Brain Drain and Brain Gain in Europe -- Thomas Straubhaar
MANAGING MIGRATION IN THE 21st CENTURY:
Winter 1998 Workshop
Center for US-Mexican Studies
University of California, San Diego
An Evaluation of the East-European Migration to Germany
Thomas Straubhaar, University of the Armed Forces, Hamburg
and Centre for Economic Policy Research, London
Martin Wolburg, College of Europe, Hamburg
Tel: 0049 40 6541 2850
Fax: 0049 40 6541 2875
Comments are very welcomed!
Revised version of June 12, 1998
Brain Drain and Brain Gain in Europe -
An Evaluation of the East-European Migration to Germany
Thomas Straubhaar, University of the Armed Forces, Hamburg
1 Introduction 12 Theoretical Considerations 32.1 Externalities from Migration 32.2 Internalisation and Compensation of Externalities 73 East European Migration and Brain Drain 93.1 The Eurostat Labour Force Survey 93.2 Some Stylised Facts 104 Is there an Empirical Source for Human Capital Externalities? 134.1 Data 134.2 Econometric Investigation 155 Calibrating the Welfare Effects of Emigration with Remittances 185.1 Parameters and Scenarios 185.2 The Gross Welfare Effect of East European Migration 205.3 Net Welfare Effect with Remittances 226 Conclusions 257 Literature 278 Annex I8.1 Annex 1: Data used for the Regressions I8.2 Annex 2: Selected Macroeconomic Parameter Underlying the Calibration III8.3 Annex 3: East Europen Countries Percental Reduction in GDP per Capita V8.4 Annex 4: Efficiency Gain and East European Migration to Germany, Mil. PPS VI
The free movement of EU workers within the Single European market is an integral part of the EU treaty (or better: treaty of the European Community). It belongs to the acquis communautaire that has to be adopted by the new members and has to be granted reciprocally to citizens from old an new EU member states sooner or later. There is no question that for the free movement of workers long transitional periods will be the compromise of the negotiation process. The old member countries of the EU are afraid of a trek to the West. They expect a mass migration from Eastern to Western Europe and fear to be overflowed by strong waves of immigrants crossing the Oder-Neisse-river like the wet backs the Rio Grande. These quantitatively impressive expectations are confirmed by almost all estimations of the potential East-West-migration within an EU enlarged by Central and Eastern European countries (e.g. Layard et al., 1992). Baldwin (1994) suggests that between 3 to 6 million Visegraders (i.e. people from Poland, the Czech Republic, Slovakia and Hungary) would move West. Franzmeyer/Brücker (1997) have estimated annual net migration flows of about 300'000 people from the five CEFTA countries. With regard to these figures it does not surprise at all, that the free movement of people has been treated extremely silent in the Europe Agreements made with all ten Central and Eastern European countries applying for EU membership. In short, the topic of free movement of people has been left more or less entirely in the competence of the single EU member states and there is (almost) no common EU regulation with regard to the immigration of Eastern Europeans. More interesting is that this rather restrictive attitude against Central and Eastern Europe stays even behind the agreement of the EC with Turkey in 1964 and far behind the European Economic Space agreement with the EFTA members.
Why are the old EU members so much afraid to grant the right of free movement of people to the possible new Central and Eastern European members? Of course this has something to do with the effects of migration and especially with the impact on immigration to certain interest groups. Migration creates winners and losers. While normally the negatively touched vested interest groups dominate the political debate, we turn around the focus in our paper. We will demonstrate in this paper that free movement of labour within an enlarged European Union would be profitable for the receiving countries in the West and North but rather harmful for the sending countries in the East. In short, our conclusions draw on the assumption that it would be most probably the rather skilled, relatively young and dynamic men who will leave Central and Eastern Europe to look for their luck in other countries of the EU - most likely Germany. But then, an old story gains new interest. It is the negative effect of the emigration of skilled people on those left behind which has been called "brain drain." By definition brain drain is the permanent emigration of qualified persons. The drain of brains from East to West does not harm the receiving Western European countries. As a matter of fact the opposite is true. The East-West migration will rather be an economic loss for the sending Central and Eastern European countries and a gain for the EU countries. Consequently, as the Central and Eastern European sending countries would be harmed by a too fast and too far reaching liberalisation of labour movements, it would be in their own interest to postpone the right of free movement of workers within an enlarged EU until the emigration incentives are sufficiently low to avoid negative impacts on their own welfare.
