH & ENVIRON. (3119) MATERIALS (2421) AVERAGE (15421) Weighted by probability. Sample size in parenthesis. 51

52 Table 3 Mean performance by STATUS of the corresponding author Observations: academic scientists in 16 countries *.10 **.05***

53 The international coverage of our survey has the advantage of providing information not only on the share of the foreign- born currently in a country, but also on the share of native scientists who currently live abroad in one of the other 15 core countries. Country of origin of the respondent was determined by asking in which country the person was living at the age of 18. We prefer this definition to country of birth, because we are interested in mobility decisions occurring for reasons of work or study of the respondents. Relocation events occurring before the age of 18 likely reflect parental decisions rather than choices of the respondent. Although we do not take country of birth as a reference, for simplicity, in the following description respondents residing in the same country as they did at age 18 are referred to as natives; those residing in a different country than where they lived at age 18 are referred to as foreign born. Of the respondents, (89%) indicated an origin in one of the 16 core countries; the origin of the remaining 1703 was a non- core country. Our survey is important because it is at present the largest and most successful effort to collect comparable data on the mobility of the academic workforce in a large number of countries. While micro- level data are available for migrants to the OECD countries, they are not specific to the academic workforce (Docquier & Rapoport, 2012). For the academic workforce, the sources of data until now have been very limited. Among these, three studies are worth noting. The first is a survey of Career of Doctorate Holders (known as CDH), conducted in seven countries by the OECD in collaboration with Eurostat and the UNESCO Institute for Statistics (UIS) (Auriol, 2007). The second (Ateş, Holländer, Koltcheva, Krstić, & Parada, 2010) is a pilot study of Eurostat in all domains of science and humanities, called The Eurodoc Survey I of doctoral candidates conducted in in twelve European countries (Austria, Belgium, Croatia, Finland, France, Germany, the Netherlands, Norway, Portugal, Slovenia, Spain and Sweden). The third is a pilot study of EU- US mobility designed to be used in combination with the Eurostat project (IdeaConsult, 2010). Unfortunately, low response rates in the latter two studies (considerably below 10%) make these data unreliable for research purposes. 53

54 5. Performance of the internationally mobile Based on the answers to the GlobSci survey, we partition the academic workforce of each of 16 countries into three statuses based on international background: NATIVE NOT- MOBILE: natives of a country who declared no prior experience of work or study abroad at the time the survey was administered RETURNEE: natives of a country who declared a prior experience of work or study abroad and were residing in country of origin at the time the survey was administered FOREIGN BORN: immigrant scientists whose country of origin differs from the country in which they were working or studying at the time the survey was administered. Status number one corresponds to non- mobile academics, trained in their country of origin, who have never taken a position outside their country. Categories two and three correspond to the internationally- mobile academics. These individuals share at least one experience of study or work away from their country of origin. The two groups differ, however, with respect to their current location. Individuals in group two have subsequently returned home; individuals in group three are either still mobile or have permanently resettled away from their country of origin. Groups one and two combined represent the share of domestic scientists within the academic workforce of a core country. Four types of international experience qualify for the status of internationally mobile: First, having an experience of non- PhD study outside the origin country (such as a BA, MA, laurea or equivalent); second, having either received a PhD degree or currently being enrolled in a PhD program (or doctorate education or equivalent) outside the country of origin; third, having taken a postdoc appointment or currently holding a postdoc position outside the country of origin; and fourth, having taken a job or currently holding a job outside the country of origin. For reasons of methodological conservativeness, stays of less than one year are not classified as an international experience Visiting periods of at least one year, such as PhD sandwich, which are quite common across EU member countries, and sabbaticals were coded and controlled for separately. 54

55 Here we compare the performance of the foreign born, returnees and native non- mobile researchers. Figure 1 shows the mean citation values for the survey article by field for the three groups. Figure 2 shows the mean IF. We see that regardless of field, mean citations of the foreign born are higher than those of native non- mobile researchers and also higher than returnees, although the differential is not usually as great and not statistically significant. In terms of IF, the sample paper for the foreign born outperforms that of native non- mobile and usually outperforms that of returnees. Figure 1 Mean total citations by STATUS and by subject +Weighted by probability. Figure 2 Mean Impact Factor by STATUS and by subject +Weighted by probability. 55

