Institution: University of Amsterdam, ASE
Academic year: 2013/2014
Name: Sofia Ghizzoni
Student number: 10257802
Specialization: Bachelor of Economics and Business, specialization in Economics
Field: International Labour Market
Number of credits thesis: 12
Title of the research: Investigating the relationship between the Italian brain drain and
the relative wage levels abroad.
1. Introduction
1.1 Motivation The topic of brain drain is a very living matter for Italian neo-graduates. In this
section the most burning issues and recent discussions rotating around this phenomenon are presented. Data are collected from both academic and non-academic sources, such as
European databases, recent researches, radio interviews and newspaper articles. The issues at stake are in particular: the awareness of Italian graduates of higher wages and better living conditions abroad and their willingness to take advantage of these opportunities. Moreover, it is taken into account the scale of the phenomenon of migration outflow from Italy and its impact on home’s society. To conclude, the main dysfunctions perceived by Italian graduates within Italian academic and working environment are described.
After five years from graduation, Italians employed abroad receive a much higher salary (€2,324) with respect to those working at home (€1,378) (AlmaLaurea 2013). It is estimated that a “wage spread” of €500 exists amongst Italians, who after 4 years from graduation are
employed abroad (€1,800) and those who work at home (€1,300) (Barone 2012). Moreover, AlmaLaurea observes that it is especially engineers, who gain the highest benefits from migration.
An exemplar case is that of Andrea Cremese, a 31 year-old mechanical engineer. After graduation he has accepted a job offer in London and subsequently in Hong Kong. In 2011 he returned to Italy, thinking that his curriculum, enriched by these international experiences, would have yielded better living standards. After less than one year, both him and his wife came to the realization that their career perspectives were not at all promising. Currently (2013) they live in New York, where they earn higher and especially equal salaries (differently from the situation in Italy, where his wife was earning 30% less). One week before this interview was taken (Sept. 2013), the Italian Minister of labour and social policies Enrico Giovannini, on the same radio programme, had blamed the crisis responsible for the lack of career opportunities in Italy. He claimed that especially in some sectors, salaries at home are much lower than abroad.
The issue of Italian brain drain itself is at the centre of Italian news today. A study shows that it is especially in the last year and a half (18 months) that the phenomenon of Italians expatriating has boomed (Triandafyllidou & Maroufof 2013). The study enumerates amongst the most frequent destinations: Germany, the UK, Switzerland and France. The first two countries are used as partner countries for this research as well. The above-mentioned study by
Triandafyllidou (2013) identifies amongst the main causes of this phenomenon of migration the lack of career opportunities for Italian neo-graduates.
This shortage of high skilled labour demand is in turn a consequence of two main factors: low investment in human capital and corruption. It is Minister Giovannini again, who appeals to companies, encouraging them to invest more in human capital, in order to gain in innovation for the future. He says the recently experienced economic crisis has created more awareness amongst companies: from the 30% observed in 2005, the percentage of firms investing in human capital has increased to more than 50% in 2010. This result is still far behind from investment levels in other countries. To provide a partial justification for these low numbers, the Minister justifies: “We have a production system composed in most industries by small
businesses, and salaries in small businesses are lower than in bigger ones, because of differences in the scale of output” (Radio24, Sept. 2013).
The second factor strongly influencing the scarce level of career opportunities is that of corruption, as the study by Ariu and Squicciarini (2013) shows. They argue that there are two parallel channels on which corruption acts: on the one hand, skilled natives leave their corrupt home country, because of their integrity, seeking for more virtuous destinations, where job positions are assigned based on meritocratic criteria. At the same time, inbound skilled
migration is low, since foreign talents would have to penetrate a highly corrupted labour market, which would require lobbying people in high positions (Ariu and Squicciarini 2013, p. 3). In the radio interview (Radio24, Nov. 2013), Squicciarini (2013) explains the findings of their research: “In corrupted countries such as Mexico, Korea, India and Italy, there is a net negative balance between skilled migration inflow and skilled migration outflow”. The levels of corruption in Italy are quite shocking: in the 2013 CPI index1, Italy is placed 69th in the world, tied up with Kuwait and Romania, with a score of 4.3 out of 10.