In our paper we take a closer look at the development of the volume and composition of East-West migration. We will discuss the conditions under which highly qualified migration really constitutes a drain of brains. Up to now the whole discussion has been mostly emotional: On the one hand the brain drain is, as implied by its name, viewed as an evil. On the other hand it has been put forward that migration of highly qualified persons contributes like any other factor movement to an efficient allocation of resources and is thus desirable. It is our aim to clarify the conditions under which both point of views hold theoretically and to judge on the empirical relevance of this topic.
The problem of highly qualified emigration from Eastern Europe is already recognised by the pertinent literature. Although migration statistics with respect to the migrant's qualification do not exist (except for the US) some remarkable evidence of brain drain can be found in the literature. Table 1-1 summarises some literature on the brain drain's volume for selected East European countries.
250 Scientists of the academy of science (20%)
600 members of the academy of science; the most productive
4% of the emigrants are students, 80-90% want to leave permanently
not specified; permanent contracts
contracts , primarily Israel, Germany & the US
Work in the West
34,4% of the emigrants are intellectuals
many want to emigrate permanently
12,1% of the emigrants were highly qualified
19.800 Engineers, 8.800 Scientists & academics, 5.500 doctors, 6.000 nurses
Germany, US, France and others
from 81-88 approx. 50-55% migrated to Germany, a great part being ethnic German emigrants (Aussiedler)
Source: Wolburg (1997: 32)
We will concentrate on the German case because Germany has been and still is a major receiving country of migration flows from Central and Eastern Europe. Furthermore, relatively good data are available.
In what remains, the structure of the paper is as follows: In section two we outline some theoretical arguments concerning the welfare effects of migration with special reference to transforming countries. We review the welfare effects that are associated with the non-pecuniary externalities of migration. These effects are the loss of investment in publicly provided education, an increase in the rate of unemployment and a possible reduction in the income level because of the existence of externalities. Traditionally these non-pecuniary externalities of migration are called the "brain drain" effects of migration. In the third section we take a deeper look at the East European migration to Germany and analyse the volume of the brain drain. In the fourth section we investigate econometrically whether externalities from human capital exist by estimating a production function for the EU (12) countries. After having confirmed the existence of externalities we argue that these externalities will also be present in case of the East European countries. Based on this assumption several scenarios concerning the net welfare effect of highly qualified emigration on those left behind are constructed. The final section draws some conclusions.
2 Theoretical Considerations
2.1 Externalities from Migration
Within the neoclassical framework income effects of international factor movements can be subdivided into a factor endowment and a terms of trade effect. The former results from a change in the factor endowment holding commodity prices constant, the latter represents the impact of changing commodity prices on national welfare. The direct income effect due to an altering factor endowment effect can best be illustrated in a simple neoclassical one good and two factor model (the basic model). Under these assumptions Grubel and Scott (1966) analysed the effect of emigration on the welfare of those left behind. Their fundamental conclusion is that an infinitesimal emigration leaves the welfare of the remaining population unaffected. Berry and Soligo (1969) in contrast pointed out that finite emigration affects the well being of those left behind by altering the factor proportions and thus per capita income. There will be no effect on income, if capital is internationally mobile and thereby equalises factor returns (Clarke, 1994a: 82) or if the capital-labour-ratio stays constant. Within this class of models one can think of highly qualified persons as being endowed with more capital than the average person, thereby implicitly assuming a perfect substitutability of physical and human capital. Consequently highly skilled emigration causes per capita income to rise in the receiving country and to decline in the sending country.
