International Journal of Educational Development 40 (2015) 106–116

Transnational higher education and skilled migration:Evidence from Australia

Antonina Levatino *

Department of Political and Social Sciences, Universitat Pompeu Fabra, Spain

A R T I C L E I N F O

Keywords:

International education

Globalisation of higher education

Offshore enrolment

Transnational education

Skilled migration

Australia

A B S T R A C T

This paper presents empirical evidence regarding the relationship between enrolment in transnational

higher education (TNE) and skilled migration into the country of the institution which provides

educational services. Based on macro-level panel data, the analysis shows a link between skilled

immigration and offshore enrolment in Australian higher education within the previous years,

suggesting that the provision of higher education offshore can constitute a successful strategy to enlarge

skilled migrants’ recruitment. The results also indicate that more caution should be devoted to this kind

of issue by developing countries when opening their education market to foreign providers.

� 2014 Elsevier Ltd. All rights reserved.

Contents lists available at ScienceDirect

International Journal of Educational Development

jo ur n al ho m ep ag e: ww w.els evier . c om / lo cat e/ i jed u d ev

1. Introduction

Skilled migration has been long considered a major issuebecause of the crucial importance of skilled people for thesocioeconomic development of a country. After a period of absencein scholars’ and policymakers’ agendas, the topic received renewedattention in the nineties, essentially due to developed economies’growing interest in attracting skilled individuals from abroad. Formany OECD countries, the continuous ageing of the populationcauses reason to worry for the sustainability of national welfaresystems and has led them to make their immigration systems moreskill-selective and to adopt policies in order to actively attractskilled migrants.

In this context, international students are increasingly consid-ered an appreciated source of potential skilled immigrants for thecountries where they have studied: they hold recognizablequalifications from familiar educational entities (Kuptsch, 2006)and they are already integrated into the host society, whoselanguage they know and where they have already establishedsocial ties. Due to reasons such as these, several OECD countrieshave eased their immigration policies to allow internationalstudents to remain in the host country after having completedtheir studies, as well as measures to increasingly attract students

* Correspondence to: Universitat Pompeu Fabra, Jaume I Building, Office: 20.135,

Ramon Trias Fargas 25–27, 08005 Barcelona, Spain. Tel.: +34 93 542 2505;

fax: +34 93 542 2372.

E-mail address: antonina.levatino@upf.edu

http://dx.doi.org/10.1016/j.ijedudev.2014.11.009

0738-0593/� 2014 Elsevier Ltd. All rights reserved.

from abroad. Recent research on student mobility and skilledmigration has effectively shown how international students arelikely to stay in the country where they studied after havingcompleted their studies (e.g. Dreher and Poutvaara, 2011;Rosenzweig, 2008).

As student mobility often constitutes the first step of futureskilled migration, it is relevant to investigate how the new trendsin the higher education industry may reshape the current scenarioof skilled migration. Several higher education institutions of someOECD countries have indeed started to offer educational services‘offshore’ (i.e. in other countries) through the establishment ofbranch campuses and/or programmes abroad. The main reason toimplement transnational education (TNE) in a foreign country is toexpand recruitment, reaching people who cannot afford and/orhave no interest in emigrating to study. Interestingly, one of theprincipal motivations for governments opening their educationalmarket to foreign providers’ is instead to satisfy the demand forhigher education within their country and to retain studentoutflows and the potential consequent skilled emigration.

There is no consensus on the relationship between enrolment inTNE and skilled migration. From one side, it has been argued thatTNE enables countries that open their educational markets toforeign providers to retain and attract human capital from othercountries (OECD and World Bank, 2007; World Trade Organisation,2001). From the other, it has also been hypothesized that TNEmight increase the brain drain (Chiang, 2012; Lien and Wang,2012). Despite the interest devoted to the phenomenon of skilledmigration by policymakers and scholars, the continuous increaseof TNE and the importance of the issue for both developed and

A. Levatino / International Journal of Educational Development 40 (2015) 106–116 107

developing countries, the relationship between TNE and skilledmigration into the country of the educational services’ provider hasremained, until now, empirically understudied, due essentially tothe scarcity of data available. Using macro-level data, this paperaims at filling this gap, exploring the association betweenenrolment in Australian TNE and immigration of skilled individualsinto Australia.

The consideration of Australia as a destination countryconstitutes a noteworthy case study because of its interest inrecruiting skilled migrants, the strong international character of itshigher education system and the diffusion of its TNE, particularlyin Asia, i.e. the continent where TNE has known the biggest growth.

