Che materia stai cercando?

The Impact of EU Accession on Human Capital Formation Appunti scolastici Premium

Dispense per il corso di Differenziali Economici e Migrazioni della Prof.ssa Paola Giacomello e del Dott. Paolo Sellari. Trattasi dell'articolo di Emily Farchy dal titolo "The Impact of EU Accession on Human Capital Formation: Can Migration Fuel a Brain Gain" all'interno del quale l'autrice espone la sua teoria sull'effetto che la... Vedi di più

Esame di Differenziali Economici e Migrazioni docente Prof. P. Giacomello



the use of lagged migration rates as an instrumental variable is that increased

enrollment in time t cannot influence the migration rates at time t−1. However,

it is possible to imagine that rational agents will be able to foresee changes in

human capital the following year, for example, in an economic downturn, it

may be expected that, as the job market shrinks, the opportunity cost of higher

education is reduced and hence more people are likely to enroll the following

year. If this is the case then the lagged migration rates, employed by Beine et

al (2007), may not entirely eradicate the endogeneity problems.

Since a failure to account for some potential reverse causality is likely to

result in biased estimates this paper utilizes the natural experiment of EU ac-

cession and the exogenous impact this has on the parameter p (the probability of

successful migration). Furthermore, to avoid the data issues that have stymied

much of the previous research, this paper has an altogether narrower focus:

does an increase in migration opportunity lead to an increase in investment in

human capital? The analysis that follows abstracts from measurements of the

brain drain—the relocation of skilled workers—focusing instead on the incen-

tivization effect of migration. In order to do this I look at the impact of an

exogenous increase in p (the probability that intended migration will succeed)

on levels of human capital in the origin country.

3 Analytical Methods and Main Results

Accession to the EU provides a natural experiment to test theories of migration

fuelled increases in human capital. This is because, as a member of the EU, a


countries citizens benefit from the free movement of persons (Article 39 of the

EC Treaty). In terms of the brain gain model outlined in section 1.4, such a

policy change represents a significant increase in the parameter p (the possibility

17 . And, whilst the available data limits analysis of

of successful migration)

16 The relevant rights are complemented by a system for the co-ordination of social security

schemes and by a system to ensure the mutual recognition of diplomas.

17 Note that freedom of movement does not increase p to the level that p = 1. If it were to

do so there would be no possibility of a brain gain because all those who invest in education

13 18

migration flows disaggregated by education level, the magnitude of the change

in enrollment that follows this increase in p should give some indication as to the

feasibility of the brain gain hypothesis that migration may increase the stock of

human capital.

To capture investment in education I use the Gross enrollment Ratio; the

number of pupils enrolled in tertiary education, regardless of age, expressed


as a percentage of the population in the theoretical age group for tertiary

education. The GER is calculated according to the following formula:




GER = (17)

t ·

(P ) 100



ˆ t

GER is the Gross enrollment Ratio at time t,

ˆ t 20

E is the enrollment at time t,

ˆ t

P is the population in age-group a which corresponds to tertiary educa-

a 21

tion in school-year t .

It should be noted at this point that, if it is the case that enrollment in ter-

tiary education increases with accession to the EU, this is unlikely to be solely

a result of students wishing to use their new rights to migrate. Accession to

the EU will almost certainly also effect domestic employment opportunities -

in order to migrate will do so. The time lag and uncertainty about the future ensure that

whilst migration is a right it remains that p < 1. This is because there are many other

factors involved in the translation of intention to migrate into reality, for example changing

personal/economic circumstances may have a substantial impact given the extended time

period required to gain a tertiary qualification. Indeed, micro-level surveys on intentions to

migrate consistently exceed actual migration figures. Note further that the indiscriminate

nature of the freedom of movement of people implies that of the two reasons a potential

migrant may wish to invest in education; firstly as a vehicle for exit, and secondly to benefit

from the increased wage advantage abroad, only the second is relevent here. Thus the findings

are likely to represent a lower bound on the estimated incentive effect of increased migrational


18 As mentioned above the recent dataset by Docquier and Marfouk has begun to relax this


19 For the tertiary level, the population used is the five-year age group following on from the

secondary school leaving age.

20 Many thanks to Lucy Mei Hong (UIS UNESCO) for obtaining the historical data not

publicly available on the internet.