If we consider an incompletely specialised open economy terms of trade effects (TOT hereafter) have to be taken into consideration as well. Standard neoclassical trade theory argues that product prices are determined by demand, production technology and factor endowment. Taking the technology within the trade theoretical framework as given TOT can be influenced via migration through demand and supply effects (Dixit/ Norman, 1993: 144). Whereas the former can take place if the tastes of the migrants and residents in the immigration country differ, the latter is due to a changing factor endowment. Neglecting demand effects of migration by the assumption of identical homothetic preferences within a Heckscher-Ohlin world it is essential to distinguish among a small and a big country: Whereas the former faces constant TOT given by the world market the latter country may influence the TOT. As a consequence the only effect of migration in case of a small country is an alternation of the capital-labour-ratio and hence of per capita income. With regard to Eastern Europe the assumption of a small country seems to be plausible since its aggregate volume of trade is small relative to the total European trade volume. From this one does not expect migration to have any impact on the factor endowment. If however non tradable goods are taken into consideration (Rivera-Batiz, 1982), the price of the traded good is given from the world market but the price of the non traded good is determined endogenously and also depends on the factor endowment. Since the price of the tradable good is exogenously given the TOT solely depend on the prices of the non tradable good. Alam (1982) argues that in developing countries highly skilled people primarily work within the service sector which is (human) capital intensive. Brain drain causes the service sector to contract and its price to increase which implies a worsening of the TOT. Note that the reverse result is obtained under the assumption of a labour intensive service sector.
In contrast to pecuniary externalities technological externalities of migration arise if some neoclassical assumptions do not hold. For the sake of simplicity we do not differentiate between institutionally induced kinds of market failure and summarise all types of market failure under the category technological externality. It is a common feature of this group of externalities that migration under such conditions does not only have distributional but also allocational implications. Within the context of migration our special focus is on indivisibilities, positive externalities from skilled labour, unemployment and publicly provided education.
1. First, if indivisibilities are allowed the assumption of constant returns to scale is violated. Skilled migration may cause a small country to become even smaller: Markusen (1988) models a two country world in which both countries produce a modern human capital intensive good under increasing returns to scale and a traditional good under constant returns in autarky. If free trade is allowed the greater country will specialise in the production of the modern good and thereby cause the wages of skilled labour to diverge which in turn creates strong migration incentives. Allowing for factor mobility skilled labour thus leaves the country making the small country smaller and leading it to specialise in the production of the traditional good. The emigration of skilled labour exerts a negative effect on per capita income. The same line of argumentation is pursued by Krugman (1991) who models the emergence of a diverging centre-periphery pattern through migration: Since the production function exhibits increasing returns to scale in the input of labour large scale emigration is likely to occur. This mass emigration clearly benefits the immigration area (i.e. the center) but leaves the remaining immobile factors of production worse off (in the periphery) and thereby contributes to income divergence.
2. The second kind of market failure associated with skilled migration are positive externalities stemming from human capital. Clearly, if the social marginal product of labour exceeds the private one the loss in per capita income due to the endowment effect is magnified. This argument is frequently put forward within the brain drain literature (e.g. Bhagwati, 1976a: 714) and supported by recent findings: Winter-Ebmer (1992) investigates wage differentials within Austrian industries and finds that the average years of schooling exert a positive influence on the wage differential. Weale (1992) identifies a positive relationship between per capita income and the share for highly educated at the population for Great Britain. Hence a significant decrease in the average stock of human capital will lower the income overproportionally.
The link between migration and growth has traditionally been discussed within a Solow-like (1956) neoclassical growth model. Consequently migration can influence the steady state income level through an impact on the saving rate, the fertility rate and the rate of technological progress. If a distinction between human and physical capital is met an outflow of non-accumulable human capital will slow down the steady state per capita income level. A reduction in the long-run income level also follows, if migration speeds up the population growth rate of a country. The welfare effect of an altered saving rate (Galor, 1989; Galor/ Stark, 1991) in contrast depends on whether the country initially over or undersaves with respect to the golden rule of accumulation. If the latter is the case the emigration of high savers will reduce the welfare of those left behind. Of great theoretical interest as these effects are, we believe that they are of minor empirical relevance: First, the flow of skilled persons compared to its stocks are not significantly different from the level of marginality and consequently leaves the population size and thus the per capita income level almost constant. Second, the population growth rate generally does not depend on the flow of migrants but is determined by socio-economic factors. Immigrants furthermore quickly assimilate themselves and adopt the fertility pattern of the host country (see OECD, 1991, esp. chapter III; Blau, 1992; Kahn, 1994). Finally, capital markets nowadays get increasingly integrated. As a consequence an alternation of the national saving rate will affect the current account of a country but leave the investment decision unaltered.
