Skip to main content
SpringerLink
Log in

The Effects of International Mobility on European Researchers: Comparing Intra-EU and U.S. Mobility

  • Published:
Research in Higher Education Aims and scope Submit manuscript

Abstract

Using econometric analysis on survey data from European-born and European-educated researchers who are internationally mobile after their PhD within Europe or to the United States, we find significant positive effects from international mobility on scientific productivity, as well as several other positive career development effects. European researchers mobile to the United States consistently report stronger positive effects on their scientific productivity and on their career development compared to their peers who are mobile within the EU. A propensity score matching analysis shows that this apparent ‘U.S. premium’, is almost entirely due to the different characteristics of those mobile researchers who move to the US compared to those who move intra-EU. After accounting for this selection, there is no longer any significant difference in reported scientific productivity effects between U.S.-mobile and EU-mobile researchers.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. This study was carried out by IDEA Consult in consortium with NIFU Step, WIFO, Logotech and the University of Manchester for the European Commission in 2009–2010.

  2. A small number of researchers are mobile to other countries, such as Australia, but these are omitted from the analysis. Canada was originally considered together with the U.S. as one destination, North America, but after cleaning no European researchers mobile to Canada remained in the sample.

  3. A mobility experience is defined in the MORE survey as being at least 3 months. Unfortunately we could not analyse any differences in the duration of the mobility event beyond this 3 months cut-off.

  4. More than half of the respondents responded that effects on patent out put and ability to work in industry were not applicable to them. A little over a hundred researchers also did not indicate an effect on access to infrastructure.

  5. Sauermann and Rauch (2012) suggest the use of a “neutral question” as reference to correct for this bias. Unfortunately the MORE survey did not contain such a question.

  6. Multinomial probit analysis is preferred over multinomial logit analysis as it does not require the Independence of Irrelevant Alternatives assumption.

  7. As the categories are ordered, ordered logit or probit would have been an alternative option. But as we are particularly interested in how the independent variables play out differently for positive compared to highly positive outcomes, we don’t want to treat the switch from negative and neutral to positive in a similar manner than the switch from positive to very positive. The multinomial analysis allows more flexibility on this.

  8. In this case, the relative risk ratio for strongly positive effects for the US mobility coefficient becomes 1.562(0.223)***; for positive effects 1.254 (0.159)* both relative to neutral or negative effects.

  9. In this case, the relative risk ratio for strongly positive effects for the US mobility coefficient becomes 1.591(0.321)**, for positive effects 1.170 (0.196), for negative effects 0.441(0.145)** all relative to neutral effects.

  10. As already said, for the case of publication output, multinomial probit and logit models give similar results (cf Table 2 versus 3).

  11. The high relative risk ratios for strongly positive effects for patent output and ability to work in industry should be handled with care in view of the smaller number of observations.

  12. The survey asked researchers to score 7 motivations for mobility on a scale from 1 to 5, ranging from not important at all to extremely important. These motivations were regrouped and averaged into three motivation factors, including career motivations, personal motivations and financial motivations. Similarly, the survey asked respondents to score 8 external influencing factors for mobility on a scale from 1 to 5, which were regrouped and averaged into regulatory factors, personal factors, concerns about funding, potential loss of contact sand language. See Van Bouwel and Veugelers (2013) for more on the analysis of the determinants of mobility.

  13. As a robustness check we also do nearest-neighbor matching. Although the outcomes are very similar, nearest-neighbor matching does not result in a ‘perfect match’, in the sense that some significant differences between the treated and untreated groups remain. We therefore only report Kernel matching results.

  14. Results on split samples by scientific disciplines (results not reported) show that the effect on access to infrastructure and professional experience are mainly attributable to researchers in the exact sciences.

  15. The share of researchers reporting strong effects for patent output is small: 8 % of U.S.-mobile researchers report strong increases in patent output, compared to 1 % of EU-mobile researchers. A split sample analysis (not reported) shows that these effects on patent output are driven by researchers in the exact sciences.

  16. This study was carried out by IDEA Consult in consortium with NIFU Step, WIFO, Logotech and the University of Manchester for the European Commission in 2009–2010.

