1
The influence of an economic crisis on labour migration:
The influence of an economic crisis on labour migration: is there a brain drain in
Portugal, Ireland, Iceland, Italy, Spain and Greece due to the recent economic
meltdown?
Stefan Klink 10560572
BSc Economics and Business Specialisation: Economics and Finance June 29, 2016
2 Statement of originality
This document is written by Klink, Stefan who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3 Acknowledgements
I would like to thank my thesis supervisor dr. D. F. Damsma for his guidance on writing this paper and providing the necessary feedback and comments.
4 Contents
1. Introduction ... 6
2. Literature review ... 8
2.1 Theoretical Push & Pull factors ... 8
2.3 Conclusion ... 15 3. Methodology ... 17 3.1 Data specification ... 17 3.2 The models ... 18 3.3 Hypothesis ... 20 3.4 Conclusion ... 21 4. Results ... 21
4.1 Results GDP per capita growth ... 21
4.2 Unemployment rate and GDP per capita growth ... 23
5. Discussion and limitations ... 24
6. Conclusion ... 25
7. Bibliography ... 26
5 Abstract
This paper analyses if there occurred a ‘brain drain’ in the countries Portugal, Ireland,
Iceland, Italy, Spain and Greece during the recent economic crisis. Despite the fact that these countries are developed OECD countries, the economic crisis had severe consequences for push and pull factors of migration. Existing literature emphasizes factors such as
unemployment and mean income as determinants of international migration. Additionally, to these theoretical literature, past empirical research has been conducted on developing countries but not in a time period during a crisis. By using a ‘before and after’ comparison with unemployment rate of tertiary educated people and GDP per capita growth as dependant variables, it is possible to compare the value of the emigration rate of highly skilled workers before and during the crisis. Whilst the theory suggest that there will be an increase in the emigration rate of highly skilled workers, no such results have been found with empirical testing.
Keywords: brain drain, highly skilled worker, international migration, developed countries,
6 1. Introduction
In recent years the euro-area suffered the worst economic crisis since 1929. Due to an economic crisis unemployment rates rose significantly, especially in southern-European countries (Jauer, Liebig, Martin, & Puhani, 2014). Most of these southern-European
countries have a negative net migration rate, however the migration rates of young people rose as well in the countries. Due to this flight of human capital and of young highly skilled workers with potential the economies of these countries are being hollowed out (Paul, 2015). This triggered a negative spiral, which diminishes the productivity of growth. Beine, Docquier, & Rapoport (2001) point out that education is a major determinant of long-term growth, and that common wisdom suggests that migration of people endowed with a high level of human capital is detrimental for the country of emigration. An example of this is Spain, which is one of the countries that produces above average graduates in comparison to other EU countries. This is very costly for the Spanish government, as the average cost to train a young professional equates to around 60.000 euro (The Guardian, 2012).
This flight of highly skilled people is characterized as a brain drain. This research paper investigates if a so called ‘’brain-drain’’ has occurred in certain European countries, namely: Portugal, Ireland, Iceland, Italy, Greece and Spain. To specify this phenomenon, brain drain is the migration of educated workers from their native country to another country. These people seek out for better opportunities of their professional career or a better standard of living in a foreign country. This situation arises mostly in lower developed countries where certain pull and push factors drive these educated workers away from their native country to more developed countries. Despite the fact that the reference countries used in this paper are developed countries they are likely to have been prone to brain drain. As a result of the recent crisis the south-European countries were severely hit and this affected push and pull factors of migration, such as unemployment and the cost of living in the host country compared with the cost of living in the native country. People outweigh these benefits and costs and choose thereby which destination yields the highest maximum utility. For that reason a brain drain may have occurred in Europe due to the recent
economic meltdown.
With the guidance of past literature and empirical testing the following central
question will be examined: The influence of an economic crisis on labour migration: is there a
7 meltdown?
The most related former research that has been conducted on brain drain is mostly theoretical and there has nearly not been any empirical testing. This research paper will be partly descriptive and partly empirical. This paper hypothesizes the effect of the growth in gross domestic product per capita and the unemployment rate of tertiary educated workers on the migration rate of highly skilled workers. This research will be done with data retrieved from OECD database and from the World data bank. Subsequently, a ‘before and after comparison’ with panel data will be performed on 6 countries for two reference years (2005 & 2010). At first this paper will review the existing literature on international migration and brain drain. Then, there is a brief summary with statistics of push factors of brain drain. Next, the way of research will be comprehensively explained in the methodology chapter.
Afterwards, the results of the models will be provided. Furthermore, these outcomes will be examined and limitations and recommendations will be provided. Lastly, the final chapter will consist of the interpreted results which will be used to define a possible resolution for the central question.