56 Although these first results are suggestive, they control for no other characteristics that are related to performance. Thus, we resort to multivariate analyses to control for potentially confounding factors. Our model for investigating the performance of the foreign- born and returnees compared to native non- mobile scientists takes the following general specification: Π i =φx i + δa i + γi i + ε i where: Π i is a performance indicator (either IF or TC) Xi is a vector of individual characteristics such as age (entered in a quadratic form), gender, trainee status, location (country were currently working); Ai is a vector of article characteristics that likely affect the rate of citation or visibility for reasons such as number of authors, international collaboration, interdisciplinary nature of the research or the article being in an emerging research topic 12 ; Ii is a dummy variable taking the value of one if the scientists has a prior international mobility experience (i.e. is either foreign- born or a returnee, depending on the model) or zero if the scientist is a native of the country with no prior international mobility. γ is a coefficient that captures the rate at which the international experience enhances performance relative to non- mobile scientists (in some models we will have split variables for international mobility and distinct γ s); φ,δ are vectors of coefficient estimates and ε i is the error term. We also initially controlled for prior visiting periods abroad, but the coefficient was not significant in any of the estimates and was omitted in the final estimations presented here. Summary statistics for all variables used in the paper are provided in Table The latter three controls were included as dummy variables based on self- reported characteristics of the article. Characteristics were rated on a 1-5 scale and the variables were set to 1 if the rating was grater or equal to 4: zero if lower than 4. 56

57 Table 4 Summary of variables and description Variable Obs. Mean St.Dev. Min Max Description impact factor (IF) Impact Factor of survey article total citations (TC) NATIVE_NOT_MOB RETURNEE FOREIGN_A~ visiting_only lowerh_country higherh_country BA_MA phd post job MA_quality phd_quality work_quality entered_work entered_edu Total Cites to survey article after 2- years from publication Natives of a country who declared no prior experience of work or study abroad at the time the survey was administered Natives of a country with experience of work or study abroad who were residing in country of origin at the time the survey was administered Immigrant scientists whose country of origin differs from the country in which they were working or studying at the time the survey was administered. Natives of a country who declared no prior experience of work or study abroad, but declared a visiting period of at least 12 months at the time the survey was administered Dummy=1 if the scientist was a returnee from a country with a higher H- Index than that of origin; 0 otherwise Dummy=1 if the scientist was a returnee from a country with a lower H- Index than that of origin; 0 otherwise Dummy=1 if scientist reported international experience of study at bachelor, master or equivalent level. Dummy=1 if scientist reported international experience of PhD level. Dummy=1 if scientist reported international experience of postdoc. Dummy=1 if scientist reported international experience of work. Interaction variable= BA_MA*H- Index of country where mobility occurred Interaction variable= phd *H- Index of country where mobility occurred Interaction variable=h- Index of country where mobility occurred if post=1 or job=1; 0 otherwise Dummy=1 if the scientist was a foreign born who entered for postdoc or job Dummy=1 if the scientist was a foreign born who entered for BA, MA, PhD hindex_origin/ H- Index of origin country/1000 years_since_returned back_forth AGE year of birth SQ_AGE Square of AGE year when returned to origin country after international mobility for study or work (first time returned was used if returned was reported more than once) Dummy=1 if scientists reported having returned, out- migrated and returned one or multiple times female Dummy=1 if gender is female; 0 if male still_training emerging interdisc inter_coop~d Dummy=1 if respondent was a student at the time the survey was administered Dummy=1 if respondent agreed or strongly agreed to statement this paper was in an emerging (new) area of research ; 0 otherwise Dummy=1 if respondent agreed or strongly agreed to statement this paper was in an interdisciplinary area of research ; 0 otherwise Dummy=1 if respondent indicated the survey paper included coauthors in other institutions and different countries of affiliation n_author Number of authors of survey paper COUNTRY_Australia COUNTRY_Belgium Dummy=1 if scientist reported country of current job location is Australia Dummy=1 if scientist reported country of current job location is Belgium