1.2 Research question and hypothesis Is the number of Italian high skilled workers, who
decide to emigrate, positively correlated with the respective trends of wages in the selected partner countries? In other words, is the brain drain from Italy due to better living perspectives in other countries? I expect there to be a positive correlation between the ratio of Italian to foreign wages and the number of Italian high skilled workers, who migrate to that country. If a positive correlation between brain drain and wages differential is found, my hypothesis is proved right and it is confirmed that the higher the wages abroad relatively to home, the more people decide to move. However, if a negative correlation is found, my hypothesis is proved wrong. In this
1
The Corruption Perceptions Index ranks countries/territories based on how corrupt their public sector is perceived to be. A country/territory’s score indicates the perceived level of public sector corruption on a scale of 0 - 10, where 0 means that a country is perceived as highly corrupt and 10 means that a country is perceived as very clean. A country's rank indicates its position relative to the other countries/territories included in the index. The 2013 index includes 177 countries/territories.
case, results would be even more interesting, in that it seams like the higher the relative wages abroad, the lesser people move to that country. Or, more simply, that the higher benefits provided by wages alone are not a big enough incentive for people to make the step of migration.
2. Related literature “Studies of the basic determinants of international migration flows are
scarce, and those that can identify causal relations are even scarcer” (Yang 2003, p. 2). Morano Foadi (2006) investigated the causes of Italian brain drain. This section presents an overview of the research and studies done so far on the topic of brain drain in general and on strictly related matters, such as the high skilled job market, job polarization and unemployment. Moreover, it analyses existing literature on issues regarding specifically Italian brain drain. This literature review section constitutes a basis and source of inspiration for the development of the research method and data gathering.
Research has started to pay particular attention to the phenomenon of high skilled migration from the 1980s, when economic globalization and the emergence of transnational corporations began to increase migration within the corporate sector. Further on, in the 1990s, the quickly changing political situation, especially in Eastern Europe and in the foreign Soviet Union gave further reason to focus on the concept of “brain waste” and brought about the implementation of policies regarding the growing “immigration market” (Koser and Salt 1997, p. 286). Moreover, a study by Goos et al. (2009, p. 61) shows that, between1993 and 2006, the share of high-paying occupations has increased substantially in Europe (6.19%). This increase in high-paying occupations has grown parallel to a net decrease (-7.77%) of mid-paying jobs and a lower increase of the lowest-paying jobs (1.58%). This phenomenon is called job polarization and constitutes the first piece of evidence for the growing phenomenon of brain circulation. The second important issue to be considered as a basis of brain circulation is European labour mobility. The study by Gill (2008, pp. 331) finds out that students, who have studied one year or more abroad, are more likely to look for a job abroad. The reason for this relationship is that these students have developed a European identity and have expanded their open-mindedness (Gill 2008, p. 332). This student mobility amongst European countries has thus given rise to European high skilled labour mobility and it has become more common in recent years to talk about “brain circulation” or “brain exchange” (Morano Foadi 2006, p. 209). However, for Italy this is not the case; in fact, Italy supplies talents to Europe and America, but fails in attracting foreign scientists from abroad (Morano Foadi 2006, pp. 209-210). The main causes of Italian brain drain identified by Morano Foadi (2006) are low investment in research,
low salaries and lack of facilities, impenetrable recruitment system, corruption, nepotism, bureaucracy and networking.