Thus the only long-term determinant that remains to be discussed is the rate of technological progress. Whereas it has traditionally been regarded as exogenously given, the recently developed New Growth Theory (see Barro/Sala-i-Martin (1995) for a survey) attempts to endogenously determine the rate of technological progress. This gives rise to interesting links between the flow of skilled migrants and long-run growth that will be briefly outlined in the aftermath.
a) First, if based on the model of Lucas (1988) the human capital endowment of an economy is significantly lowered, then it can be shown that under reasonable parameter constellations (Xie, 1994) the income of the emigration country can never catch up with the leading countries. The basic assumption for this channel of work is the existence of locally limited knowledge spillovers.
b) Second, production costs in countries with a large skilled labour force will decrease and make those countries more competitive. Through the increased competitiveness on the world market this country will specialise in the production of high tech goods. In the same spirit Bretschger (1993) argues that within a Grossman and Helpman (1991) multi-sector growth model with trade the outflow of skilled labour will cause its wage rate to rise and thereby worsens the country's competitiveness on the world market. The R&D sector shrinks and long-run growth is likely to be slowed down.
A completely different approach is chosen by Wolburg (1996) who constructs a model where income growth is driven by an increase in the division of labour. As far as the immigrant with an average skill level imports new skills into the country the division of labour and hence the income is increased. The total effect of immigration can only be negative if the immigrant possesses a skill level far under the average skill level of the receiving country. However, within a dynamic setting divergence in income levels among countries is possible. Rich countries clearly attract the inflow of additional skilled labour and thereby enhance the divergence pattern.
Summing up, it follows from arguments of the New Growth Theory and Agglomeration Theory that on theoretical grounds the principle of cumulated causation is also of importance in the field of migration.
3. Apart from those effects brain drain can additionally reduce per capita income additionally through another channel: If education is positively related to ability the emigrants will be substituted by less able persons thus lowering overall productivity and output (Berry & Soligo, 1969: 792). As far as the production of human capital (i.e. schooling and professional or academic education) is subsidised, the emigration country loses human capital when people with human capital leave their origin. Consequently (and according to the theory of public goods) the production of human capital in the emigration countries is too low in comparison to a world without migration. This is the comparative static equilibrium analysis of the brain drain as it was discussed by Bhagwati and his scholars (see Johnson, 1967; Adams, 1968; Bhagwati, 1976b, 1979, 1985, 1985b; Bhagwati and Rodriguez, 1975; Grubel and Scott, 1977).
4. Fourth, highly qualified persons might constitute a narrow pass for certain kinds of development projects. In this case the output loss is magnified and may go along with thereby induced unemployment: If skilled labour is a narrow pass, its emigration induces unemployment because a substitution by unskilled labour is impossible. Generally, skilled emigration can hinder economic development if specialists emigrate on a large scale. This effect has been intensively discussed with respect to developing countries in the 1960s. According to Rhode (1991) the emigration of doctors from Poland in the early 1980s led to a regional undersupply.
5. Fifth, brain drain induces a negative fiscal externality on the emigration country, if education is publicly subsidised. If the economy wide education is expanded in response to emigration the governmental deficit increases (Bhagwati, 1976a: 715) ceteris paribus. This effect is accelerated, if the newly educated are less gifted. Furthermore educational subsidies can be regarded as an investment of the old generation into their pension (Grubel & Scott, 1977: 39p.) which is lost in case of permanent emigration.
Summing up, on theoretical grounds there is a strong conjecture that the emigration of skilled labour will harm those left behind whereas the emigration of unskilled labour will in contrast be beneficial for the sending country. The migration of highly skilled labour might lead to a divergence between the centre and the periphery ('the Mezzogiorno Problem'). Unless policy measures are introduced to offset the effects of migration of skilled workers, polarisation effects may become self sustaining and peripheral regions may find it very difficult to recover economically. Keeping this in mind, what can be done in order to avoid those negative effects ? Traditionally the literature discusses possibilities either to internalise or to compensate for externalities.