58 COUNTRY_Brazil COUNTRY_Canada COUNTRY_Denmark COUNTRY_France COUNTRY_Germany COUNTRY_India COUNTRY_Italy COUNTRY_Japan COUNTRY_Netherlands COUNTRY_Spain COUNTRY_Sweden COUNTRY_Switzerland COUNTRY_UK COUNTRY_USA FIELD_biology FIELD_chemistry FIELD_earth&env FIELD_materials Dummy=1 if scientist reported country of current job location is Brazil Dummy=1 if scientist reported country of current job location is Canada Dummy=1 if scientist reported country of current job location is Denmark Dummy=1 if scientist reported country of current job location is France Dummy=1 if scientist reported country of current job location is Germany Dummy=1 if scientist reported country of current job location is India Dummy=1 if scientist reported country of current job location is Italy Dummy=1 if scientist reported country of current job location is Japan Dummy=1 if scientist reported country of current job location is Netherlands Dummy=1 if scientist reported country of current job location is Spain Dummy=1 if scientist reported country of current job location is Sweden Dummy=1 if scientist reported country of current job location is Switzerland Dummy=1 if scientist reported country of current job location is UK Dummy=1 if scientist reported country of current job location is US Dummy=1 if focal article was in journal classified by ISI in Biology Dummy=1 if focal article was in journal classified by ISI in Chemistry Dummy=1 if focal article was in journal classified by ISI in Earth or Environmental sciences Dummy=1 if focal article was in journal classified by ISI in Materials science The results of the estimates of the general model are reported in Table 5. The left panel of the table shows results of the OLS estimate of the model in which the dependent variable of performance is the IF. The right panel shows results of the Negative Binomial estimate of the model in which the dependent variable of performance is the measure of TC. The performances of mobile individuals both returnees and foreign born- - are estimated against the baseline of the performance of domestic scientists who were never mobile. All models control for field and country. All models were also estimated using probability weights. Results are consistent with those shown in Table 5 and are reported in the Appendix. In both cases, we computed heteroschedasticity- robust standard errors. The robustness of the estimates is discussed in section 6. Regardless of which performance indicator is chosen, the estimates are largely consistent with the univariate analyses, pointing to positive performance differentials of scholars with either type of mobility experience. Moreover, in both instances, returnees have a performance advantage over those who are foreign born; the difference is 58

59 significant (at the 1 percent level) in the case of the Impact Factor measure of performance. This suggests that positive selection is not at work in determining who remains outside the country. This is supported by the finding, reported in the Appendix, that for most countries the performance of returnees is no different than that of compatriots who remain abroad after controlling for other effects. Exceptions, however, exist, as noted there. In particular, we find that returnees to India, Brazil, and Italy perform at a lower level compared to their fellow natives who have remained abroad. Table 5 Performance of the internationally mobile Robust OLS - v1- - v2- Robust Neg.Bin. - v1- - v2- RETURNEE *** *** RETURNEE *** *** (0.058) (0.061) (0.025) (0.027) FOREIGN_AT_ *** *** FOREIGN_AT_ *** *** (0.077) (0.079) (0.027) (0.028) visiting_only ** visiting_only (0.070) (0.035) AGE *** *** AGE (0.017) (0.017) (0.008) (0.008) SQ_AGE *** *** SQ_AGE female *** *** female *** *** (0.060) (0.060) (0.023) (0.023) emerging *** *** emerging *** *** (0.057) (0.057) (0.021) (0.021) interdisc interdisc (0.051) (0.051) (0.021) (0.021) inter_cooperated inter_cooperated ** ** (0.062) (0.063) (0.023) (0.023) still_training *** *** still_training *** *** (0.128) (0.128) (0.064) (0.064) n_author *** *** n_author *** *** (0.015) (0.014) (0.004) (0.004) field dummies yes yes field dummies yes yes current country dummies yes yes current country dummies yes yes Constant Constant *** *** (0.461) (0.461) (0.017) (0.304) Ln_alpha *** *** (0.017) (0.017) Dep. Variable IF IF Dep. Variable TC TC Observations 14,433 14,433 Observations 14,428 14,428 R- squared Adj. R- squared F stat *.10 **.05*** 0.01 (Heteroschedasticity- robust st. errors in parenthesis). Baseline: NATIVES_NOT _MOBILE 59