The remainder of this paper is organized as follows: Section 2offers a literature review, focused on theories that have tried toexplain the drivers of international migration and how previousresearch has addressed the relationship between student mobilityand skilled migration. It also briefly presents the recent discussionon how new trends in higher education could influence themobility of students and skilled individuals. In Section 3 the choiceof Australia as a case study is explained. Sections 4 and 5,respectively, present the methodology and the data used. Section 6shows the results after having presented some descriptiveevidence. It then discusses them and concludes.

2. Literature review

2.1. Determinants of international migration

Past research on international migration has put forwardseveral theoretical explanations to understand what drivesinternational migration.1 The neoclassical migration model mainlyconceives international migration as the result of differences inwage between countries (Harris and Todaro, 1970; Todaro, 1976).At a micro level, this theory considers migration as a rationalindividual decision made to maximize benefits, predicting thatbefore emigrating an individual evaluates costs and potentialbenefits related to it (Sjaastad, 1962). Costs related to migration areusually calculated based on the geographical distance betweencountries of origin and destination, colonial links, and linguisticproximity. The neoclassical theory of migration has normally beenmodelled using the ‘push-pull model’ and empirically implemen-ted using the gravity equation (Bessey, 2012; Karemera et al., 2000;Mayda, 2005). The new economics of migration (NELM) theoryexplains migration as a household strategy to get away fromrelative deprivation within sending societies (Stark and Taylor,1989). Both the neoclassical and the NELM theories explainmigration as a rational decision taken at a micro level (Masseyet al., 1993), a prevalent approach in the empirical literature onskilled migration.

Other theories focus on structural forces that operate at aninternational level. For example, the segmented labour markettheory (Piore, 1979) connects immigration to the developedcountries’ need for cheap immigrant workers (pull factors/demandside). Piore’s theory focuses on low skilled migrants, but it mayhave inspired some research on skilled migration. Several authorshave definitely explained increased skilled migration from theperspective of the demand side, by pointing to changes inimmigration policies introduced in main destination countries,which are increasingly selective and favour skilled workers (Abella,2006; Taran, 2007; Zeugin and Van Dok, 2007). Nevertheless, asnoted by Brucker et al. (2012: 170), the favourable admissionpolicies towards skilled workers ‘‘may not suffice to attract them’’.Another theory that focuses on macro level forces is the world

1 For exhaustive and comprehensive reviews of the theories on the determinants

of international migration, see De Haas (2011) and Massey et al. (1993).

system theory. According to it, international migration flowsfollow the dynamics of the international market and the structureof the world economy, and are a consequence of neoliberalglobalisation. In the framework of this theory, Sassen (1988: 119)states that in order to fully understand what mechanisms explainmigration it is necessary to research the ‘‘context of linkages’’ – thestructural and ideological linkages existing between sending andreceiving countries and in which way they have been established.

As argued by De Haas (2011) and Massey et al. (1993); differenttheories of migration can help to give light to the phenomenon ofmigration from different and complementary perspectives. Indeed,as highlighted by Massey et al. (1993: 433), it is completelyconceivable that ‘‘individuals act to maximize income, whilefamilies minimize risk’’. It is also quite reasonable that the need forworkers on the demand side influences the choices of potentialmigrants, although this impact is probably bound to the capabilityof obtaining information about visas, opportunities, and the jobmarket in destination countries. In this sense, the exploration ofSassen’s ‘context of linkages’ becomes fundamental.

Access to information, which depends on structural forces at amacro level, could be a way to facilitate the realization of anexisting aspiration to go abroad, but it can equally constitute theroot of the desire to emigrate. In such a given context, foreigneducation can be seen as one of the most significant informationchannels for young people.

2.2. Higher education, student mobility and skilled migration

The degradation of higher education systems in manydeveloping countries has been identified as one of the main ‘pushfactors’ causing people to emigrate for education abroad (Agarwaland Winkler, 1985; Mazzarol and Soutar, 2002; McMahon, 1992).The brain drain literature furthermore often points out that skilledemigration leads to the worsening of welfare and public services inorigin countries; this is especially worrying for the sustainability oftheir health and education sectors. Hence, as argued byDocquier and Rapoport (2012), the global competition fortalent could weaken the capacity of sending countries to investin public education. Literature on skilled migration as a braingain instead focuses on the potential positive impacts ofmigration prospects in promoting individuals’ skill acquisitionin the origin country, thus fostering human capital accumula-tion (Beine et al., 2011; Vidal, 1998).