8 2. Literature review
In the following chapter the determinants of international migration will be elaborated with use of past literature and present world figures of these determinants of migration. Firstly, theory will identify push and pull factors of emigration, further there will be theory
explained about brain drain and some empirical research will be reviewed. Secondly, the figures of these push and pull factors of the reference countries will be reviewed. Lastly, there will be a concluding subchapter building up to the methodology chapter and summarizing the theory and figures.
2.1 Theoretical Push & Pull factors
The impact of the past economic meltdown has been immense, and as expected there has been a lot of research conducted and documented about the causes, consequences and the policy responses of the countries of this meltdown (Young, 2014). However, the influence of a crisis on brain drain is scarcely recorded, especially for developed countries. Yet, we can take guidance from existing theory about international migration and brain drain.
Nearly all recent neo-classical theories of determinants of international migration are based on the framework of Larry Sjaastad (Bodvarsson & Van den Berg, 2009). According to the framework of Sjaastad (1962) the individual agent outweighs his cost and benefits of migration and calculates the value of the opportunity of migration, eventually he will choose the destination with the maximum utility. This relocation is seen by Sjaastad (1962) as an investment of the individual in human capital, where the distance travelled to the host country is the proxy for migration costs. He validates this argument by pointing out that the greater the distance travelled the greater the monetary costs of migration, such as
transportation, living costs and the interruption between jobs. While he realises the returns of migration in the form of labour income, he disregards nonmonetary benefits such as better climate, recreational opportunities and a more desirable environment because these differences are accounted for in the living costs (Bodvarsson & Van den Berg, 2009).
Building on the framework of Sjaastad, Borjas (1990) developed a model that people not only outweigh their maximum utility based on wage and costs to leave, but on several additional aspects. Each individual agent has his own set of skills, talents and perceptions to the labour market. According to Borjas (1990) the size of the migrant flow depends on more factors than only the factors stated by Sjaastad. The decision to migrate of an individual
9
depends not only on the average of differences in wages across countries, but where the immigrant would fit better in the foreign labour market and how well his abilities would be applied there (Borjas, 1990). Borjas states the following hypotheses: the emigration rate is higher the greater the mean income in the host country, the emigration rate is lower the greater the mean income in the source country and the emigration rate is lower the greater the level of migration costs. These three hypotheses have been widely tested and form their theoretical core from the framework of Sjaastad. Besides these hypotheses he developed the following hypothesis: the emigration rate is higher the greater the pay-off to the
observed demographic variables in the host country relative to payoff of the source country. This explains his addition to the model of Sjaastad. More generally, the greater an individual fits in the foreign labour market based on his abilities, the higher his utility would be if he would emigrate to the foreign country. He stipulates in particular that this prediction follows from: individuals with a certain level of schooling are more likely to migrate, the higher the rate of return to schooling in the host country relative to the rate of return to schooling in the source country. He concludes succinctly that educated people migrate to the country that values educated labour the most. Therefore, international labour flows are no different to the flows of international goods. Workers, like goods, flow to the country that is willing to pay the most for them (Borjas, 1990).
On the other hand, Kwok & Leland (1982) suggest another cause for the flight of foreign-trained students. Kwok & Leland refer to the definition of brain drain: ’’skilled professionals who leave their native land in order to seek more promising opportunities elsewhere’’. In their opinion the skilled students do not return to their native country after studying abroad due to an asymmetric information problem. Kwok & Leland derive their assumption that this asymmetric information problem could lead to serious market failures from George Akerlof (1970) and John Riley (1975). According to Akerlof there are many markets in which buyers use some market statistics to judge the quality of prospective purchases. In our case the worker in a host country tries to distinguish him or herself in the labour market by following an education. However, employers of the native country cannot assess the exact quality of the foreign education and therefore employers of the native country cannot offer an exactly matching wage for an individual who studies in a foreign country. On the contrary, the employers of the host country could assess the value of the education and offer a more tailor-made wage for the worker. Akerlof gives a more concrete
10
example for the labour market further in his paper. He imposes that good quality schooling could serve as a good statistic to indicate the quality of the ability of the worker, but when asymmetric information occurs about this indicator this leads to disadvantages for the worker in the minority group, in this case the workers who are educated in a foreign country. Generally: ‘’ This lack may be particularly disadvantageous to members of already disadvantaged minority groups. For an employer may make a rational decision not to hire any members of these groups in responsible positions, because it is difficult to distinguish those with good job qualifications from those with bad qualifications’’. To be more specific in our case, employers in the host country of the students have a more accurate judgement of the skills and level of productivity of the student than an employer in the native country would have. Thus, the employers of the host country can analyse a student’s attributes more accurately with respect to required job qualifications. This is a direct result of the familiarity of the host country’s employers with the academic system with which the student follows, and their past experience with hiring graduates from universities in that same host country. Therefore, the employers in the native country only know the average productivity of returning students, thus an employer in the native country would only to be willing to offer an average wage. Whereas, the employers of the host country could assess the true
productivity of an individual and offer a more tailor-made wage.