60 5.1 Performance of the foreign born We continue our investigation, by focusing on the performance of the foreign born. We exclude the potentially confounding effect of having returnees in the baseline and run the estimates on observations. Table 6 reports the results of two sets of estimates. The left panel has as the dependent variable the IF and the right panel has TCs as the dependent variable. As before, we treat IF estimations with standard OLS and TCs with negative binomial. On average the superior performance of the foreign born residing in a country compared to natives of the same country who did not experience mobility is confirmed. Note that the estimates are net of location effects, with the baseline being the country where the current job is located. They are also net of controlling for the H- index of the country of origin, included here to capture average differentials in the quality of basic and secondary education that the scientist received (Model 1). The results indicate that, holding all else equal, the average foreign- born scientist outperforms a homegrown scientist by about 0.84 in IF (left panel) and by as much as 2.29 TCs (right panel). 13 When we re- estimate the model using only the US observations (3772), the foreign- born performance premium is confirmed. 14 Some of the foreign- born entered the country during graduate or PhD studies. Others entered after having completed their tertiary education to take a postdoc or job position. Those who entered during higher education should in principle be more homogeneous to the non- mobile natives, having shared to a greater extent the same academic environment beginning with the training years. To account for this potential source of variation, we split the foreign born based on the career stage at time of entry to the current country. The positive and significant effect holds, regardless of the stage of entry (Model 2), suggesting that foreign born outperform natives even when they were educated in the same country of the non- mobile natives. The performance 13 The TC effect is calculated for changes in the conditional mean E(y i X i) of the dependent variable y i when the j- th independent variable x ij changes from 0 to 1, based on the formula. 14 Omitted for brevity, but available upon request from the authors. 60

61 premium varies by the stage of entry, ranging between a plus 0.66 in IF and plus 1.86 TCs for the average foreign born who entered during training to a plus 0.91 of IF score and 2.57 TCs for the average foreign born who entered after training. This effect holds invariant even after controlling for individuals who visited more than one country for postdoc training and job experience (results omitted for brevity). In sum, foreign- born scientists on average show superior performance than natives who were never mobile. Moreover, the performance premium exists even after controlling for country of origin and is stronger for people who entered after training than for those who entered during training, although positive in both cases. Table 6 Performance of the FOREIGN BORN Robust OLS - v1- - v2- Robust Neg.Bin. - v1- - v2- FOREIGN_AT_ *** FOREIGN_AT_ *** (0.096) (0.036) entered_work *** entered_work *** (0.105) (0.038) entered_edu *** entered_edu *** (0.124) (0.047) hindex_origin *** *** hindex_origin *** *** (0.150) (0.154) (0.052) (0.054) AGE *** *** AGE (0.019) (0.019) (0.010) (0.010) SQ_AGE *** *** SQ_AGE (0.000) (0.000) (0.000) (0.000) female *** *** female *** *** (0.072) (0.072) (0.026) (0.026) emerging *** *** emerging *** *** (0.067) (0.067) (0.025) (0.025) interdisc interdisc (0.061) (0.061) (0.024) (0.024) inter_cooperated inter_cooperated ** * (0.073) (0.073) (0.028) (0.028) still_training still_training * * (0.135) (0.136) (0.066) (0.066) n_author *** *** n_author *** *** (0.017) (0.017) (0.004) (0.004) field dummies yes yes field dummies yes yes current country dummies yes yes current country dummies yes yes Constant ** ** Constant *** *** (0.540) (0.540) (0.261) (0.261) Ln_alpha *** *** (0.020) (0.020) Dep. Variable IF IF Dep. Variable TC TC Observations 10,046 10,046 Observations 10,043 10,043 R- squared Adj. R- sq F stat *.10 **.05*** 0.01 (Heteroschedasticity- robust st. errors in parenthesis). Baseline: NATIVE_NOT_MOBILE 61