Recent research has shown how the seed of skilled emigrationlies often in international student mobility, highlighting howinternational students are highly likely to stay in the host countryto work after having completed their studies there (Rosenzweig,2008; Tremblay, 2001). Dreher and Poutvaara (2011) showed howthe number of foreign students studying in the U.S. is a goodpredictor of skilled immigration, whereas Felbermayr andReczkowski (2012) determined an average ex-internationalstudents’ retention rate of about 70%. This seems to be due tothe fact that a foreign education increases the probability of findinga job in the training countries (Rosenzweig, 2005). It is howeverunclear if young people decide to study abroad to facilitate theirdesire to emigrate (which precedes the decision to study abroad) orif the decision not to return is mainly due to circumstances thatoccur during or after the period of study abroad. According toCoulon and Paivandi (2003: 45), international students often leavetheir home country with a ‘‘residential strategy’’; they studyabroad in order to stay there afterwards. Although it is quiteobvious that a strong link exists between the educational decision-making and employment purposes (Hashim, 2007), some scholarshave connected the non-return of ex-international students withthe effect of the so-called ‘second acculturation’ or ‘re-entrytransition’; i.e. the difficulties that many ex-international students

A. Levatino / International Journal of Educational Development 40 (2015) 106–116108

experience when readapting to their home culture after havinglived a long period abroad (Arthur, 2003; Viguier, 1966, 1968). Thedecision to stay abroad can be however also sometimes connectedto the fact that the training received in the host country is notadapted to the needs of the home country labour market.

2.3. TNE and skilled migration: the need for empirical evidence

As recently underlined by Becker (2012), with regards tostudent mobility as directly related to skilled migration, it isinteresting to introduce into the academic debate on these topics areflection on how and if TNE is related to skilled mobility. Due tothe difficulty in finding harmonized and comparable data, thewhole question has been empirically neglected. Hence, until now,the discussion has essentially remained speculative.

If we look at the perspective of countries that have opened theirmarket to foreign countries’ institutions, they mainly ‘‘aim atsatisfying the demand for higher education that was not being metby local existing providers’’ (Wilkins and Huisman, 2011: 300–301) and to enhance their higher education system’s quality. Indoing so, they can reduce skilled emigration and themselvesbecome a destination country for international students fromabroad (Wilkins and Huisman, 2011; Zhang, 2003). Theoretically, ithas been argued that TNE enables countries to foster the quality ofeducation institutions through a greater competition, a broadersupply and the internationalization of curricula, furthermorecreating new job opportunities and a more stimulating academicmilieu (Bashir, 2007; Kapur and Crowley, 2008). This will permitcountries to retain and attract human capital (OECD and WorldBank, 2007; World Trade Organisation, 2001). It has also beenhypothesized that TNE may contribute to the retention of homecountry students, who would have otherwise gone overseas andspent ‘‘money there’’, constituting a ‘‘less-worse option’’ (forcountries of origin) compared to students’ emigration (Kapur andCrowley, 2008: 30). Larsen and Vincent-Lancrin (2002) argue thatthe increase of TNE may reduce the risk of a brain drain becausestudents will be able to study without having to go abroad. If formany international students a reason for not returning is theproblematic re-adaptation to their home culture after a periodabroad (Viguier, 1966, 1968), the possibility to study in aninternational institution in the home country may diminish thepermanent emigration of tertiary educated people, thereforehelping countries to retain their human resources (Vincent-Lancrin, 2005). This kind of position essentially views TNE as away to retain students in the home country during and after theirstudies. They also seem to assume that international mobilestudents and transnational students mainly constitute the samesegment of clientele and that studying in the framework of TNEdoes not involve future emigration. In their article focused ondescribing the context of student mobility and skilled migration inMalaysia and Australia, Ziguras and Law (2006) note that peoplewho received their degrees from Australian education institutionsoffshore appear to migrate less into Australia and receive lesspoints in the admission’s system than those who studied directly inAustralia.