Moreover, Kwok & Leland (1982) concluded in their paper that even when students prefer returning home (at equal salaries) and employment opportunities exist at the native country, the students stayed in their host country. This is related to the information the employers have at the time of hiring, which allows them to determine precisely the productivity and offer a tailored wage, while native employers have to assume average productivity for a returning worker, resulting in a different wage. In their own research Kwok & Leland take Taiwan as the reference country. After their research they conclude that despite the fact that the native country, Taiwan, offers the same wage as the host country, students remain at the host country. It seems plausible that this is not only because of the personal preference of a student who prefers living abroad, but also due to the asymmetric information problem. Their paper shows that the brain drain may exist even when students have a preference for returning home (at equal salaries) and employment opportunities exist at comparable average pay.
11
less developed countries (LDC’s) rather than developed countries. Armed with ambitious economic development plans many LDC’s are short of professional and technical personnel (Kwok & Leland, 1982). However, in this research paper, developed European OECD
countries (Iceland, Ireland, and PIIGS) will be analysed to show if the recent economic recession resulted in a stronger outflow of skilled migrants. According to the
aforementioned migration theory, an individual will seek maximum utility based on the benefits and costs of migration. While some factors play an important role in those benefits and costs, like the mean income of the host and native country and migration costs,
asymmetric information too plays a role in determining whether a person stays in the host country.
Beets & Willekens (2009) analysed the impact of the recent economic crisis on international migration. They based their research on international migration theory, on data of global economic crises in the past and their influence on international migration and published expert opinions. They point out that the impact of an economic recession would generate ‘conjuncture shocks’ on migration, but in a more developed country these shocks would be absorbed by social security and social capital cushions. They mention in particular Spain, reporting that the unemployment rate rose to 18 percent and the unemployment rate of youths went up to 36 per cent in a time frame of two years (2007-2009) (The World Bank, 2016).
However, when Beets & Willekens (2009) analyse further in their research paper they state that it is unclear yet whether migrants becoming jobless stay or go home and whether migrants in service jobs are laid off or have their wages reduced. The main assumption is thus that migrants are unlikely to return home in large numbers. ‘’But highly skilled economic migrants are often young and single, and may more easily stay because of their ability to quickly find another job as several have good language skills’’ (Beets & Willekens, 2009).
In a study from Docquier, Lohest, & Marfouk (2007) the focus was on brain drain of only developing countries. They report that developed countries like Germany, Canada and the United Kingdom worried about losing talented skilled workers; since past literature stressed that the consequences would be detrimental for the economic development of their country. The article presents new estimates of the brain drain experienced by developing countries based on a new self-developed dataset.This dataset consists of
12
emigration rates of tertiary educated people in the period of 1990-2000 of developing and OECD countries in the world. Their main findings were, on the whole, that the brain drain is stronger in countries that are not too distant from OECD countries and where the average level of schooling of natives is low. These results coincide with the before mentioned literature about brain drain and determinants to migrate. The brain drain was more pronounced in developing countries where a lower level of education prevails. Leaving to developed OECD countries would improve the standard of living for these people due to better education and a higher initial level of GDP per capita. Besides these push factors these findings of Docquier, Lohest, & Marfouk (2007) coincide with the earliest theories of migration developed by Sjaastad (1962) because Sjaastad stated that people outweigh the benefits and costs of migration and they choose the destination with the highest maximum utility. As a proxy of costs he used the distance to travel, and if we examine the origin of people’s native country in the paper of Dohier, Lohest & Marfouk we experience that the brain drain occurred more in countries close by OECD countries. This implies that the costs of migrating are lower for people living closer to OECD countries and therefore are more determined to leave their native country.
2.2 Push & Pull factors in practice
Nonetheless, due to the recent economic meltdown several factors of international migration have been influenced. In particular GDP and unemployment rates were affected during the crisis. Where the crisis resulted in severe consequence in southern –Europe countries, not all countries were equally affected. Whereas the unemployment rate rose in Greece and Spain by 18 percentage points over the same period, it actually declined in Germany, by more than 2 percentage points. At the same time, there is a significant amount of free mobility in Europe whose scale has risen sharply following the EU enlargements in 2004 and 2007. According to the OECD standardised migration statistics, which cover only permanent migration movements across borders within OECD Europe, in 2011
intra-European free circulation was four times more common than migration from outside of the free mobility zone, exceeding more than 900 000 migrants (Jauer, Liebig, Martin, & Puhani, 2014).