62 5.2 Performance of returnees We move on to consider how the performance of returnees compares to that of fellow natives who have not been mobile. As stated above, the standard selection model postulates that returnees would be the worst of the best who migrated if initial outward migration was composed of positively- selected scholars (Borjas & Bratsberg, 1996). 15 We also noted that mobility might be associated with greater skills that the scientist acquired while training or working in a richer environment (Sjaastad, 1962) as well as with a broader and better network established while abroad (Scellato, Franzoni, & Stephan, 2012). We exclude from the model the foreign born and include in the model only the natives of the sixteen countries who are currently based in their country of origin. Table 7 reports the results of step- wise estimates for IF (left panel) and TCs (right panel). The estimates of Model (1) confirm that, on average, returnees outperform natives with no prior experience of international mobility. The difference is sizable: holding all other things equal, returnees have on average a 0.63 higher IF than natives with no prior mobility and 1.69 higher TCs. We noticed earlier in the univariate tests (Table 3) that the US exhibited a different pattern than the other countries, with no apparent difference in the average performance between returnees and non- mobile natives. When we run Model (1) for only the 2597 US observations, we find that this result holds for the IF measure of performance. However, we find that on average US returnees perform at a significantly higher level of TC than natives. 16 In Model (2) we split the returnees into two groups: those who had a prior international experience in a country with a stronger science base than their country of origin and 15 In the Appendix we show that we find no compelling evidence of negative selection regarding return, when comparing natives with an experience of international mobility who have subsequently returned to natives who have remained abroad. In the simple mean- comparisons we see that returnees to India, Brazil and Italy perform at a lower level compared to their fellow natives who have remained abroad. However, this correlation does not hold for other countries. We find no significant difference between returnees and those who remained abroad after we control for the effect of current country and other potentially confounding factors when countries are pooled (see Appendix A- 2). 16 The estimates are omitted in the interest of brevity but are available upon request from the authors. 62

63 those who had a prior international experience in a country with a weaker science base than their country of origin. To measure movement we use the country ranking based on the country H- Index 17. In case of multiple moves, we choose the highest ranked among the destination countries. As expected, the great majority (72.4%) of returnees was mobile towards countries with a better H- Index. However, both upward and downward mobility are associated with a positive performance differential. Based on the estimates of Model (2), returnees from a country with a stronger science base produce papers with a 0.81 larger IF. Mobility downwards is associated with a 0.26 larger IF. The two coefficients are statistically different from one another (Prob. F=0.000). Differentials in TCs (1.8 more citations for upward mobility and 1.6 for downward) are also found but are not statistically different from one another (Prob. Chi 2 =0.487). Note that in all models we control for the incidence of internationally- coauthored papers. In sum, the results suggest that mobility of either kind is positively associated with performance. We move on to investigate whether the observed superior performance of returnees is associated with any international experience, or if it is exclusively the experience of work or education abroad that correlates to a performance differential. The literature on the mobility of skilled- human capital highlights that workers tend to move abroad in anticipation of a salary premium exceeding the cost of mobility. If salary is a function of personal skill- levels, movers are positively selected from their country of origin. In our case this implies that those who move to take job positions are positively selected among the top achievers in their country of origin. To the extent that PhD study is supported with financial assistance, we should also observe a similar positive selection. We know, however, that the amount of financial support available varies by country and largely dependent on government budgets (Bound, Turner, & Walsh, 2009), and quite inelastic to actual demand in a given year. As a result, we expect the positive selection effect to be less evident for those who leave for a PhD than for those who leave for work. In the extreme case, i.e. for bachelor and master degrees, where it is typically the 17 H- Index is the number of papers that received as a minimum an equal number of citations. We use H- Index by country and by subject category, computed for all Scopus publications during the interval by the Scimago Journal and Country Rank. Retrieved from on April 18,