The perspective of higher education institutions from devel-oped countries opening campuses and programmes offshore isinterestingly different. They view TNE as a way to enlarge theirclientele (Ziguras, 2008), reaching people who do not have theaspiration and/or capability to study abroad. The InternationalEducation Advisory Council, an agency established by theAustralian government, discusses the issue of internationaleducation in a 2012 discussion paper, stating the following:

TNE offers students who are unable or unwilling to acquiretheir qualification in Australia access to a more diverse range ofquality programs than those available in their home country,

including the opportunity to undertake tuition in English(International Education Advisory Council 2012, 19; emphasismine).

If TNE is expanding international student recruitment, thenclaims that TNE is a good way to retain potential internationalstudents in the home country fail, meaning that the spread of TNEexposes more people to a foreign education experience. Stein(1996) remarks how the participation in international educationand training stimulates the interest of young people to workabroad. Chiang (2012) argues that TNE could constitute a ‘‘TrojanHorse’’ that might aggravate the brain drain, and furthermoreincrease the influence of western culture on East-Asia. Lien andWang (2012) provide an analytical model to examine the effectsof the presence of a foreign branch campus on the social welfareof the host country. According to their research, more studentswill end up immigrating into the foreign country, deepening thebrain drain problem. Due to the scarcity of actual data, the twoscholars were not able to test their assumptions and conclusions,which rely upon numerical simulations. It must also be pointedout that TNE programmes often entail stay periods in the countryof the educational institutions. In a survey-based study onstudents who had participated in exchange programmes, Kingand Ruiz-Gelices (2003) highlight how just a ‘‘year abroad’’ haseffects in subsequent migration behaviour: they found thatpeople who experienced the ‘‘year abroad’’ are almost twice aslikely to have migrated abroad and three times as likely to becurrent residents in a foreign country. According to Venturini(2012: 304), through their foreign student programmes, the U.S.,Canada, Australia and the U.K. generate educated workers thatare somewhat connected (at least through knowledge of thelanguage), to the country where they have studied. This,combined with their skill-selective immigration policies,increases skilled immigration. It is undeniable that TNE somehowconnects people with the country of the higher educationinstitutions attended. Additionally, through transnational pro-vision, qualifications become international and that increases thecapability of their holders to move (Vincent-Lancrin, 2008). In thecase of Australia, for example, immigrants who have anAustralian degree from 1998 onwards received extra points inthe point systems (Ziguras and Law, 2006). This kind of positionassumes that TNE is expanding recruitment and that theparticipation in foreign education is related to skilled mobility.They therefore hypothesize that an increase in TNE correspondsto an increase of skilled immigration to the country of theinstitution which provided educational services offshore.

As outlined by the above, the debate on TNE and its associationwith skilled mobility is characterized by the existence of manydifferent arguments which rely upon heterogenous assumptionsand by a scarcity of empirical evidence. This paper constitutes oneof the first attempts to provide an empirical analysis of therelationship between enrolment in offshore education and skilledmigration into the country of the education provider.

3. Case selection

Data on enrolment in TNE are exclusively available on anaggregated level. Only two countries, the UK and Australia,systematically collect data on students enrolled in TNE withintheir institutions operating offshore. Nevertheless, the differencesbetween the measurements and the variables in the two datasetsprevent a comparison of them. UK data, provided by the HigherEducation Statistics Agency (HESA), makes a distinction betweenenrolment in programmes delivered in the framework of apartnership (franchising, twinning arrangement, etc.) and enrol-ment in UK branch campuses. This distinction is not made by

4 As the Australian Bureau of Statistics (2009) indicates, a very high proportion of

onshore skilled visa applicants (45% in the period 1997/98–2007/08) are ex-

A. Levatino / International Journal of Educational Development 40 (2015) 106–116 109

Australian data, provided by the Department of Industry, Innova-tion, Science, Research and Tertiary Education (DIISRTE) of theAustralian Government2. Moreover, Australian data give informa-tion about enrolment by country of birth, whereas UK data onlytells us how many people are studying in every country withoutreference to the country of origin of students. Australian data arethus preferable because they allow us to evaluate the impact offactors associated with the country of origin of the students.Furthermore, they cover a longer period of time, being availablefrom 2002, whereas UK data have only been collected from 2007.

The case of Australia is remarkable for the aims of this paper.Together with Canada and New Zealand, Australia can beconsidered a pioneer in the selection of migrants. It is also amongthe six main destination countries of skilled migrants. Moreover,the Australian higher education sector has one of the mostinternationalized enrolments in the world (Hall and Hooper, 2008;Nelson, 2003: 35; OECD, 2004, 2009). Australian universities haveknown in the recent decades an unprecedented growth ofinternational student enrolments, especially in offshore pro-grammes (IDP Education Australia, 2002). Offshore enrolmentsin Australian universities have more than tripled from 1996 to2008, constituting approximately one-third of the total ofinternational enrolments in Australian universities in 2008 (IDPEducation Australia, 2008). Finally, the presence of Australian TNEis particularly important in Asia, i.e. the continent with the biggestsupply of TNE.