Koser and Salt offer four different methods for analysing this topic: macro-level, meso-level, micro-level and multi-level models (1997, pp. 291-292). This research will make use of a model of the macroeconomic type. What characterizes these kinds of models is that they focus on international as well as national economic conditions and their influence on the demand and supply of highly skilled people (Koser and Salt 1997, p. 291). My analysis will be conducted by investigating the relationship between changes of the domestic variable (migration outflow), the
foreign nation’s state of the economy and the home nation’s state. My research aims at finding a
correlation between the number of migrants from Italy and both, variables in the destination and in the home country. A similar analysis was conducted by Yang (2003, p. 21), who investigated the relationship between the number of new departures for overseas work from the Philippines, given rainfall shocks within the country and shocks in the foreign exchange rates generated by the Asian financial crisis: Yang (2003) aims at finding a causal relationship between the
dependent variable (migration outflow), domestic shocks as well as a foreign one. The
econometric model used by Yang (2003,) is of fundamental importance in defining the model for this analysis. This will be addressed more closely in the research method section.
3. Data and empirical model
3.1 Data Framework According to a general supply and demand model of labour mobility, the
more labour moves abroad, the higher the wage becomes in the home country (see Fig. 1 in the Appendix). We observe a movement of labour from home to foreign and a subsequent decrease (from point C to point A) of real wages (MPL =!!) at home. The Marginal Product of Labour is defined as the change in output resulting from employing an additional unit of labour. The formula shows how it is productive to hire an extra unit of labour up to the point where the additional benefit of it equals its cost (real wage). Migration inflow in Italy is especially of low skilled workers, whereas migration outflow from Italy is increasingly amongst the high skilled workers (Morano Foadi 2006, p.217). This should yield, as the graph predicts, higher wages for Italian graduates at home. Data collected on Eurostat show that unemployment rate of graduate workers in Italy has an upward slope from 2011 and is on average around 8%, whereas the average European level is 6.5%, as it can be seen in Fig. 3. Moreover, Italian annual earnings for graduates are much lower in terms of Purchasing Power Standards (41,900) compared to European ones (43,200) and those of other countries such as Germany (59,700), the UK
(43,400), France (42,400) and Switzerland (69,400). These last data are taken from a survey on EUROSTAT conducted in 2010. These findings are unexpected, knowing that the level of high skilled migration outflow in Italy over the past decade has been quite significant (Morano Foadi 2006, p. 217). Looking at Table 1, we do observe a 16% increase in wages for high skilled workers in Italy, however it has to be taken into account that these measurements are not PPP-adjusted. Why then, do Italian graduates continue to expatriate? By focusing on relative data, this research will try to show the reasons.
In this analysis the partner countries are selected amongst European ones. The reason for this choice is that, by looking at Fig. 2 we can see that the largest portion of Italians (aged 30-34) graduates and non- decides to migrate to countries within Europe. This could be the result of specific measures implemented within the European Union to facilitate labour mobility. Some of these policies are cited in the European Job Mobility Action Plan (2007 – 2010), which has as main goals to improve existing legislation regarding workers mobility and to foster awareness of the advantages of mobility. For example, in 2008 a small-scale pilot project was launched called “Your First Job Abroad”, to offer support for young international job seekers. Moreover, in order to choose the destination countries, we first take a look at the European countries with the highest level of total immigration (graduates and non), since specific data on the level of education (graduates) are not available. In Table 2 we observe that the highest percentage of Italians (aged 30-34, regardless of their education level) emigrate to Germany, the UK, Switzerland and France. The two countries with the largest share of migration inflow are selected for our analysis (Germany and the UK) not only for their relevance, but also for reasons related to availability of data.
The Italian brain drain is measured through Italians with tertiary educational qualification, emigrating abroad. Koser and Salt (1997, p. 287) argue that it may not be sufficient to be a graduate in order to be regarded as highly skilled on the labour market. However, this research makes this simplifying assumption. Data on polarized migration are not yet fully available, because of the novelty of this phenomenon. Moreover, another issue related to measuring migration is that of the terms of stay (permanency, long-term, shorter-term). In fact, it cannot be discerned easily whether the emigrants will make return to their home (Koser and Salt 1997, p. 299).
3.2 Variables description In this section the variables that are used to perform the econometric
is presented. Last, the observations that are gathered to identify them are provided and the source from which these data were extrapolated is cited.