21 See http://esa.unorg/unpp/forUNDPdataonline.


through increased Foreign Direct Investment (henceforth FDI), as well as in-

creasing trade flows - thus increasing returns to investment in human capital

may also result from an improved domestic climate. Nevertheless, the impact of

increased FDI and trade flows will likely have been felt for several years preced-

ing the accession date. Indeed research into the effect on FDI of EU accession

(Bevan et al 2001) finds that as far back 1994 when the EU announced its com-

mitment to enlargement, front-runner candidate countries in Eastern Europe

experienced significant increases in inflows of FDI. Barriers to trade were also

reduced in anticipation of accession. Thus, if a positive effect on education

is identified, the role of the labor market mobility resulting from accession, in

increasing enrollment in tertiary education, should not be undermined.

The relative importance, in the decision to invest in education, of FDI flows

and the migration opportunity, can be examined through the lens of economic

geography models as outlined above. The degree to which accession encourages

dispersion across Europe (and hence inward flows of FDI to accession countries)

or agglomeration in Central Europe (and with it outward flows of skilled labor

from accession countries) is dependent on the relative importance of exogenous

parameters such as economies of scale and transport costs. And, although trans-

port costs are likely to be less signicant in high skilled sectors, the impact of

increased FDI and trade flows represents a possible source of upward bias on

the results. Thus, in the panel regression that follows, I control for FDI levels

as well as exports.

In order to assess the impact of European accession on enrollment in tertiary

education this paper employs two strategies. Firstly, in section 3.1, with a

simple difference-in-difference approach which, along the lines of a simplified

propensity score matching methodology, exploits the division of Czechoslovakia

in 1993. This enables the use of Slovakia as a counterfactual against which to

assess the impact of the EU negotiations that began with the Czech Republic

in 1997. Secondly, in an attempt to support these results, section 3.2 examines

the impact of EU accession on tertiary enrollment across a panel of 13 countries



that have joined the EU.

3.1 Czech Republic/Slovakia Case Study

3.1.1 Method

The goal of this section is to assess the impact that EU accession negotia-

tions had on the GER of the Czech Republic. To do this I follow Abadie’s

methodology using a difference-in-difference estimation but rather than create

a synthetic counterfactual, I utilise the disparate timing of the opening of EU

accession negotiations with the Czech Republic (in November 1997) and Slo-

vakia (in January 2000) to use Slovakia as a Czech counterfactual. If the brain

gain hypothesis is correct, and agents incorporate information relating to the

probability of a successful migration into their investment decision, I expect to

see a rise in the disparity between Czech and Slovak GER following the 1997

announcement of negotiations with the Czech Republic and a fall following the

2000 announcement of negotiations with Slovakia.

I have chosen these countries for two reasons. The first is obvious; until 1993


they were just one country and whilst there were economic differences they

were at least subject to the same policy environment and political shocks. The

second motivation behind this choice is the micro level survey data (IOM 1998)

suggesting these are the two countries with the highest potential migration.

Ultimately the Czech Republic and Slovakia joined the EU at the same

time, and so this example can only be utilized to look at the effect of accession

negotiations on enrollment. This creates a number of problems with the timing.

22 Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia,

Slovenia, Greece, Portugal, and Spain.

23 st

Prior to their separation on January 1 1993 the Czech and Slovak republics operated

under similar institutions and policies and by the time of their separation were considered to

have progressed more rapidly toward a market system than their East European neighbours

(with the exception of East Germany). However differences do remain: the Czech Republic

privatized faster, offered less generous social security benefits, and had a higher proportion of

smaller firms offering more opportunities for entrepreneurial activities. In addition the Czech

Republic’s long border with Germany enabled many workers to commute beyond its borders.

And, whilst the supply of more educated workers, following the fall of communism, increased

in both the Czech Republic and Slovakia; labor demand did not. The Slovakian economy

experienced higher unemployment than the Czech Republic.


Figure 1: Trends in Gross enrollment Ratio of the Czech Republic and Slovakia

Firstly there may be credibility delays; given that the investment required to

enrol in tertiary education is significant, both in terms of time and forgone

income, it may be the case that risk-averse agents did not respond immediately

to this signal of accession (and concomitantly future higher returns to investment

in education). Indeed it is likely that agents would wait to see how negotiations

progressed, to examine their credibility and content, before enrolling in a tertiary

course. Secondly there are likely to be delays in the data. This is because the

GER does not measure new students enrolled each year but only the total

number of students enrolled in tertiary education in any given year. As a result

the impact of increasing enrollment on GER is likely to be gradual. In order to

account for these potential delays in the impact of accession negotiations, after

looking first at the immediate impact I then consider a lagged impact, of three



3.1.2 Results

Below (figure 2) is a plot of the trend behaviour of the GER of the Czech Repub-

lic and Slovakia following their separation in 1993. The vertical lines represent

the opening of accession negotiations: the first with the Czech Republic, the

24 Three years is half of the average delay between the accession of the Czech Republic

and Slovakia, and the opening of accession talks. It is therefore used as a benchmark for

establishing the credibility of talks and the speed of progression.


second with Slovakia. The third line represents the accession of the Czech

Republic and Slovakia to the European Union in 2003.