  17. The search identified html-pages or pdf files which match a few keywords that identify an academic CV and likely mobility between the US and the EU. The resulting list of e-mail addresses was the primary direct sampling source. Indirect sampling methods were also used, including publishing the survey on LinkedIn and forwarding it to the Euraxess community, the EU Centres of Excellence in the U.S., and the coordinators of the ATLANTIS Programme on EU-U.S. Cooperation in Higher Education and Vocational Training.

  18. We are grateful to the Belgian Federal Science Policy Office for allowing us access to the data.

  19. For example, one of the questions in the EU-U.S. mobile group is ‘To what extent were the following aspects important as factors motivating you to become mobile to the U.S.?’, whereas the mirror questions for the non-mobile group is ‘To what extent were the following aspects important as factors dissuading you to become mobile?’. Aspects such as family considerations may be given little weight by the first group if they became mobile despite family considerations, even if family considerations received a lot of weight in the overall mobility decision, and vice versa for the non-mobile group.

References

  • Albarraan, P., & Ruiz-Castrillo, J. (2012). The measurement of scientific excellence around the world, Economics Working Paper, Universidad Carlos III, Madrid.

  • Almus, M., & Czarnitzki, D. (2003). The effects of public R&D subsidies on firms’ innovation activities: The case of Eastern Germany. Journal of Business & Economic Statistics, 21(2), 226–236.

    Article  Google Scholar 

  • Altbach, P.G. (2004). Higher education crosses borders: Can the United States remain the top destination for foreign students? Change: the magazine of higher learning 36(2), 18–25.

  • Azoulay, P., Zivin, J., & Sampat, B. (2012). The diffusion of Scientific Knowledge across Time and Space. In J. Lerner & S. Stern (Eds.), Rate and direction of inventive activities revisited (pp. 107–155). Chicago, IL: University of Chicago Press.

    Chapter  Google Scholar 

  • Becker, G. S. (1962). Investment in human capital: A theoretical analysis. The Journal of Political Economy, 70(5), 9–49.

    Article  Google Scholar 

  • Black, G., & Paula, E. S. (2007). In P. E. Stephan & R. C. Ehrenberg (Eds.), The importance of foreign Ph.D. students to U.S. science, in science and the university. Madison: The University of Wisconsin Press.

    Google Scholar 

  • Brand, J., & Halaby C. (2003). Regression and matching estimates of the effects of elite college attendance on career outcomes. Working paper. Madison: University of Wisconsin.

  • Edler, J., Fier, H., & Grimpe, C. (2011). International scientist mobility and the locus of technology transfer. Research Policy, 40, 791–805.

    Article  Google Scholar 

  • Finn, M. G. (2010), Stay rates of foreign doctorate recipients from US universities 2007. Oak Ridge: Institution for Science and Education.

  • Gaulé, P. (2010). Return migration: Evidence from Academic Scientists, working paper presented at NBER productivity lunch seminar on November 16th, 2010, Boston, MA.

  • Goldin, I., Cameron, G., & Balarajan, M. (2011). How migration shaped our world and will define our future. Princeton, NJ: Princeton University Press.

  • Ham, J., Li, X., & Reagan, P. (2003). Propensity score matching, a distance-based measure of migration, and the wage growth of young men. Working Paper. Columbus: Department of Economics and CHRR Ohio State University.

  • Heckman, J., Ichimura, H. & Todd, P. (1997) Matching as an econometric evaluation estimator: Evidence from evaluating a job training programme. Review of Economic Studies 64, 605–654.

  • Hoisl, K. (2007a). Does mobility increase the productivity of inventors? Journal of Technology Transfer, 32, 195–215.

  • Hoisl, K. (2007b). Tracing mobile inventors: The causality between inventor mobility and inventor productivity. Research Policy, 36(5), 619–636.

  • Horta, H., Francisco M.V. & Rócio G. (2010). Navel gazing: Academic inbreeding and scientific productivity. Management Science, 56, 414–429.

  • Hunter, R. S., Andrew, J. O., & Bruce, G. C. (2009). The elite brain drain. Economic Journal, 119, 231–251.

  • Kahn, S., & McGarvie, M. (2012). The Effects of the foreign fulbright program on knowledge creation. In J. Lerner & S. Stern (Eds.), The rate and direction of inventive activity revisited (pp. 161–197). Chicago: University of Chicago Press.

    Chapter  Google Scholar 

  • Laudel, G. (2005). Migration currents among the scientific elite. Minerva, 43, 4, 377–395.