In the past literature there has not been any research conducted on a possible brain drain in a developing country after an economic crisis, especially not on developed countries
13
after the recent economic meltdown. In this research paper there will be examined if the latter economic crisis had a significant influence on the flight of human capital in comparison to before the crisis. With data retrieved from the OECD database (Dumont & Hovy (2013)) a statistical ‘before and after’ approach will be used to research if the push factors GDP per capita and unemployment rate had a significant influence in the outflow of highly educated people between the years 2005 and 2010.
Before this statistical approach we scrutinize firstly the figures of the unemployment, GDP growth and total emigration of our reference countries (PIIGS, Iceland) over the period 2006-2014. We retrieve the following graphs (The World Bank, 2016):
Graph 1: GDP per capita annual % change. (The World Bank, 2016).
Graph 1 shows that the economic meltdown caused a decrease in the gross domestic product per capita in all the reference countries. After 2007 every country enters a
downward spiral which leads to annual percentage decline. Mostly all countries experience their most substantial declines in 2009. Except for Greece which suffers a loss of nearly 9 percent, while it still in 2012 it produces a loss of 6.79 percent relative to 2011. None of the countries produces a stable growth in the years 2011 and 2012; this indicates that the countries still suffer from the economic crisis. While in other EU countries like Germany and
-10 -8 -6 -4 -2 0 2 4 6 8 2005 2006 2007 2008 2009 2010 2011 2012 an n u al % ∆ gr o wt h Years
GDP per capita
ITA GRC PRT ESP ISL IRL14
UK even though the GDP per capita declined as well in 2009, both countries produced a stable growth of around 1-2 percent in the year 2012, becoming more attractive for potential high-skilled migrants (The World Bank, 2016). Thus, this situation hints that the standard of living in the reference countries relatively to Germany and UK diminished during the period of the crisis and according to the theory this push factor would affect the decision to migrate.
Graph 2: Unemployment rates, total in % of total labour force. (The World Bank, 2016).
Graph 2 shows the unemployment rates of the reference countries throughout the years 2005-2012. All reference countries experience a higher rate of unemployment at the end of the period in comparison with 2005, two years before the beginning of the crisis. In particular in Spain and Greece the unemployment rates rose significantly to nearly 25 per cent in the year 2012. These ascending unemployment rates cause uncertainty for potential job seekers. It reduces the chance to obtain a job in the future and pushes the labour force out of the country.
0,00 5,00 10,00 15,00 20,00 25,00 30,00 2005 2006 2007 2008 2009 2010 2011 2012 Une m p lo ym e n t % Years
Unemployment
ITA GRC PRT ESP ISL IRL15
Graph 3: Emigration of total population. Eurostat (2016)
The third graph shows the emigration by absolute number of emigrants. To be clear these numbers are the total population of emigrants so not of highly educated emigrants. This gives an indication if emigration changed in a developed country due to influences of the economic meltdown. Notably, Spain experiences an increase of emigration throughout the period (2006-2012) the absolute number of emigrants nearly tripled in size in 2012 relative to 2006.
2.3 Conclusion
Concluding, the discussed literature and figures about GDP growth per capita,
unemployment and the total emigration lead us to that several pull and push factors of migration were active during the recent economic crisis. The past literature expresses that people choose to migrate based on an examination of benefits and costs of migration. Furthermore, Kwok and Leland elaborate that due to an asymmetric information problem people who enjoyed their education in a foreign country are most likely to stay there after graduation. Borjas build his theories on the framework of Sjaastad, in addition to the model of Sjaastad, Borjas included that migrants not only outweigh their benefits and costs in terms of wage income and costs to migrate, but he added another variable: the applicability of the migrant’s skills in the labour market. These papers are all theoretical approaches to the brain drain problem, in a more recent empirical research conducted by Docquier, Lohest, & Marfouk (2007) there has been tested wheter brain drain occurred in developing
countries. Their findings coincide with the literature of determinants of brain drain.
0 100000 200000 300000 400000 500000 600000 2006 2007 2008 2009 2010 2011 2012 2013 2014 N u m b e r o f e m ig ran ts Years
Emigration
IRE GRC ESP ITA PRT ISL16
According to Docquier, Lohest and Marfouk individuals who live in a developing country with a lower level of education and live close by an OECD country are most likely to migrate. These results emphasize that brain drain would be more present in less developed countries than in developed OECD countries. However, as stated in chapter 2.2 the chosen reference countries suffered relatively more from the recent econmic crisis than the north-european countries (Germany and UK). The unemployment rates rose significantly during the crisis in the south-european countries, while the GDP per capita growth was negative and unstable in the same period. Also, the easy possibility to migrate due to the free borders in the European Union contributes to a potential flight of highly skilled human capital from the reference countries to the north-european countries. Standing on the theories build by Sjaastad and Borjas regarding determinants of international migration this paper expects that the flight of highly educated people will be significantly more during the recent crisis in the reference countries since push and pull factros are affected due to the recent meltdown. Eventually leading to a possible brain drain in the reference countries despite the fact these are not developing countries. This research will be conducted with guidance of empirical research.