64 student who pays to receive an education, we would expect the selection effect to be virtually nonexistent. Table 7 Performance of RETURNEES Robust OLS - v1- - v2- - v3- - v4- Robust Neg. Bin. - v1- - v2- - v3- - v4- RETURNEE 0.636*** RETURNEE 0.212*** (0.057) (0.026) lowerh_country 0.270*** lowerh_country 0.194*** (0.096) (0.043) higherh_country 0.808*** higherh_country 0.227*** (0.065) (0.028) BA_MA BA_MA (0.149) (0.124) phd phd (0.122) (0.064) post 0.742*** post 0.209*** (0.064) (0.026) job job (0.100) (0.035) MA_quality MA_quality (0.213) (0.178) phd_quality 0.279* phd_quality (0.167) (0.082) work_quality 0.752*** work_quality 0.180*** (0.065) (0.025) AGE 0.048*** 0.045** 0.045** 0.047** AGE (0.019) (0.019) (0.019) (0.019) (0.009) (0.009) (0.009) (0.009) SQ_AGE *** *** *** *** SQ_AGE (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) female *** *** *** *** female *** *** *** *** (0.063) (0.063) (0.063) (0.063) (0.026) (0.026) (0.026) (0.026) emerging 0.937*** 0.932*** 0.936*** 0.932*** emerging 0.339*** 0.338*** 0.339*** 0.339*** (0.063) (0.063) (0.063) (0.063) (0.024) (0.024) (0.024) (0.024) interdisc interdisc (0.056) (0.056) (0.056) (0.056) (0.023) (0.023) (0.023) (0.023) inter_cooperated 0.150** 0.151** 0.161** 0.170** inter_cooperated 0.101*** 0.101*** 0.105*** 0.108*** (0.070) (0.070) (0.070) (0.071) (0.027) (0.027) (0.027) (0.027) still_training still_training (0.140) (0.140) (0.140) (0.141) (0.075) (0.075) (0.075) (0.076) n_author 0.189*** 0.189*** 0.189*** 0.189*** n_author 0.066*** 0.066*** 0.066*** 0.067*** (0.016) (0.016) (0.016) (0.016) (0.005) (0.005) (0.005) (0.005) field dummies yes yes yes yes field dummies yes yes yes yes current country dummies yes yes yes yes current country dummies yes yes yes yes Constant Constant 2.100*** 2.100*** 2.114*** 2.111*** (0.509) (0.509) (0.509) (0.514) (0.348) (0.347) (0.329) (0.328) Ln_alpha *** *** *** *** (0.020) (0.020) (0.020) (0.020) Dependent variable IF IF IF IF Dependent variable TC TC TC TC Observations 11,227 11,227 11,227 11,072 Observations 11,223 11,223 11,223 11,068 R- squared Adj. R- sq F stat *.10 **.05*** 0.01 (Heteroschedasticity- robust st. errors in parenthesis). Baseline: NATIVES- NOT MOBILE We examine these effects in Model (3) of Table 7, where dummy variables have been included which take the value of one depending upon whether the returnee had a mobility experience for pre- doctoral studies (MA_BA), doctoral studies (phd), postdoctoral study (post- doc) or job positions (job). Results indicate quite clearly that pre- doctoral education (Bachelor & Master or PhD) abroad is not associated with 64

65 superior performance after returning. The same is true for the natives who went abroad to take a PhD or take a job. The superior performance effect holds only for returnees who went abroad to take a postdoc position. Although this result is not conclusive, it suggests ex- ante selection effect in explaining the differentials in performance. This conclusion is consistent with the findings in Model (4). Here, we have substituted the dummy- variables for the different types of experience abroad with variables that take a value of zero if the person did not move and take the value of the H- index of the country where the respondent experienced mobility. This model is meant to control for differentials in the strength of the science base of the destination country. Experience of a postdoc and job here are grouped into a single variable (work_quality) because we do not have the information on destination country of postdoc separately from that of job. In sum, after controlling for the relative strength of the country where the mobility experience occurred, it is only mobility for reasons of work (and we infer for reasons of a postdoc) that is associated with superior performance differentials of the returnees, compared with the non- mobile natives. In sum, our data are consistent with the existence of long- term differentials in the performance of the natives with prior mobility experiences from the performance of natives without international experience, although this is not the case in the United States. In general, the performance edge accrues to those who did a postdoc abroad. While the effect is consistent with a selection effect, it may also be due to the networks that returnees build while abroad and are able to maintain upon their return. 6. Robustness checks As highlighted in the data section, the surveys were administered separately in the 16 countries. Consistently, the rate of response varies by country. Thus, when the data are pooled together to estimate the models, response rate differentials result in over- weighting the answers from scientists in countries whose response rate is above the average (Australia, Brazil, Canada, Denmark, Italy, Spain) and under- weighting the answers from scientists in countries whose response rate is below the average (Belgium, Germany, India, Japan, Netherlands, Sweden, Switzerland, UK, US). To neutralize this potential source of bias, all models were re- estimated by including probability weights. All results hold; see the Appendix. 65