Figure 2: Trends in Gross enrollment Ratio of the Czech Republic and Slovakia

Initially table 1 appears to present pretty damning evidence against the brain

gain hypothesis. The opening of EU accession talks does not appear to have had

a significant effect on the GER in the corresponding period; there was very little

difference in the GER of Czech Republic and Slovakia in the years following the

opening of accession talks with the Czech Republic; the difference in difference

between 1997 and 2000 is in fact negative. The difference in difference following

the opening of talks with Slovakia (table 2) is similarly unimpressive. Here we

would hope for a negative sign, indicating that Slovakia had begun closing the

GER gap, and in fact we find a positive.

If however, the possibility of a delay in the impact of EU accession is con-

sidered, and lags the GER 3 years, as discussed above, a discernable difference

emerges (Figure 3).

Between 2000 and 2002 (the opening of negotiations with the Czech Republic

lagged three years) the difference in GER between the Czech Republic and


Pre-Treatment: Talks Post Treatment Difference

with Cz

Slovakia 0.222346 0.261852 0.03951

Czech Republic 0.240419 0.264245 0.02383

Difference 0.018073 0.00239 -0.01568

Table 1: Difference-in-Difference of GER at time t (Treatment 1: The Opening of

Accession Negotiations with the Czech Republic)

Pre-Treatment: Post Treatment Difference

Talks with Slovakia

Slovakia 0.261852 0.303742 0.03951

Czech Republic 0.264245 0.308823 0.04458

Difference 0.00239 0.00508 -0.00269

Table 2: Difference-in-Difference of GER at time t (Treatment 2: The Opening of

Accession Negotiations with the Slovakia)

Slovakia increases by nearly 280% of its pre-treatment difference (Table 3).

Pre-Treatment: Post Treatment Difference

Talks with Cz (t+3)

Slovakia 0.287256 0.323255 0.035999

Czech Republic 0.294225 0.349701 0.055476

Difference 0.006969 0.026446 0.019477

Table 3: Difference-in-Difference of GER at time t (Treatment 1: The Opening of

Accession Negotiations with the Czech Republic (t+3))

Similarly it can be seen from Table 4 that in the last observations the Slovak

GER begins to rise at a greater rate than the Czech GER thereby beginning

to close the difference. This finding supports the hypothesis that the widening

difference-in-difference was a result of the disparate timings of the opening of

accession negotiations.

However, whilst the lack of immediate impact can be justified by considering

risk-averse agents, two concerns remain. Firstly, it is important to bear in mind

that Slovakia does not represent a perfect counterfactual to the Czech Republic.

From 1993 the two countries separated and became subject to separate policies


Figure 3: Trends in Gross enrollment Ratio of the Czech Republic and Slovakia

Pre-Treatment: Post Treatment Difference

Talks with Slovakia

Slovakia 0.361551 0.453154 0.0916

Czech Republic 0.431591 0.498499 0.06691

Difference 0.07004 0.04535 -0.0247

Table 4: Difference-in-Difference of GER at time t (Treatment 2: The Opening of

Accession Negotiations with the Slovakia (t+3))

and shocks. As a result there is the possibility that the trends identified above

are affected by omitted variable bias. A second problem, that has already been

mentioned, is the impact of FDI on increasing domestic returns to education.

As argued above, the EU’s announcement of its intention to enlarge in 1994 led

to significantly increased FDI inflows to the front-runner candidates in Eastern

Europe. Nevertheless, at this point, following the announcement of the open-

ing of accession negotiations the effects of remaining increases FDI flows may

impact upon domestic opportunities. This is because the opening of negotia-

tions represents a clear signal of the credibility of the candidate economy. After

this point the impact on FDI of actual accession could be expected to be less


In order to address these two concerns and further test the relationship be-

tween EU accession and increased enrollment in education, section 3.2 analyzes


a cross-country regression using panel data comprising of 13 countries that have

completed accession.

3.2 Cross-Country Panel Regression


The cross-country panel consists of the AC-10 as well as Spain, Portugal, and

Greece, who joined the EU in the 1980s. This earlier Southern EU expansion

is included whilst the incorporation of Austria, Finland, and Sweden in 1995 is

omitted, because in the 1995 expansion, there was no significant GDP per capita

gap between the accession countries and the EU member states. Furthermore

free movement of people was already in place in these countries as they were

members of the European Economic Area.