17 3. Methodology
In this chapter will be discussed which research method will be conducted to investigate the influences of unemployment rates and GDP growth per capita on the emigration rate of highly educated workers. Firstly, there will be a specification of the dataset and how it is originated. Secondly, there will be a further explanation of the used statistical models. Lastly, the hypothesis will be formulated.
3.1 Data specification
The dataset of the migration figures is retrieved from a joint contribution of the OECD and the United Nations, who collaborated for a high-level dialogue on migration and
development. They wrote a report of the world migration in figures. They constructed a database of immigrants in the OECD countries and it is based on a database of educational attainment of Barro & Jong (2013) and on population censuses and registers around 2005/2006 and 2010/11 in OECD member countries (Dumont & Hovy, 2013). The data covers the population of age 15 and over. Migrants are defined on the basis of their country of birth.This data collection made it possible to calculate emigration rates by skill level. These OECD rapports are published every five years, with reference years 2000/2001, 2005/2006 and 2010/2011 causing less available observations.
Unfortunately, the datasets from the OECD is the only dataset available where migrants are characterized by skill and education. However, the reference years 2005/2006 and 2010/2011 suit fine with the research since it contains both a year before the crisis and a year during the crisis.
To specify, the migration rate of skilled workers is the percentage of skilled workers from the total population of emigrants. In our case the skilled worker will be defined as approached by the OECD database, a worker who followed a tertiary education. Furthermore, the data of the annual difference of gross domestic product per capita is originated from the world data bank (The World Bank, 2016).
To offer a more well-rounded and holistic view of the emigration of educated skilled workers it would have been more appropriate to use more reference years in the period of 2005-2011. As aforementioned there is no such database where the emigrants per country are sorted by skill and education, leading this to an unbalanced dataset with only two
18
reference years.But, Stock and Watson (2015) stipulate that despite the fact data is unbalanced all the methods of panel data can be used with an unbalanced panel.
3.2 The models
Because the research is on several countries and with the use of two reference years the data consists of observations on the same n countries at two time periods. Further there are cross-sectional relations within the data, this results that our dataset is a panel dataset. Hence a panel dataset is created to look how the regressors; annual difference in GDP per capita and unemployment rates of tertiary educated workers influences the emigration rates.
According to Stock & Watson (2015) the best suitable regression method would be a ‘before and after’ comparison. Stock and Watson state that with this difference comparison we could eliminate potential omitted variable bias, if the factors excluded from the model remain constant over time in a given country. Let 𝑍𝑖be such a variable that determines the emigration rate in the 𝑖𝑡ℎ country but is constant over time. This could be for example the cultural attitude of the country towards migration. Because the dataset is a panel dataset, it is possible to hold these factors constant even though we cannot measure them. This is done with an OLS regression with fixed effects. In this case the data from the countries are
obtained for T=2 time periods, it is possible to compare values of the dependant variable in the second period to values in the first period. By focusing on the changes in the dependent variable, this ‘before and after’ comparison in effect holds constant the unobserved factors that differ from one country to another but do not change over time within the country.
Accordingly, the linear regression relating the annual difference in GDP per capita and 𝑍𝑖 to the emigration rate is for the two reference years:
𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2005= 𝛽0+ 𝛽1𝐴𝑛𝑛𝑢𝑎𝑙 (∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2005+ 𝛽2𝑍2005+ 𝑢2005 (1)
𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2010= 𝛽0+ 𝛽1𝐴𝑛𝑛𝑢𝑎𝑙 (∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2010+ 𝛽2𝑍2010+ 𝑢2010 (2)
Because 𝑍𝑖 is constant over time it will not produce any changes in the emigration rate
19
analysing the change between the periods rather than look at the both individually. Subtracting equation (1) from (2) eliminates the effect of𝑍𝑖, resulting:
𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2010− 𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2005 =
𝛽1{(∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2010− (∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2005} + 𝑢2010 − 𝑢2005 (3)
By eliminating 𝑍𝑖 we state that the emigration rate was not influenced by any constant
factors that not vary over time. As stated in the literature review chapter unemployment rates are push factors for migrants and affect the decision to leave. Besides, the
unemployment rate is a variable that does change over time and so will influence the model when it would be omitted. This raises a new model where a ‘before and after’ comparison will be executed but now included with an extra independent variable, the unemployment rate of tertiary educated people.
𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2010− 𝐸𝑚𝑖𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑒2005 = 𝛽1𝛿 + 𝛽2𝛾 + 𝑢2010 − 𝑢2005 (4)
Where 𝛿𝑖 = (∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2010− (∆%)𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎2005
and 𝛾𝑖 = 𝑈𝑛𝑒𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡 𝑟𝑎𝑡𝑒 𝑇𝑒𝑟𝑡𝑖𝑎𝑟𝑦2010− 𝑈𝑛𝑒𝑚𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡 𝑟𝑎𝑡𝑒 𝑇𝑒𝑟𝑡𝑖𝑎𝑟𝑦2005
Other changes in the emigration rate over time who are not captured by the GDP per capita arise from other sources and are captured by the error term 𝑢2010 𝑎𝑛𝑑 𝑢2005, for 2010 and
2005 respectively. By specifying the equation (3) in differences in Y and X the model has the effect of controlling for omitted variable bias. Finally, a constant is added to the model. As Stock and Watson (2015) state that the inclusion of an intercept with a ‘before and after’ analysis provides the same estimate of the slope coefficient as the OLS regression when using entity and time effects (Panel regression of T>2).Furthermore, Stock & Watson (2015) indicate that a regression using panel data with fixed effects is conditional to the following four assumptions:
1. εi,t has a conditional mean zero: E(εi,t|Xi,t, αi, λt) = 0,
2. (Xi,t, εi,t) are independently and identically distributed draws from their joint
distribution.
3. Large outliers are unlikely: (Xi,t, εi,t) have nonzero finite fourth moments.
20
These conditions are met, since there is no case of large outliers (see appendix for
descriptive statistics). Additionally, in the fourth model could multicollinearity occur due to the inclusion of an extra variable. However, the correlation of the variables of GDP per capita growth and unemployment rate for tertiary educated workers is (-0.297), this figure
confirms for no multicollinearity (Stock & Watson, 2015).
3.3 Hypothesis
With this ‘before and after’ comparison we scrutinize the difference of emigration rates of highly educated workers in the years 2005 and 2010, and to see if there has been a negative relationship between the annual change in GDP per capita and emigration rate of highly educated workers and if there is a significant difference between the reference years. Likewise, we analyse if there is a positive relationship between the unemployment rates of tertiary educated and the emigration rate of highly educated worker. This is done by a simple t-test with a five per cent significance level. According to the theory the growth of GDP per capita has a negative relationship with the emigration rate of highly skilled workers, this leads to the following hypotheses:
𝐻0: 𝛽1 = 0
𝐻1: 𝛽1 < 0
Furthermore, the literature suggests that the unemployment rate of tertiary educated have a positive relationship with the emigration rate of highly skilled workers.
𝐻0: 𝛽2 = 0
𝐻1: 𝛽2 > 0
Where 𝛽1 is the coefficient of the annual change in GDP per capita and 𝛽2 is the coefficient
of the unemployment rate of tertiary educated people. Generally, there will be a statistically significant negative relation between the annual percentage change per capita and the emigration rates. For the second hypothesis, there will be a positive relationship between the emigration rates of highly educated people and the unemployment rate of tertiary educated people.
21 3.4 Conclusion
To summarize, this paper will execute a ‘before and after’ comparison as statistical approach with a data set retrieved from the OECD database in order to answer the research question, is there a brain drain in the PIIGS and Iceland due to the recent economic meltdown? Two key hypotheses were formed; first the GDP per capita growth has a negative effect on the emigration rate of highly skilled worker. The second states that the unemployment rate of tertiary educated people has a positive effect on the emigration rate of highly skilled workers. Due to the ‘before and after’ comparison the difference will be measured from before and during the crisis, and will be examined if there was a significant increase of migration of highly skilled workers difference during the crisis in the reference countries.
4. Results
In the following chapter the results of the statistical regression will be represented.
Subsequently, in the next chapter there will be some critique and recommendations about the descriptive research adduced.
4.1 Results GDP per capita growth
Table 1: Results model 1, 2 and 3
Dependant variable Emigration rate High educated ‘Before after’ comparison. T=2 ∆2010−2005 Eq.(3) Regression T=2010 Eq.(2) Regression T=2005 Eq.(1) Annual %∆ GDP per capita 0.8904891 (1.52) -0.7298559 (-0.42) 7.70092 (9.01)*** Constant 0.0188971 (0.72) 0.2396173 (4.84)*** 0.1054171 (4.81)*** R2 0.3662 0.0421 0.9531 F-statistic 2.31 0.18 81.25*** N 6 6 6
Note: The t-statistics are represented between the parentheses. The abbreviation of: ***significant at 1% level, ** at 5 % and *at 10% level. 𝑅2 Is provided to show what percentage of variance can be explained by this model.