66 Information on age and gender were asked at the end of the questionnaire. As a result, information on these variables is less complete than information on past mobility. Collectively, there are 988 responses for which the information of age or gender is missing, largely because the respondents dropped out before completing the survey. Given that gender and age are typically important control variables when exploring scholarly performance and have thus been included in our estimates, the missing items result in dropping observation when we run the multivariate estimates. Although we have shown in Section A1 of the Appendix that the evidence of potential biases caused by dropouts is very limited, nonetheless, the units dropped because of incomplete information in the control variables could potentially bias the estimates. In order to check the robustness of our estimates to these dropped observations, we imputed the age and gender of the respondent and re- estimate the models with the imputed values, using the multiple imputation method (Rubin, 1987) with 5 imputations for each variable with missing instances. The imputation models include both variables of observation (country, and foreign experience of higher education, PhD, postdoc, job and experience of visiting abroad) and auxiliary variables from our dataset (job position/tenured- untenured, affiliation type, existence of secondary affiliations, field of research). They also include probability weights. The imputation model generates imputed variables for age and gender in all but 6 cases. The variable of squared- age is produced as a passive imputation of age. Overall the imputation procedure yields a sample of from a possible observations. Model estimates that take into account observations with imputed values for gender, age and age- square are presented in the Appendix. The main findings remain after including these observations. 7. Discussion and conclusion We find compelling and consistent evidence that mobile individuals perform at a higher level than non- mobile individuals, using several definitions of mobility and two measures of performance. These results are consistent with mobility being positively correlated with selection. They are, however, also consistent with treatment effects in the sense that mobility can enhance networks and lead individuals to work in environments that enhance their performance. The cross section nature of our data does not permit us to differentiate between the two types of explanations. 66

67 Our results suggest that on average foreign- born have a performance premium over natives who were not mobile of about 0.84 as measured by the Impact Factor of the survey article and 2.29 as measured by the total citations that the survey article received. The positive foreign- born premium holds for both those who were trained away from the host country but immigrate for a job as well as for those who were trained in the same country of the non- mobile natives. We also find that domestic scientists who studied or worked abroad and subsequently return to work and live in their country of origin (the returnees) perform at a higher level than natives who were not mobile. Returnees have on average 0.63 higher IF than natives with no prior mobility and 1.69 higher TCs. The empirical results also suggest that not all mobility experiences are equal. To be more specific, the mobility experience that is systematically associated with a performance premium is that of a postdoc; other types of mobility experience are uncorrelated with the performance of returnees. An exception is the US, where returnees perform in general at the same level as natives who were not mobile. Experience of visiting alone is not a substitute for other types of mobility and is never associated with a performance premium. When returnees are compared to fellow nationals who remained abroad, we find little evidence that they perform less well. Exceptions are returnees to India, Brazil and Italy. Regardless of country, we do, however, find evidence of decay in the number of citations received after one returns. The decay is modest and amounts to receiving approximately one fewer citations in the two years between publication and measurement for scientists who have been back in their native country for ten years. 67

68 APPENDIX The appendix is composed of four sections: A.1 Representativeness of sample A.2 Who returns. Selection of returnees A.3 Estimates weighted by probability of response A.4 Imputation of missing control variables A.1 Representativeness of sample We performed three different checks concerning potential differences in scientific impact, as assessed by the journal impact factor of source articles. A final check is performed on a self- reported assessment of article representativeness. Table A- 1 Two- groups comparisons of Impact Factor in 16 countries by 4 subject categories. T- Tests. Hypothesized difference (respondents non- respondents)=0 COUNTRY DIFFERENCE (RESPONDENTS - NON- RESPONDENTS) Biology Chemistry Earth & Environment Materials Science mean st.err. mean st.err mean st.err mean st.err Australia Belgium Brazil * Canada Denmark France Germany India Italy * Japan Netherlands Spain * Sweden Switzerland UK * USA * * * OVERALL * * * +Weighted by probability. *p<0.5 First, we asses potential bias due to unit- non response by comparing the Impact Factor of respondents in each of four subject categories and each of 16 countries, against those of non- respondents (Table A- 1). Results indicate modest potential biases in the samples from Brazil (Materials science) Italy (Biology), Spain (Materials Science) and 68