4. Methodology

Inspired by previous research (Bessey, 2012; Docquier andRapoport, 2012; Karemera et al., 2000; Lewer and Van den Berg,2008; Mayda, 2005), skilled migration from one country of origin(i) to Australia in a year (t) is modelled here as function of severalvariables, which aim at controlling for economic, demographic,legal and institutional factors which may affect the volume ofskilled migration.

Considering the recent developments in econometric literature(Santos Silva and Tenreyro, 2006, 2011), the relationship of interestis estimated using a multiplicative model, using the Poissonpseudo-maximum-likelihood (PPML) estimation technique, amodel derived from count data analysis.3

This estimator overcomes the methodological challengesconnected with the logarithmic specification, which has longerrepresented the standard approach in the analysis of trade andmigration flows. One of the biggest problems related to thelogarithmic specification is the fact that it cannot deal with zerosbecause the logarithm of zero in undefined. Thus, the observationswith zero values are dropped and omitted from the calculations. Intrade and migration flows, the dependent variable is zero for alarge number of observations. Omitting them can be considered aloss of important information and a source of selection bias due tothe fact that zero values are not randomly distributed (Burger et al.,2009; Eichengreen and Irwin, 1995). As pointed out by Santos Silvaand Tenreyro (2006), the PPML estimator constitutes a solution forthe problem of the presence of zero values in the dependentvariable because the dependent variable in this model does notenter in log. Another problem of the logarithmic specification isrelated to the assumption made on homoscedasticity, i.e. theassumption that the error terms all have the same variance. Through

2 Since August-September 2013, Higher education section moved from DIIRSTE

to the Department of Education in line with the Australia’s machinery of

government (MOG) changes.3 For a detailed presentation of the PPML estimator, see the ‘‘The Log of gravity’’

page by Santos Silva and Tenreyro: http://privatewww.essex.ac.uk/�jmcss/LGW.

html (last accessed July 25, 2014).

a set of Monte Carlo simulations, Santos Silva and Tenreyro (2006:641) demonstrate that the PPML estimator is ‘‘robust to the differentpatterns of heteroscedasticity’’. Furthermore, it has been found togive consistent results even when the distribution is not rigorouslyPoisson, and is used more frequently in the literature (Beine et al.,2011; Beine and Parsons, 2013; Bessey, 2012).

Hence, following Wooldridge (1999), a panel data analysis isconducted for estimating a fixed-effects (FE) Poisson pseudo-maximum likelihood regression with robust standard errors. Thebasis of the empirical analysis is Eq. (1) below:

skillimmit ¼ exp½b0 þ b1 lnðO f fshoreenrolit�tÞ

þ b2 lnðGDP per ca pitaitÞ þ b3 lnðDemogitÞ

þ b4ðBilagreeitÞ þ b5 lnðUnem plitÞ þ ai þ g t� þ eit (1)

Skilled immigration into Australia (skillimmit), is defined as thenumber of people from country i who are granted an Australianskilled migration visa offshore. Counting only visas grantedoffshore can be considered a reasonably accurate measure ofskilled immigration because it excludes from the estimations ex-international students already living in Australia who changestatus from a student visa to a skilled migrant visa onshore.4 In thisway, it is assured that the individuals counted are already ‘skilled’before entry into Australia and they were not living there in theperiod that precedes the grant of the visa.5

The main variable of interest is Offshoreenrolit�t, whichmeasures the stock of students from country i enrolled inAustralian offshore higher education at a period of time sufficientlyfar from t in the past (t � t) so that the current stock of students in t

is not accounted for in the stock of skilled immigrants. From thetotal enrolment, the data used permits the exclusion of first yearstudents, because it is unlikely that they are eligible for a visa basedon qualifications after only one year of commencing their studies.Actually, higher education can be considered the process throughwhich skills are acquired and a time (at least two or three years)must elapse so that the observed values of this explanatoryvariable can result in a change in the dependent variable. It isfundamental to remark that the main explanatory variable ofinterest is not the presence of TNE in one country (which surely isnot random), but rather the stock of people from whichevercountry (also from countries where no Australian TNE is available)enrolled in Australian TNE. More and more people are indeedenrolled in TNE in countries other than their own. Consequently,the analysis conducted includes also countries where no AustralianTNE is available.