The dependent variable in the regression analysis is the dependent variable Brain Drain (BD). This variable measures the expatriation of Italian graduates to other countries. To be more specific, BD!"!!"#$%&',! is measured as the number of Italian graduates aged 30-34, who have migrated to a selected destination country. So that, for example BD!"!!" is the number of Italian graduates in Germany. The reason why this variable was chosen is that Italians, who graduate abroad are more likely to look for a job abroad, as Gill (2008) testifies: they are most likely entering in the labour force of that country. The data are collected on EUROSTAT in the section Education. The database provides precisely the number of Italian students who have obtained a bachelor and master degree in the partner country. These data are available only in the time lag from 2007 to 2011.
To move on, we take into consideration the independent variable, wage!. This variable measures the relative wage abroad with respect to home. This variable is therefore defined as follows: wage! = !"#$!"#$%&'!!"#$!"#$
!"#$!"#$ The subscript "i" indicates i = Germany, UK. Data on this
variable were harder to collect, given the only recent rise of the phenomenon of wage
polarization. The reason why this variable is constructed in this way is, because the research aims at finding a relationship between the brain drain and the relative wages. The relative data are used to take into account a term of comparison between the two countries. For each country, different sources were used. Statistics about Italy were collected on Almalaurea (footnote 1) and show the monthly earnings in euros after five years from graduation. Observations are available from 2002 until 2011. Data on Germany were collected on the ECB website, which in turn, has drawn from the EUROSTAT database. These data show annual compensation per employee occupied in professional, scientific and technical activities. The currency of this database is also euros, and the data are not seasonally nor working day adjusted. These data are available from 2002 until 2012. Finally, data on wages in the UK were collected on the Office for National Statistics website (ONS), in particular on the annual survey of hours and earnings. The data are about weekly wages of people occupied as professionals; they are measured in pounds and represent the pay unaffected by absence and the observations are available from 2009 until 2013. Since all the observations were in the home currency, in order compare the wage levels in the different countries, these data have been PPP-adjusted. The historical values of the Purchasing Power Parities Index were found on the OECD National Accounts Statistics.
In order not to incur in omitted variable bias, some control variables are included in the model. As we have previously seen in the related literature section, there are a number of control variables that are most likely relevant for this regression. First of all, corruption is used as a control variable. This variable measures the amount of perceived corruption in the public sector within a certain country. The control variable of corruption is measured in the same relative term as that of wage: corruption! =!"##$%&'"(!"#$%&'!!"##$%&'"(!"#$
!"##$%&'"(!"#$ . Ariu and Squicciarini
(2013, p. 3) use the ICRG corruption index as their measure for corruption. However, these data are not freely available. For this reason, the measure for corruption used in this research is the Corruption Perception Index (CPI), which is available for every year from 2001 until 2013 on Transparency International. The second control variable is that of R&D expenditure (RD). This variable measures the amount of expenditure on intramural R&D as a percentage of GDP, no matter what the source of funds is. As the other dependent variables, RD is measured as follows: RD!=!"!"#$%&'!" !!"!"#$
!"#$ . This value is reported on the Global Innovation Index, which is
published every year by the World Intellectual Property Organization (WIPO) and the source is the UNESCO institute for statistics, which yearly cooperates with other research institutions. Observations are available from 2008 until 2013. The third and last control variable is that of unemployment (unemp). This variable measures the percentage of people holding a bachelor and a master degree aged between 30 and 34 years old, who have no paid occupation. Unemployment is included in the regression as follows: unemp! =!"#$%!"#$%&'!"#$%!!"#$%!"#$
!"#$ . The
reason why this variable is chosen is that relative unemployment levels of high skilled workers might have a significant influence on the decision of a graduate to emigrate abroad. These data are found on EUROSTAT and are available from 2005 until 2013.
3.3 Model specification Given the variables described in the previous section, we come now to
the formulation of the basic model:
BD!"= α + βwage!+ u!