22
Table 1 represents the results of the regressions on the dependant variable emigration rate of high educated people. On the regressions of one single time period (T=2010 and T=2005) an OLS regression was performed, they show opposite results. The coefficient of the GDP is negative and not significant in 2010, while in 2005 the coefficient is large and significant at the one per cent level. The table shows that with the ‘before after’ comparison the independent variable GDP per capita growth is not significant, not even at a 10 per cent level. Hence, with this information the null hypothesis could not be rejected. There is not enough information to state that the difference in GDP growth causes a significant difference in emigration of highly educated people.
Moreover, the coefficient regarding the independent variable is positive while as stated in the theory the relation between GDP per capita growth and the migration rate should be negative. Besides, the F-statistic is 2.31 and therefore as well not significant at a ten per cent level. This means that the model is not fit for estimating the data. The
statistical results stated above do not provide an acceptation of the null hypothesis. Due to the insignificants results it is not possible to state if the difference of the migration rate influenced by the GDP per capita growth has been significantly higher in 2010 than 2005. Figure 1: Scatterplot of the change in emigration rate and the change in capital growth between 2005 and 2010 for 6 countries. -. 0 8 -. 0 6 -. 0 4 -. 0 2 0 .0 2 .0 4 D if f_ H ig h e d u c -.08 -.07 -.06 -.05 -.04 -.03 -.02 -.01 0 .01 diff_GDPcapitagrowth
Fitted values diff_Higheduc
23 4.2 Unemployment rate and GDP per capita growth
Table 2: Results regression model (4)
Dependant variable Emigration rate High educated ‘Before after’ comparison. T=2 ∆2010−2005 Eq.(4) 𝛿𝑖 0.8966284 (1.32) 𝛾𝑖 0.0935321 (0.12) Constant 0.0248261 (0.74) R2 0.3691 F-statistic 0.88 N 6
Note: The t-statistics are represented between the parentheses; 𝑅2 is provided to show what percentage of variance can be explained by this model.
Table 2 represents the results of the statistical regression of the fourth model as stated in the methodology chapter. In this model neither the coefficient of the difference in unemployment rate of tertiary educated people, nor difference in the GDP per capita growth have a significant influence in the difference between emigration rate of highly educated people for the years 2010 and 2005. Hence, with this information we cannot reject the null hypotheses and cannot state that the coefficients are significantly different from zero.
The coefficient regarding the unemployment rate of tertiary educated people is positive, but since the coefficient is not significant, not even at a weak level, it is not possible to draw any conclusions about this coefficient.
24 5. Discussion and limitations
After doing the descriptive research and interpreting the results, some discretion and points of critique are needed. After all, the regression from above resulted in unpredicted
outcomes. The coefficient of GDP growth per capita is positive, determining a positive relation between GDP growth per capita and the migration rates. This indicates that a growth of the GDP per capita leads to higher migration, while the contrary is suggested by the theory. This coefficient could be positive due to fact that maybe the GDP growth per capita in the destination countries of the migrants accelerated significantly more than the reference countries and causing migration. Yet, we cannot state after empirical research that the coefficient of GDP growth per capita is significantly different from zero.
Additionally, the fourth model included an independent variable, the unemployment rate of tertiary educated workers; the results of this regression were as well not acceptable to draw any conclusions about the hypotheses. The inclusion of the extra independent variable did not lead to expected results, as the graph 3 in the second chapter suggest an increase of emigration during the crisis and the unemployment rates rose during the same period.
Moreover, the most substantial point of critique is the lack observations in the empirical research for both models. With only six countries and two reference years the observations are scarce, while the ‘before and after’ comparison is a regression on the difference and therefore bisects the dataset. For the empirical research it would be more suitable if the regression contained more observations. Unfortunately, as mentioned before there is no such data base that contains the emigration rates of highly skilled migrants for every year. As recommendation regarding this limitation it would be more appropriate for the research to make use of more reference years. In that way a panel regression could be executed for a more accurate well-rounded empirical research. Also, as indicated the before and after model eliminates omitted variable bias for variables who are constant over time, point of critique could be the omission of more variables who are not constant over time. However, these variables could only be added to the model if there would be more observations.
25 6. Conclusion
The empirical nature of this research paper accepts that there is no significant evidence for the difference in the emigration rate between 2010 and 2005. However, as stated in the chapter of discussion and limitations the research would be conducted better if there were more reference years possible to insert in the dataset. Based on the results retrieved from the produced regression, we cannot confirm the aforementioned theory. The theory behind international migration and brain drain suggested that the influence of the crisis would affect push factors of migration. As stated in chapter two, the unemployment rates rose and the growth of GDP per capita diminished during the recent economic crisis in the reference countries. Whilst there may be some weight in the literature part, one must take empirical evidence, in the context of a solid framework, in consideration to properly estimate a possible brain drain in developed countries. Therefore, this research found no significant evidence on difference of the emigration rates between 2010 and 2005, further research is required to give a more well-rounded understanding of the emigration of highly educated workers in the chosen reference countries.