2.2 Internalisation and Compensation of Externalities
With regard to the internalisation, the standard suggestion to deal with the brain drain has been the introduction of an exit tax for brain drain in the form of a surcharge on their income tax in the destination country (Bhagwati & Dellafar 1973). The migration tax is a pure allocationally oriented tax in the sense of Pigou with no direct fiscal motivation. The only focus is to internalise the non-pecuniary externalities of the brain drain. The immigration country should collect the tax and then transfer this money to the country of origin to compensate for the losses in human capital.
There have been many discussions about the precise design of the migration tax, the optimal tax rate and the moral justification of taxing migration (for an overview cf Bhagwati & Wilson 1989). The most important problem seems to be the turning around of the international (tax) law. Today, with the exception of the United States, most countries view citizenship as irrelevant for tax purposes and residency is the relevant jurisdictional nexus. That is to say: Residents are taxable on their world-wide income but non-residents are not taxable. People leaving their home country do not fall under the tax laws of the country of which they are citizens. A migration tax as proposed from Bhagwati would mean that the country of origin could levy taxes on her citizens even if they lived outside the country. That is to say: Citizens are taxable world-wide and not all residents are fully taxable by the country of residence.
The problem with the brain drain tax proposal is that it does not take into account the net gains of immigration in the receiving countries, the possibility of return migration and the beneficial effects of remittances and the acquisition of human capital in the case of temporary migration. Furthermore, we do not know (as it is usual with a Pigou tax) what size the non-pecuniary externalities have and what tax rate should be applied to the migrants. An optimal Pigou migration tax would internalise completely the non-pecuniary externalities. But the general problem is that the proper Pigovian migration tax is not a tax equal to marginal non-pecuniary externalities at the existing level of migration but is a tax equal to marginal non-pecuniary externalities at the optimal level of migration. This means that we have to look first for the optimal level of migration to determine the level of the proper migration tax (Cropper & Oates 1992: 685).
The taxation of the migration legitimised by the non-pecuniary externalities of subsidised human capital production in the emigration country assumes that the migrants are able to transfer at least a part of their human capital - acquired in the home country - into the immigration countries. Do the immigrants earn higher incomes as a consequence of higher human capital they acquired by subsidised schooling? Taxing immigrants is only legitimised if the human capital acquired in the home country leads to significantly higher income in the immigration country relatively to people with less human capital. Only in that case we have some non-pecuniary externalities of migration that give a reason for taxing migrants and for remitting the taxes to the countries of origin where the human capital has been produced with public subsidies.
In a world without immigration restrictions and a high elasticity of migration for all individuals Wilson (1992) has shown that a Pareto-efficient policy in all communities is to impose a poll tax equal to marginal congestion cost associated with public goods provision. In equilibrium governments do not redistribute income. If only one group of citizens has a high migration elasticity then it is optimal to tax low skilled citizens even more than their marginal cost which amounts to a redistribution from "the poor" to "the rich".
Clarke (1994b) has applied some of these thoughts to a world in which, as it usually is the case in the real world, migration restrictions do exist. A Pigovian tax can then serve two purposes: First, it can ensure efficient pricing of congestion externalities as long as transaction costs do not exceed the tax-related efficiency gains. It would, however, be impractical to reimburse revenues to residents alone which would be required if the goal was to maximise national welfare. Second, it can provide a second best solution in the case where an arbitrary quota creates excessive demand for entry. The fee should then reflect the size of the quota.
Summing up, the real world possibilities as well as moral founding of taxing skilled emigrants are very limited. Thus it is unlikely that the implementation of measures to internalise the brain drain's effects will be implemented. We therefore draw our attention now towards endogenous compensation measures. More specifically, we will consider the welfare effect of remittances, and capital mobility.
If emigrants remit a certain amount of their earnings under the assumption of a non-tradable good by means of tradable goods, Djajic (1986) argues that the remaining population is enabled to hold the internal TOT constant. Remittances can be viewed as an endogenous compensation mechanism for the sending country. Within the brain drain literature it is even stated that the presence of remittances can alter the brain drain into a brain gain for the country of emigration. Abu Rashed and Slottje (1991) argue that remittances can be used for productive investment and thereby raise the per capita income for the remaining population. This theoretical possibility is empirically confirmed by Brown (1994) who finds that remittances made a significant contribution to saving and investment in the South Pacific Islands. The same author argues that remittances primarily take place within families and also consist of goods which are transferred via informal channels. Hence the amount of remittances from the balance of payments will generally underestimate the true volume of remittances.