However, in order not to incur in omitted variable bias, other variables are added. As we have seen in the related literature review and motivation section, the other main factors influencing Italian brain drain are: corruption, R&D expenditure, unemployment.
4. Results Performing the multiple regression described above, with the little data available,
yields unobservable results for the UK, whereas for Germany, though insignificant, we can observe some results. The problem for the UK regression is that of high dimensionality. In fact, only four observations per variable are available and in total there are six variables. The lack of data availability and, as a consequence, that of high dimensionality, can be overcome by
removing one or more variables from the model. We try to run the regression again by removing one or more control variables.
What I expect to find as a result of the multiple regression is the following: I expect there to be a positive correlation between the brain drain and the wage variable, in that as the wage differential grows, more people are expected to exploit the opportunity of a relatively higher salary abroad. Second of all, I expect there to be a positive correlation with the corruption perceived index. In fact, it has to be kept into mind that the higher the score, on a scale from 1 to 10, the less corrupt the country is perceived to be. Third, the investment on human capital variable is expected to have a negative correlation with the brain drain. This is because R&D expenditure as a variable is measured in world ranking, so that the lower the number, the better positioned the country. Therefore, given the definition of the R&D variable provided above, the lower the value, the higher the expenditure in the abroad country with respect to the home country, the more Italians are expected to flee. To conclude, the unemployment variable is expected to have a negative correlation with the brain drain. In fact, the lower the
unemployment in the foreign country and the higher in Italy, the higher are the incentives for migrating.
The results of the regression are shown in Table 3 for Germany: omitting CPI (3), omitting RD (4) and omitting Unemployment (5). For the UK, results are shown in Table 4: regressing Brain Drain with wage and CPI (9), with wage and RD (10) and with wage and Unemployment (11).
Table 3. Germany: Brain Drain equations for assessing the validity of the models
Variables (1) Model (3) Model (4) Model (5) Model
Wage -4365 (-3.89) -3874 (-7.01) -2664 (-2.35) -2298 (-1.62) Corruption -752 (-0.55) (omitted) 1499 (1.27) 2516 (3.82) RD expend 7292 5597 (omitted) -460
(1.98) (3.41) (-0.12) Unemployment -4565 (2.48) -3583 (-9.44) -1472 (-0.96) (omitted) Constant 14783 (4.51) 13208 (10.14) 8494 (6.68) 7930 (2.38) Observations 6 6 6 6 Adj. 𝐑𝟐 0.9554 0.9709 0.8902 0.8404
Table 4 The UK: Brain Drain equations for assessing the validity of the models
Variables (9) (10) (11) Wage -2355 (-0.32) -2548 (-0.33) -6770 (-2.26) Corruption 1494 (0.22) (omitted) (omitted) RD exp (omitted) 1251 (0.08) (omitted) Unemp (omitted) 11628 (2.61) Constant 12721 (1.17) 14723 (1.81) 26081 (5.01) Observ 4 4 4 Adj. 𝐑𝟐 -1.4843 -1.5918 0.6666
Looking at the adjusted R!, one would be tempted to choose model (3) for Germany. However, by leaving all of the variables in the model, though less significant, the regression is closer to reality. Therefore, for Germany model (1) is used. In the case of UK, the variables chosen for the regression are indeed those used in model (11): wage and unemployment. This choice of variables comes from the problem of high dimensionality mentioned above. Now, for each of the two models, the t-values, confidence intervals and assumptions of the models will be checked.
BD!"= α + βwage!"+ γcorruption!"+ δRD!"+ εunemp!"+ u!" The final model (11) used for the UK is:
BD!"= α + βwage!"+ εunemp!"+ u!"
Starting off with model (1), we observe a t-value of -3.89 for the wage variable. Therefore, there is not enough statistical evidence to infer that β!"#$%& is negative at any significance level (p-value=0.160). Moreover, we find more evidence of this by looking at the 95% confidence interval, which has extremes both on the positive and negative side. To move on, we observe that the t-value for the RD variable is 1.98.There is enough statistical evidence that β!"#$ is positive (p-value=0.298). Also, the confidence interval ranges from a negative to a positive value, so there is ambiguity about the sign of this variable. Regarding the
unemployment variable, its t-value is -2.48 yielding no secure sign for β!"#$%&' (p-value=0.244).