69 US (Biology, Chemistry and Materials Science) 18. Except for Spain, biases are in the direction of over- representing authors with higher- impact papers. Second, we compare the impact factor of early respondents against those of late respondents. Late respondents are characterized as those who completed the questionnaire during the third (final) round, as opposed to those who completed the questionnaire during the first and second rounds (Table A- 2). This screening is useful to assess the potential existence of biases due to item non- response, but can also be helpful to assess the severity of bias for unit- non- response, if we expect that late- respondents would be more similar to non- respondents (or to those who would have responded if we had solicited the questionnaire one more round). T- test comparisons highlight modest biases concerning the Japan sample in Biology, and the Sweden and US samples in Earth & Environmental sciences, where authors of higher- impact contributions were disproportionately distributed among late respondents 19. Table A- 2 Two- groups comparisons of Impact Factor in 16 countries by 4 subject categories. T- Tests. Hypothesized difference (early respondents late respondents)=0 COUNTRY DIFFERENCE IMPACT FACTOR (EARLY - LATE) Biology Chemistry Earth & Environment Materials Science mean st.err mean st.err mean st.err mean st.err Australia Belgium Brazil Canada Denmark France Germany India Italy Japan * Netherlands Spain Sweden * Switzerland UK USA * OVERALL Weighted by probability. *p<0.5 Third, we compare respondents who took the entire questionnaire against respondents who dropped- out before completing the survey (Table A- 3). This check is meant to 18 Note that only that Bonferroni- adjusted p- value would indicate significant differences and only for the US sample. 19 Note that no difference would be significant if Bonferroni- adjusted p- values were used. 69

70 assess potential biases due to item- non- response, for example caused by the fact that certain incomplete observations (like country of origin) made the response not- usable for our purposes. Moderate biases are highlighted for Belgium (biology), Germany and India (Chemistry), Japan and US (Earth & Environment), where authors of higher- impact papers were comparatively more likely to complete the questionnaire 20. Table A- 3 Two- group comparisons of Impact Factor in 16 countries by 4 subject categories. T- Tests. Hypothesized difference (complete respondents dropped- out respondents)=0 COUNTRY DIFFERENCE IMPACT FACTOR (COMPLETE - DROPPED- OUT) Biology Chemistry Earth & Environment Materials Science mean st.err mean st.err mean st.err mean st.err Australia Belgium * Brazil Canada Denmark France Germany * India * Italy Japan * Netherlands Spain Sweden Switzerland UK USA * OVERALL * Weighted by probability. *p<0.5 In sum, the controls performed on the set of information known ex- ante point to moderate evidence of bias. If biases exist, they seem to be more likely in the direction of slightly over- sampling correspondent authors of higher quality papers. A consistent picture is given by the last screening we performed, which, unlike the previous ones, is based on a self- reported assessment of representativeness. In particular, we asked the respondents the degree to which the selected paper represented their average production Note that Bonferroni- adjusted p- values would only indicate as significant the differences between samples in the Belgian sample in Biology. 21 The question was formulated as asking the respondent to report on a 1-5 scale the agreement to the following statement: This particular paper of yours is of a higher quality, with respect to my other papers. 70

71 Figure A- 1 shows the histogram of the distribution of the answers. The distribution is overall well- behaved, with a slight right- skewness (mean 3.20; median 3.0). Figure A- 1 Self- reported assessment of the representativeness of sample article (1-5 scale) A.2 Selection of returnees In addition to looking at the performance of returnees, it is interesting to examine evidence concerning selection effects in return migration. In other words, do academics who were mobile from a country and later decide to return exhibit comparable performance to their compatriots who have been mobile and remain abroad? The standard selection models for return migration suggest that the direction of the initial selection, be it positive or negative, will be further strengthened by return migration (Borjas & Bratsberg, 1996). In other words, if migrants are positively selected, as our data suggest, those who return will be the worst of the best, those who remain are the best of the best. Thus, the foreign born remaining abroad would be survivors of a second selection process where only the brightest receive sufficiently good enough offers to convince them to stay and returnees would be those who did not receive such offers. However, if the scientist is not or not primarily motivated by earnings in her decision to return, but, for example, motivated by family or personal reasons, the correlation may 71