Relaxing the assumption of identical preferences, Basu (1992) argues that if emigrants do not take all their capital with them, the capital-labour-ratio and income per capita of those left behind tends to increase. If furthermore emigrants stick to their consumption habits, they can thereby raise the home country's exports which exerts a positive effect on the home country's TOT. Apart from this remittances will compensate those left behind for the use of publicly provided goods, e.g. education.
From the above discussion it can generally be concluded that brain drain causes a negative income effect either because there is not sufficient capital mobility or through capital dilution when trade is absent or the country is completely specialised. Within an open diversified economy migration has no impact on goods and factor prices, if the change in factor endowment does not lead to a change in good prices. If the TOT are altered instead the human capital intensive sector contracts which in turn is welfare reducing, if the export good is relatively labour intensive. Capital mobility eliminates the welfare effects, if owing to it the factor prices are determined by good prices. Increasing returns to scale in the human capital intensive sector magnify the income loss and cause much more easily complete specialisation. In a Harris-Todaro world we generally expect a negative welfare effect from the influence on income as well as on unemployment (see Wolburg (1998) for the detailed arguments).
Summing up, the East-European brain drain is from a theoretical point of view likely to harm those left behind through a decrease in per capita income and the generation of unemployment. On the other hand remittances, international capital mobility and capital that migrants do not take with them can at least partly compensate those left behind. In the following we analyse both the degree to which the average level of human capital is lowered by means of brain drain and whether externalities are empirically existent.
3 East European Migration and Brain Drain
More precisely this section is organised as follows: At first (3.1.) we take a look at the development of migration to Germany (3.1) and differentiate among qualified and unqualified immigration by making use of unpublished data from a Labour Force Survey (LFS).
3.1 The Eurostat Labour Force Survey
The data used within this article are based on Eurostat's Labour Force Survey (LFS) which is a sample based on yearly standardised interviews within the European member states. Since a more detailed description of the LFS data can be found in Wolburg/Wolter (1997) we only outline its essential features in this paper. The variable "qualification" is distinguished in highly qualified and unqualified persons according to the highest level of completed education obtained. Highly qualified persons according to this definition basically comprise persons that exhibit some kind of third-level qualification. Within the classification of the LFS this corresponds to the holding of a third-level qualification but not a university degree, a university degree or recognised equivalent and a post-graduate qualification (Eurostat, 1992: 25). A person is identified as being a migrant, if the place of residence in the previous year is another than the present. Unfortunately the disaggregation of sending countries is not very deep prior to 1992, a fact which renders the analysis for the period of 1988-94 possible only for two East European countries, namely Poland and the former Yugoslavia. As a result we concentrate on the East European immigration to Germany for the period of 1992-94.
3.2 Some Stylised Facts
Table 3-1: East European Cumulated Immigration Flows into Germany According to Qualification, 1992-94, in 1,000 Persons Sending Country Aggregated Immigrants according to Qualification
Table 3-1 demonstrates that the cumulated share of highly qualified immigrants varies across countries. Whereas it is highest for Hungary, immigrants from the former Yugoslavia exhibited the lowest skill ratio. This is not very surprising since the civil war in the former Yugoslavia induced asylum migration which cannot be compared to voluntary economically induced migration as it takes place in case of the other countries. From the first row of Table 3-2 we see that the share of highly qualified persons at the German population is 0.13 for the period considered. Compared to the average qualification of persons residing in Germany, the human capital content of immigration from Poland, the former Czechoslovakia, Hungary and Bulgaria lies significantly above this value, whereas the opposite pattern holds in case of Romania, the former Yugoslavia, Albania and the former USSR. On the aggregate level the share of highly qualified persons is below the German average at the population.