  • OECD (2008) The global competition for talent: Mobility of the highly skilled, ISBN978-92-64-04774-7.

  • Sauermann, H. & Roach, M. (2012). Taste for Science, Taste for Commercialization, and Hybrid Scientists, Georgia Tech Working Paper.

  • Stephan, P. E. (2012). How economics shapes science. Cambridge, MA: Harvard University Press.

  • Stephan, P. E., & Sharon G. L. (2001) Exceptional contributions to US science by the foreign-born and foreign-educated. Population Research and Policy Review, 20, 59–79.

  • Topel, R. H., & Michael P. W. (1992). Job mobility and the careers of young men. The Quarterly Journal of Economics, 107(2), 439–479.

  • Train, K.E. (2002). Discrete choice models with simulation. Cambridge, MA: Cambridge University Press.

  • Van Bouwel, L. & R. Veugelers, (2013), An “elite” brain drain: Are foreign top PhDs in the US less likely to return home? In S. Uebelmesser & M. Gerard, Cross Border Mobility of Students & Researchers. Cambridge, MA: CESifo Venice Summer Institute, MIT Press.

  • van Heeringen, A., & Dijkwel, P. (1987). Effects of mobility on productivity. Scientometrics, 11, 267–290.

  • Veugelers, R. (2010). Towards a multipolar science world: Trends and impact. Scientometrics, 82(2), 439–456.

    Article  Google Scholar 

Download references

Acknowledgments

This paper has benefited from the comments of Dirk Czarnitzki, Otto Toivanen, Paula Stephan, Henry Sauerman, Paul David, Frank erboven and the participants of the ENID STI Conference (Rome, September 2011), the TEMPO Conference (Vienna, November 2011) and the BRICK Workshop (Turin, April 2012). We are very grateful to IDEA Consult, in particular A. Verbeek and E. Lykogianni for allowing us access to the MORE data. Linda Van Bouwel acknowledges the support of the FWO through an aspirant-grant. Financial support from FWO (G.0825.12) and KULeuven (GOA/12/003) is gratefully acknowledged. This paper is part of the SCIFI-GLOW Collaborative Project supported by the European Commission’s Seventh Research Framework Programme, Contract number SSH7-CT-2008-217436.

Author information

Authors and Affiliations

  1. KU Leuven, Leuven, Belgium

    Reinhilde Veugelers & Linda Van Bouwel

  2. Bruegel, Brussels, Belgium

    Reinhilde Veugelers

  3. CEPR, London, UK

    Reinhilde Veugelers

Authors
  1. Reinhilde Veugelers
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Linda Van Bouwel
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Reinhilde Veugelers.

Appendices

Appendix 1: The More Survey and Our Sample

The analysis is based on survey data from the extra-EUMORE survey. The extra-EUMORE survey is part of a group of surveys carried out in the context of a study on mobility patterns and career paths of EU researchers (‘MORE’).Footnote 16 In the MORE survey, mobility is defined as a minimum 3-month stay in a country different from the country where the highest degree was obtained. The target group for the extra-EU MORE survey are researchers who obtained their highest degree in the EU and who worked in the U.S. for a minimum of 3 months. There are three additional target groups: researchers who obtained their highest degree in the U.S. and who worked in the EU for a minimum of 3 months, researchers who have been mobile but who do not belong to the previous two groups, and researchers who have not been internationally mobile.

The main sampling method of the survey was a web-based search.Footnote 17 The sampling method allowed to construct a relatively large group of respondents, but the sampling method also has drawbacks. The major drawback is a lack of information on the representativeness of the sample relative to the underlying population. The survey sample cannot be corrected for possible biases that distort its representativeness. The findings drawn from these data therefore cannot be generalized to the whole population, and are valid only for this particular sample. Despite these caveats, given the scarcity of data on EU researchers and particularly on mobility, this sample provides an interesting source of information for analysis.

The survey was initially sent to 93,183 e-mail addresses. Out of these, 22,206 people viewed the email and 5,572 responded (6 % of the total invited and 25 % of those who viewed the e-mail), of which 4,571 respondents fully completed the questionnaire. An additional 1,393 fully completed surveys were received from non-panel individuals, adding up to a total of 5,964 fully completed questionnaires. After cleaning out responses, a total of 5,544 responses remained. This sample was used for the MORE report for the European Commission (MORE 2010).