3.2.1 Method

Having ensured, via a Dickey-Fuller unit root test, that the assumption of non-


stationary appears valid, I can be confident that OLS results will not suffer

from the spurious regression problem. However, initial inspection of a GER

time plot suggests that a linear trend may not be appropriate thus, in order to

stabilize the variance I use a natural logarithmic transformation of the GER.The

empirical approach is therefore to a run panel data regression on 13 countries

that have joined the EU of the following form:


ln GER = α + β acc + δD + ε (18)

ct ct c ct

Where ln GER represents the natural logarithm of the Gross Enrollment


Ratio of country c in time t, acc is a dummy that switches on when country


c has completed accession, δD is the time fixed effect, γD is the country

t c

fixed effect and ε is an error term. Fixed effects at the country level control

for the usual array of cross-country differences in history, geography, culture,

25 Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia,


26 using the Dickey-Fuller unit root test approach given by Holden and Perman in B Bhaskara

Rao Conintegration for the Applied Economist 1994


policy and economic structure that have been constant over the sample period.

Whilst the year dummy controls for general trends and shocks affecting tertiary

enrollment across countries. β is the coefficient of interest measuring the effect

of EU accession on the change in tertiary GER.

3.2.2 Results (OLS) (OLS) (Clustered) (Clustered)


Accession 0.207*** 0.291*** 0.207* 0.178

(0.0542) (0.0597) (0.0992) (0.105)

FDI -0.00612 -0.00219

(0.00568) (0.00305)

Exports -0.00381*** 0.000954

(0.00128) (0.00374)

Year Fixed Effects yes yes yes yes

Country Fixed Effects yes yes yes yes

Observations 222 174 222 174

Number of country 15 15 15 15


R 0.888 . 0.888 0.885

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Estimate of impact of accession talks on ln GER. The data are for the 13

countries over the period 1970–2004. Czechoslovakia split into Czech Republic

and Slovakia in January 1993, and, after this date, they are included as separate

observations. The panel therefore consists of a total of 442 possible observations

with deviations accounted for by missing data.

The results in table 5 suggest that accession to the EU does indeed have

an impact on the levels of enrollment in tertiary education. Indeed, ceteris

paribus, it can be expected that on average the GER will be approximately


20 percent higher following EU accession. Column 2 presents the OLS results

controlling for FDI and Exports, column 3 preesnts the results clustered at


the country level to assuage concerns of serial correlation. Clustering does

indeed substantially raise the standard errors indicating that there may be serial

correlation among countries; nevertheless the results remain highly significant

at the 6 percent level. Finally column 4 presents the results clustered at the

country level and additionally controlling for FDI and exports, in this final

regression the results are less significant. Nevertheless, they remain significant

at the 12 percent level and the robustness of the coefficient to these various

specifications is encouraging.

The impact of EU accession on enrollment in tertiary education identified

above, lends support to the assumption of the endogeneity of human capital

formation and hence the hypothesis that increased migration opportunities en-

courage greater levels of human capital formation that may, given the right

parameters, offset or even overpower the brain drain that results from the relo-

cation of skilled labor. Such a substantial increase in enrollment suggests that

even in the presence of large outflows of human capital, there is the potential

for a positive net effect.

Further work may wish to redo the above analyses after some time has

passed. This would allow a number of refinements. Firstly, it would enable

a comparison of the post accession trends of the Czech Republic and Slovakia

therefore providing a check on the validity of the difference in difference ap-

proach. Secondly, the pass of time would provide more data to analyse the

impact of the Eastern expansion of 2004, it would also increase the sample by

enabling the inclusion of the most recent expansion of 2007 to include Romania

and Bulgaria. Lastly, given data continuing to a later date, it may be interesting

to look more closely at the possibility of lagged effects.

27 Bertrand, Duflo, and Mullainathan (2004) present an analysis of the dangers in ignoring

serial correlation in panel data, using Monte Carlo data and randomly generated placebo laws

to illustrate frequency of inaccurate conventional standard errors. Using standard difference-

in-difference standard errors they find a statistically significant effect in 45% of the placebo

interventions. 23

Another avenue for further work would be an analysis of the effects of dif-

fering levels of public funding for education on the above results. This would

enable an analysis more focused on the welfare implications of labor market

liberalization, and as such may identify more policy prescriptions. Clearly a

positive net effect on human capital stocks is not enough to conclude that the

economic impact of labor market liberalization is positive if the gains from in-

creased participation in tertiary education are funded entirely from the public

purse. The positive implications of the brain gain rely on an increase in the

private investment in human capital.