However, we already hinted at the importance of the long-run perspective in which the migration pattern can differ from the instantaneous one due to remigration of unsuccessful persons. In order to analyse the long-run mobility pattern which also takes past migration into account we take a look at the stock of foreigners residing in Germany differentiated according to nationality and qualification. In analogy to the former table we compute the share of highly qualified persons at the entire stock of foreigners residing in Germany according to nationality. The resulting ratio is called the aggregate cumulated skill ratio and is plotted in the following Table 3-2:
Table 3-2: Aggregate Cumulated Skill Ratios of Migrants in Germany, 1992-94 Aggregate cumulated skill ratio (stock) Average (1992-94)
As deduced from the above Table the average cumulated skill ratio exhibits a value of 0.09 for the 1992-94 period which is significantly lower than the corresponding share of highly qualified persons at the German population. However, this is due to the very low educational attainment of persons from the former Yugoslavia. Taking them out of the calculation it gets clear that the cumulated skill ratio for all countries included and for all nationalities separately (except Albania) considerably exceeds the share of highly qualified persons at the German population.
In this respect long-run immigration into Germany can be called self-selective: The share of highly qualified immigrants is above the share of highly qualified persons at the German population. If we depart from the very plausible assumption that the share of highly qualified persons in the population of the East European sending countries does not exceed the share of highly qualified persons in the German population, it still holds that highly qualified persons emigrate overproportionally. Thus emigration is self-selective as well. For the sake of clarification our definition of self-selectivity is an ex-post one which only considers the resulting and empirically observable mobility pattern. This is to be distinguished from qualificational specific differences with regard to the incentives to move.
Referring to the theoretical arguments outlined in section two we can thus draw the conclusion that immigration from Eastern Europe on the aggregate has positively contributed to the German average stock of human capital. The aggregate effect of migration on the average stock of human capital can be evaluated by comparing the present average level of human capital with the level of human capital without mobility. This is done in the following Table 3-3.
Table 3-3: The Brain Drain's Effect on the Average Stock of Human Capital Country Absolute Change Percental Change
Bulgaria -0.0003 -0.0003 -0.0002 -0.51 -0.32 -0.20
In the above Table we evaluated the impact of emigration on the share of highly qualified persons in the population () of the sending country. We did so by computing the change in the share of highly qualified persons in the country examined. That is, e.g. in the case of Poland, we calculated the change in the share of highly qualified persons as a result of migration by comparing the assumed initial share of highly qualified Polish persons in the entire Polish population (e.g. 0.08) with the corresponding share of highly qualified Polish persons residing in Germany from Table 3-3. In order to compute the impact of emigration on this share in Poland we weighted the share from Table 3-3 with the share of the Polish people living in Germany in the entire Polish population, that is Polish people residing both in Germany and in Poland. In order to take into account fluctuations we based our computations on the average value for the period 1992-94.
Clearly the alternation of the average stock of human capital depends on the initial stock of human capital. As a matter of fact relevant statistics are not available for East European countries. We therefore had to construct different scenarios with regard to the average stock of human capital. With respect to the average level of qualification the share of highly qualified nationals at all nationals ranges for the rather lower developed EU countries from 0.05 (Italy) to 0.11 (Ireland) (Wolburg/Wolter, 1997: Annex 7). We believe that those countries are a relatively good approximation for the East European countries as well. Taking the heterogeneity of countries into account and considering furthermore that some East European countries lack behind Western European ones, we believe that an average share of highly qualified persons at the population of 8 % is a good approximation of reality.
In any way, as one can easily deduce from Table 3-3, the absolute decrease in the average stock of human capital is relatively robust to the assumption of different initial average stocks of human capital - this is true either within or between the Eastern European countries. The brain drain effect remains low in absolute terms and clearly decreases with the assumed initial share of highly qualified persons at the population. In any country and for every assumed stock of human capital the brain drain effect remains far below 2%. With the assumption of being made it follows from our previous analysis that the former Yugoslavia and Albania gained through emigration by raising the share of highly qualified persons at the population. In contrast the German average level of qualification is clearly raised by immigration from Eastern Europe if, as previously argued, the former Yugoslavians are left out of the calculation. Without immigrants from this region the share of highly qualified persons at the German population would on average be 0.0008 lower which equals an increase of the average level of qualification by 0.65 % in 1994.
Summing up, on the aggregate level one could talk of a brain drain with regard to the emigration pattern from Eastern Europe and talk of a brain gain with regard to the German immigration pattern: Whereas the average level of human capital for the relatively wealthy receiving country is raised it is lowered for the relatively poor country of emigration. However, brain drain cannot be more than a marginal phenomenon - the average relative effect remains far below 2% of potential human capital loss or gain. However, the qualitative impact of small marginal effects is sometimes much more important than we might assume from low absolute figures! For the brain drain phenomena it is much more crucial whether the migration of highly skilled people generates some positive technological externalities. We therefore investigate econometrically within the following section whether externalities from human capital exist in reality as well.