We have no data available to assess the extent to which this sample is biased towards EU-U.S. mobility. As a small check, we compared the researchers in our sample who are currently residing in Belgium to the Belgian sample of the Careers of Doctorate Holders survey (CDH) carried out in several OECD countries in 2006 in cooperation with the OECD, Eurostat and UNESCO Institute of Statistics.Footnote 18 The CDH sample only includes PhD holders currently working in Belgium and does not take into account researchers who moved abroad permanently or who have not yet returned. This biases the mobility rates in CDH downwards. The comparison reveals that our sample picks up four times as much ‘career mobility’ (i.e. mobility after the highest degree is obtained) as the CDH sample, and this mobility is more likely to be geared towards North America: 52 % of career mobility goes toward the U.S. in our sample, versus 12 % in the CDH sample. This indicates that the MORE sample is strongly biased towards EU-U.S. mobility. The true population mobility rates are likely to lie somewhere between the MORE estimate and the CDH estimate. As we have no good information to correct for the bias, we hope that the non-representativeness of our sample affects the alternative-specific constants in our econometric analysis, but not necessarily the estimates for the determinants of mobility destination outcomes (Train 2002). Nevertheless, results should be interpreted with caution, especially the descriptive statistics.

The survey itself consists of two parts: the first addresses all mobility groups and asks about researchers’ personal and family situation, education and training, current employment as a researcher and experience of mobility. The second part asks respondents about their views on mobility, including personal motivations for mobility, external influencing factors for the decision to become mobile and the effects they experienced from mobility. The second part differs by target group. Although the questions to the immobile group were designed to be ‘mirror questions’ to those for the mobile group, a different wording was used, which may have caused a different interpretation by mobile and non-mobile respondents.Footnote 19 This implies we cannot use the data to compare mobile researchers to non-mobile researchers. However, we can address the question whether the effects of mobility differ among researchers mobile to different destinations.

Appendix 2: Propensity Score Matching

To deal with a selection bias affecting the analysis of causal effects, matching techniques have become a popular approach. It is especially popular in studies evaluating labour market policies (e.g. Heckman et al. 1997) or R&D subsidies (e.g., Almus & Czarnitzki 2003), but it is also widely applied in other fields of study. It can be applied to all types of situations that can be considered as having a “treatment”, and a group of “treated” individuals and a group of “untreated” individuals which one wants to compare. The nature of the “treatment” may be very diverse. For instance, Ham et al. (2003) use the technique to study the effect of a migration decision on the wage growth of young men. Brand and Halaby (2003) use the technique to analyse the effect of elite college attendance on career outcomes.

The selection bias which the PSM addresses arises when we would like to know the difference between the participants’ outcome with and without the “treatment”, but we cannot observe the outcome of the same participant with and without the treatment. So we need to construct a proper counterfactual (control group) for the non-treatment outcome. Simply taking the mean outcome of non-participants as an approximation for the non-treatment outcome of the treated participants is not advisable, since participants and non-participants usually differ, even in the absence of treatment.

The matching approach is one possible solution to the selection problem. Its basic idea is to find a group of non-treated individuals who are similar to the participants in all relevant pre-treatment characteristics X. Differences in outcomes of this well selected control group and of the “treated” group can be attributed to the “treatment”. Since conditioning on all relevant covariates is limited in case of a high dimensional vector X, functions of the relevant observed co-variates X are used. Matching procedures based on the propensity score, i.e. the probability of participating in a “treatment” given observed characteristics X, are known as PSM and are used in this paper. The “treatment” is considered to be “mobility to the US”, where the comparison non-treated group is formed by intra-EU mobile researchers. The propensity score comes from estimating the probability of mobile PhD holders to go the US, rather than be intra-EU mobile. These results are reported in Appendix 3.

Appendix 3

See Table 7.

Table 7 Logit model for mobility to the U.S. versus intra-EU (N = 998)
Full size table

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Veugelers, R., Van Bouwel, L. The Effects of International Mobility on European Researchers: Comparing Intra-EU and U.S. Mobility. Res High Educ 56, 360–377 (2015). https://doi.org/10.1007/s11162-014-9347-6

Download citation

  • Received:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11162-014-9347-6

Keywords

Search

Navigation