Also in this instance the 95% confidence interval ranges from a negative to a positive value, confirming the ambiguity of the sign. To conclude, the constant term has a t-value of 4.51. Thus, there is enough statistical evidence to infer that β!"#$%&' is positive with a 15% significance level (p-value=0.139). The confidence interval is however ranging from a negative to a positive value.
Let’s now take a closer look at model (11) for the UK. We apply the same analysis in order to find out the significance of the single variables. The wage variable has a t-value of -2.26, which brings us to conclude that there is not enough statistical evidence at a 10% level for the β!"#$%& is different from zero (p-value=0.265). Also, the confidence interval ranges from a negative to a positive value, which yields an ambiguous sign for the variable in consideration. To move on, we consider the t-value of the unemployment variable, which is 2.61. This also doesn’t allow us to state with evidence that β!"#$%&' is different from zero (p-value=0.233). In this case as well the confidence interval ranges from a negative to a positive value. To conclude, the constant term has a t-value of 5.01, thus there is enough evidence to infer that β!"#$%&' is positive with a 5% significance level (p-value=0.125). Once again, the confidence interval is between a negative and a positive value, thus yielding ambiguity.
Last but not least, it is checked whether the overall brain drain towards these two destination countries is positively influenced by the single wage differential of the countries. A new variable is created BD = BD!"+ BD!" and it is regressed with the wages variables of
Germany and the UK, as the formula hereunder indicates: BD = β!wage!"+ β!wage!"
The model yields an Adj. R! of 0.3534, so we can already see that it doesn’t hold. The t-value for the variable wage Germany is -2.08. Thus, there is enough evidence at a 15%
significance level (p-value=0.13) that β!"#$%& is negative. However, the confidence interval ranges from a negative to a positive value, yielding ambiguity. With regard to the wage variable of the UK, the t-value is 0.84; thus, there is not enough statistical evidence to infer that β!"#$%& is positive (p-value=0.46). In addition, the confidence interval is between a negative and a positive value. The t-value of the constant term is 3.65, so there is enough statistical evidence at 5% significance level that β!"#$% is positive (p-value=0.036). The confidence interval is all above zero.
5. Conclusion To sum up what was found in the results section, we draw our conclusions first
about the UK regression, then about Germany and, at last, a general conclusion is provided, which is overall aimed at answering the research question.
In the UK regression, no significant results are found, given the high p-values. However, if more data were available and if the results were significant, we could conclude that the brain drain to the UK is negatively correlated with the wage differential. So that if the wage variable goes up by one unit, the brain drain variable goes down by 6770 units, all other variables kept constant. In the Germany regression, results are also not significant. Also in this case, if more data were available, Italian brain drain would be negatively correlated with the higher wages differential. So that, if the wage variable increases by one unit, the brain drain to Germany decreases by 3874 units, all other variables kept constant.
To sum up, it seams like there is no positive correlation between the Italian brain drain and the relatively higher wages abroad. On the contrary, from the results of the regressions run in this research, there seam to be a negative correlation between Italian high skilled workers’ migration and the wage differentials. Thus, my hypothesis has been proved wrong. It appears that the higher the wage differential, the lesser Italian high skilled workers are willing to move abroad. The reason for this negative correlation could be, for example, that the level of specialization required to work in German and British companies is relatively higher than that required in Italian ones, ceteris paribus, so that what is referred to “high skilled” in Italy is not competitive for what it is defined “high skilled” in Germany. Another possibility is that
unemployment for the UK and corruption, unemployment and RD for Germany are not enough control variables to ensure the validity of the regression. Maybe, adding other control variables, such as family relationships abroad, quality of public services, pension levels, taxes on earnings, would even yield a positive beta for the wage variable.