4 Is there an Empirical Source for Human Capital Externalities ?
As shown by Goto and Suzuki (1989), with the assumption that real investments (I) grow with a constant growth rate () and that they are by a time invariant depreciation rate (), the capital stock (K) at a certain time (t) for a certain country (i) can be determined recursively by means of the following formula:
Thus in order to estimate a country's capital stock information about the depreciation rate and the growth rate of real investment are needed. Following a standard procedure from the literature, the growth rate of gross capital formation can be estimated by regressing the growth rate of investment against time (where c is a constant):
This yields the following results:
Table 4-1: Estimated Gross Investment Growth Rates for the EU(12) Countries, 1960-94 Var. i= ...
0.018 0.033 0.030 0.022 0.022 0.024 0.022 0.039 0.014 0.039 0.033
-23.882 -53.189 -54.641 -32.409 -33.745 -41.440 -31.570 -68.808 -18.471 -65.394 -55.208
After having estimated the capital stock in the way described above it is taken into account that due to business cycle impacts the degree of capacity utilisation varies over time. In order to adjust for this the estimated capital stock is multiplied with the average yearly rate of capacity utilisation taken from the European Commission's Business Survey (various years). Only in the case of Spain data were not available over the entire period. The corresponding capacity utilisation rates are therefore based on OECD data that are taken from the Eurostat (1996) CD-ROM as well.
In order to divide the employed labour force into qualified and unqualified labour, two concepts of qualification can be applied: The first approach defines qualification by the formal level of education, whereas the second approach measures qualification according to the profession.
According to the educational specific operationalisation the category "highly qualified" comprises all persons within the survey with third-level non-university education, third-level university education or post-graduate studies and is in detail described in Wolburg and Wolter (1997). This categorisation does not take into account the on-the-job learning. Furthermore, the share of highly qualified persons according to this operationalisation refers to the entire population and we got no data on highly qualified persons according to education that are employed.
The second approach that is used here adopts a job-specific measure of qualification based on data from the Labour Force Survey (LFS). Within the survey the working population can also be splitted according to professions based on the ILO's International Standard Classification of Occupations (ISCO-88(COM)). Within this methodology professions consist of ten main groups. Those groups are legislators, senior officials and managers (group 1), professionals (group 2), technicians and associate professionals (group 3), clerks (group 4), service workers and shop and market sales workers (group 5), skilled agricultural and fishery workers (group 6), craft and related trades workers (group 7), plant and machine operators and assemblers (group 8), elementary occupations (group 9) and armed forces (group 0) (Eurostat, 1992: 35p.). Clearly, highly qualified occupations like managers, scientists and technicians fall into the first three groups which are therefore defined as highly qualified occupations.
The advantage of this second method to operationalise qualification is, that according to information from Eurostat the data base is larger than in the first case. Furthermore it should be noted that there exists a severe correlation between both types of operationalisations. This correlation is highest for the persons contained in the occupational group 1. However, there is a missing value for Belgium in 1992 which was estimated by taking the average value between 1991 and 1993. We also got data on the share of highly qualified persons according to education at the population. Furthermore the required data on qualification are generally available for 1988-94 except for Spain and Italy (1992-94) and for the Netherlands (1990-94). Thus the available data from an econometric perspective constitute an unbalanced panel.
As outlined in the theoretical section, there is a strong theoretical conjecture that externalities from the average level of human capital exist. In order to test this conjecture we define the average level of human capital (HA) with regard to education as the share of highly qualified persons at the entire population. In principle one could also choose an operationalisation with regard to the occupation. However, there exists a positive correlation between both formulations and both possibilities are detailly discussed and tested in Wolburg (1997). Within our context we only want to explore the question whether migration of highly skilled generates positive externalities. Since the migration data only allow an operationalisation of qualification with regard to the education, our specification of the average level of human capital is directly compatible to the migration figures. The data on which our econometric investigation is based are plotted in Annex1.