26 7. Bibliography
Akerlof, G. A. (1970). The Market for "Lemons": Quality Uncertainty and the Market Mechanism. The Quarterly Journal of Economics, 84, 488-500.
Barro, R. J., & Jong, L. W. (2013). A new data set of educational attainment in the world, 1950–2010. Journal of Development Economics, 184–198.
Beets, G., & Willekens, F. (2009). The global economic crisis and international migration: an uncertain look. Vienna Yearbook of Population Research, 7, 19-37. Retrieved 4 4, 2016
Beine, M., Docquier, F., & Rapoport, H. (2001). Brain drain and economic growth: theory and evidence. Journal of Development Economics, 64, 275-289.
Bodvarsson, O., & Van den Berg, H. (2009). the economics of immigration. New York: Springer. Borjas, G. (1990). Economic theory and international migration. The international migration review,
23, 457-485.
Docquier, F., Lohest, O., & Marfouk, A. (2007). Brain drain in developing countries. The World Bank Economic Review, 21, 193-218.
Dumont, J.-C., & Hovy, B. (2013, October 1). World migration in figures. Retrieved from www.OECD.org: http://www.oecd.org/els/mig/World-Migration-in-Figures.pdf
Jauer, J., Liebig, T., Martin, J. P., & Puhani, P. A. (2014). Migration as an adjustment mechanism in the crisis? A comparison of Europe and United States. IZA discusson paper, 7291.
Kwok, V., & Leland, H. (1982). An Economic model of the Brain Drain. The American Economic Review, 72, 91-100. Retrieved 4 21, 2016
Paul, J.-M. (2015). Spain's brain drain is a eurozone problem. Retrieved from
www.bloombergview.com: http://www.bloombergview.com/articles/2015-09- 28/spain-s- brain-drain- poses-a- threat-to-
Riley, J. G. (1975). "Competitive Signalling". Journal of Economic Theory, 174-186.
Sjaastad, L. A. (1962, October). The cost and returns of human migration. Journal of Political Economy, 80-93. Retrieved April 21, 2016, from http://www.jstor.org/stable/1829105 Stock, J. H., & Watson, M. W. (2015). Introduction to econometrics; updated third edition. Harlow:
27 The Guardian. (2012, 6 1). Brain drain in Spain as 1m graduates swell the ranks of the unemployed.
The Guardian, pp. -. Retrieved from
https://www.theguardian.com/world/2012/jun/01/europa-brain-drain-spain-graduates The World Bank. (2016, 6 3). World Development Indicators. Retrieved from
http://www.worldbank.org/:
http://databank.worldbank.org/data/reports.aspx?source=world-development-indicators&Type=TABLE&preview=on#
Young, B. (2014). Financial crisis: causes, policy responses, future challenges. Brussels: EUROPEAN COMMISSION.
28 8. Appendix
Descriptive statistics
Data used for regressions
Country code Year GDP capita
growth 2005 Emigration rate High educated 2005 Unemployment Tertiary 2005 ESP 2005 0.008123192 0.249 0.232999992 GRC 2005 -0.005932523 0.183 0.113999996 IRL 2005 0.028692193 0.417 0.108000002 ITA 2005 -0.008686881 0.156 0.173999996 PRT 2005 -0.007184787 0.086 0.183999996 ISL 2005 0.031267437 0.44 0.135
Country code Years GDP capita
growth 2010 Emigration rate High educated 2010 Unemployment Tertiary 2010 ESP 2010 -0.023030461 0.287 0.187999992 GRC 2010 -5.49963E-05 0.218 0.116000004 IRL 2010 -0.020043741 0.334 0.103999996 ITA 2010 -0.074581969 0.174 0.22 PRT 2010 -0.004592095 0.099 0.21 ISL 2010 -0.052574192 0.372 0.135 unemployme~e 12 .09425 .0466673 .026 .202 gdpcapitag~h 12 -.0107166 .0300466 -.074582 .0312674 higheducated 12 .25125 .1195074 .086 .44 unemployme~t 12 .1600833 .0468333 .104 .233 Variable Obs Mean Std. Dev. Min Max . summarize unemploymenttert higheducated gdpcapitagrowth unemploymentrate
gdpcapi~2010 6 -.0291462 .0289231 -.074582 -.000055 gdpcapi~2005 6 .0077131 .0182877 -.0086869 .0312674 diff_GDPgr~h 6 -.0368593 .0363508 -.0838416 .0058775 diff_Unemp~t 6 .0041667 .0307663 -.045 .046 diff_Highedu 6 -.0078333 .0534917 -.083 .038 Variable Obs Mean Std. Dev. Min Max > pitagrowth2010