Empirical analyses that attempt to quantify the net brain effect continue to

be severely hampered by data deficiencies. Nevertheless incremental steps such

as those presented in this paper, can help to clarify what has for a long time

been, and will likely continue to be, a highly contentious area.


The brain gain literature relies for its positive prognosis on the assumption that

human capital is endogenous. This paper has shown that this assumption is well

founded. It has examined the relationship between migration opportunities and

the private decision to invest in higher education, looking first at the impact of

the announcement to open negotiations, through a difference-in-difference com-

parison between the Czech Republic and Slovakia, and secondly at the impact of

accession on a cross-country panel of 13 current EU members. Taken together

these results represent evidence that an increase in the opportunity to migrate,

heralded by accession to a free-trade area such as the EU, has a significant ef-

fect, both economically and statistically, on the proportion of the population

that choose to invest in higher education, and hence the skill level of the country.

While it is important to remember that this result cannot be entirely attributed

to the increased migration opportunities, as increased FDI and trade flows will

also raise the domestic return to education, micro-level survey data regarding


intentions to migrate indicate that migration to achieve a higher wage abroad

plays a significant role.

The example of the brain gain analyzed here represents a very narrow focus

on the gross impact of migration opportunities on human capital formation.

Clearly the net brain effect will depend also on outward flows of skilled labor.

However, if recent policy trends towards a focus on increasing temporary and

return migration are successful, the gross effect of migration on human capital

formation, identified in this paper, could enable migration to be utilized as a

powerful tool for development.


and (2001). “Human Resources in Science and Technology:

Auriol Sexton

Measurement Issues and International Mobility”, International Mobility of the

Highly Skilled,OECD

(2003) “The Economic Costs of Conflict: A Case Study of the Basque


Country” American Economic Review, 93(1) p113-132.

and (2001) “Brain Drain and Economic Growth:

Beine, Docquier Rapoport,

Theory and Evidence” Journal of Development Economics 64 p275-289

and (2003) “Brain Drain and LDC’s Growth:

Beine, Docquier Rapoport,

Winners and Losers” IZA Working Paper

and (2007) “A Panel Data Analysis of the Brain

Beine, Defoort Docquier,

Drain”, Discussion Papers 2007024, Universite catholique de Louvain, Departe-

ment des Sciences Economiques.

and (2004) “How Much Should we Trust

Bertrand, Duflo, Mullainathan

Differences-in-Differences Estimates?” Quarterly Journal of Economics p249-

275 and (2001) “The Impact of EU accession Prospects

Bevan, Estrin Grabbe

on FDI Inflows to Central and Eastern Europe”, ESRC Policy Paper

and (1974) “The Brain Drain, International Integration

Bhagwati Hamada

of Markets for Professionals and Unemployment” Journal of Development Eco-

nomics. 1(1) p19-42.

(1990) “Self-Selection and the Earnings of Immigrants: Reply”, Amer-


ican Economic Review. 80(1): p305-308.





277.40 KB




+1 anno fa


Dispense per il corso di Differenziali Economici e Migrazioni della Prof.ssa Paola Giacomello e del Dott. Paolo Sellari. Trattasi dell'articolo di Emily Farchy dal titolo "The Impact of EU Accession on Human Capital Formation: Can Migration Fuel a Brain Gain" all'interno del quale l'autrice espone la sua teoria sull'effetto che la partecipazione al mercato comunitario genera sulla formazione ed accumulazione di capitale umano negli stati membri.

Corso di laurea: Corso di laurea magistrale in analisi economica delle istituzioni internazionali
A.A.: 2011-2012

I contenuti di questa pagina costituiscono rielaborazioni personali del Publisher Atreyu di informazioni apprese con la frequenza delle lezioni di Differenziali Economici e Migrazioni e studio autonomo di eventuali libri di riferimento in preparazione dell'esame finale o della tesi. Non devono intendersi come materiale ufficiale dell'università La Sapienza - Uniroma1 o del prof Giacomello Paola.

Acquista con carta o conto PayPal

Scarica il file tutte le volte che vuoi

Paga con un conto PayPal per usufruire della garanzia Soddisfatto o rimborsato

Ti è piaciuto questo appunto? Valutalo!

Altri appunti di Differenziali economici e migrazioni

Europa e flussi migratori
Immigrazione straniera in Italia - Dati ufficiali
Condizioni della popolazione straniera in Italia
Ruolo economico degli immigrati in Italia