However, a larger n (number of observations) is required to find more significant results. This is the biggest limitation of this research. The fact that so little observations are available doesn’t allow us to use a bigger amount of variables, because of high dimensionality. At the same time, not including more control variables, as for example those mentioned above, could raise the problem of omitted variable bias. Given the novelty of the phenomenon of high skilled migration, only recently the main statistical institutions have started to be interested in this phenomenon, conducting surveys and creating databases. For further research, it would be interesting to conduct surveys to fill in the observations from the years previous to the available ones. For example, one survey could be conducted amongst Italians, who graduated in
Germany and the UK from year 2007 backwards. Also, a survey about earnings of graduates in Italy and abroad in the past would be a relevant source of information, which would allow the research to be more complete. To conclude, for further research, it is left to establish, whether other control variables can be added that provoke the brain drain of Italian high skilled workers to be positively correlated with wages differentials.
Fig. 1 KRUGMAN ET AL. (2012), International Economics Theory and Policy, Pearson Ninth
Edition, Fig. 4.13
Fig. 2 Proportion of Italians (30-34) emigrating to EU countries. EUROSTAT
Table 1 Annual wages in Italy of people employed in middle management positions
Time 2005 2006 2007 2008 2009 2010 2011
Private Sector €22574 €23178 €23704 €24463 €25086 €25699 €26200
Table 2 Number of Italians migrating to the listed destination countries
GEO/TIME 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 TOTAL EU
TOT 7.917 9.068 9.978 9.845 11.077 10.755 12.985 12.788 11.748 12.478 EU 3.636 4.240 5.773 5.629 6.233 6.551 7.947 7.834 6.817 7.014 Belgium 220 260 259 246 278 223 254 238 238 301 Germany 1.434 1.863 2.047 1.794 1.731 1.203 1.228 1.143 1.060 1.226 Spain 173 192 509 586 739 763 1.001 921 877 794 France 459 588 803 795 736 748 814 771 885 1.003 Luxenburg 57 40 52 49 60 56 96 93 87 107 Netherlands 85 76 136 164 187 258 223 215 197 218 Austria 110 116 168 135 155 183 205 184 218 220 Romania 155 147 110 151 191 596 1.192 1.428 959 938 UK 535 545 1.193 1.157 1.360 1.575 1.805 1.634 1.442 1.271 Switzerland 916 1.156 1.003 880 965 796 1.042 958 868 1.008
Fig. 3 Unemployment rates of people (aged 30-34) with a bachelor and master degree
Bibliography
ARIU, A. & SQUICCIARINI, P. (2013) The Balance of Brains: Corruption and High Skilled Migration, Institute de Recherches Economiques et Sociales de l’Université catholique
de Louvain,Discussion paper.
BARONE, C. (2012) Le trappole della meritocrazia, Università di Trento, Il mulino.
BRUNELLO, G., FORT, M., WEBER, G. (2009) Changes in compulsory schooling, education and the distribution of wages in Europe, The Economic Journal, vol. 119, pp. 516–539.
BUDRIA, S. & MORO-EGIDO, A. (2008) Education, educational mismatch, and wage inequality: Evidence for Spain, Economics of Education Review, vol. 27, pp. 332–341.
GOOS, M., MANNING, A., SALOMONS, A. (2009) Job Polarization in Europe, American
Economic Review, vol. 99, no. 2, pp. 58–63.
KOSER, K. & SALT, J. (1997) The geography of highly skilled international migration,
International Journal of Population Geography, vol. 3, pp. 285-303.
MORANO-FOADI, S. (2006) Key issues and causes of the Italian brain drain, Innovation, The
European Journal of Social Science Research, vol. 19, no. 2, pp. 209-223.
SALT, J. (1992) Migration processes among the highly skilled in Europe, International
Migration Review, vol. 26, pp. 484–505.
YANG, D. (2003) Financing constraints, economic shocks, and international labor migration: understanding the departure and return of Philippine overseas workers. Dissertation chapter, Harvard University.