The model contains the following controls. Following the macromigration literature (Dreher and Poutvaara, 2011; Hatton, 1995;Pedersen et al., 2008) GDPper capitait is added. The stock ofpotential skilled emigrants (Demogit) is approximated by a variablewhich controls for the total number of people enrolled in tertiaryeducation. This measure can surely be considered quite a truthfulway to control the stock of potential skilled immigrants intoAustralia, who are usually fairly young, especially for thecategories of skilled visas considered here. In fact, Australianimmigration policies clearly favour the entrance of immigrants intheir twenties and early thirties.

international students.5 It is possible that the measure includes ex-foreign students that have come back

to their country of origin and immigrate back to Australia for a second time.

Nevertheless, as underlined by the Australian Bureau of Statistics (2009), skilled

arrivals are normally quite young (within the decade of 1997/98–2007/08, 57% of

arrivals were aged 29 years old and under); this suggests the improbability of this

migration pattern.

A. Levatino / International Journal of Educational Development 40 (2015) 106–116110

One of the originalities of the model is the introduction of thevariable Bilagreeit which accounts for the existence of bilateralagreements which can foster or facilitate the mobility of workers.Specifically, the variable controls for the existence of SocialSecurity Agreements between countries of origin and Australia.Under these agreements, residents of Australia and of countriesthat stipulate them will be able to move from one signing countryto another with the knowledge that their right to social benefits isrecognized in both countries. The relevance of Social SecurityAgreements that allow the portability of social security’s rights inaffecting cross-border workers’ mobility has been highlighted inthe literature (Bertoli et al., 2013; Bertoli and Fernandez-HuertasMoraga, 2013; Pasadilla, 2011). This kind of opportunity favours amore positive attitude of people towards labour migration into aparticular country. It is thus expected that the introduction of anagreement of such type is positively related with workers’mobility. Indubitably, it is equally important to control for thepresence of agreements which easily enable overseas educationalqualifications to be recognized. In Australia there is no singleauthority that assesses or recognizes all overseas qualifications.However, Australia has signed agreements of mutual recognitionof qualifications with some countries. All the agreements of thiskind have been signed prior to 2003, the starting year of theanalyses. Thus, their effect is constant over the whole periodconsidered. Hence, origin countries’ fixed effects control for this.

As Docquier and Rapoport (2012: 285) note, ‘‘some skilledindividuals emigrate because they would not find a job or would beineffectively employed in their countries of origin’’. A furthercontrol which measures the unemployment rate will be thereforeincluded, because unemployment has usually been considered inthe literature as one of the main macroeconomic determinants ofmigration (Harris and Todaro, 1970; Hatton, 1995; Van Hear et al.,2012). The introduction of this variable causes many observationsto be dropped because of many missing values; therefore it will beconsidered at a later point in the analysis.

ai is a country specific intercept (origin country‘s fixed effects).Countries of origin are quite different from one another, thereforeit is fundamental to consider the existence of differences in theanalytic model.6 The use of countries’ fixed effects controls fortime-invariant characteristics related with culture, climate,geography and other factors relevant for migration patterns. As Iconsider only one country of destination, ai also captures country-specific economic and cultural links with Australia, such as thesame commonwealth membership, common language and geo-graphical distance. Thus, the use of these kinds of fixed effectsmakes it possible to account for all the time-invariant character-istics of one country of origin which can influence both the numberof students enrolled in Australian TNE and the number of peopleimmigrating as skilled migrants into Australia. The influentialtime-invariant factors might also include the Australian recogni-tion of degrees from a particular country, a particularly friendlyattitude of Australia towards skilled immigrants from a specificcountry7 and other time-invariant unobserved factors. Thus, the

6 It would be worthwhile to have the possibility to add a variable to control origin

countries’ TNE policies. To my knowledge, any dataset is available where such

information is available for every country in the world. It should however be

pointed out that where Australian TNE is more present (e.g. Singapore, Malaysia,

etc), this phenomenon had begun and was promoted before the period considered

in these analyses, so that countries’ fixed effects may control for this issue in a

reasonable manner.7 It can be assumed that, for the time period considered in this analysis, fixed

effects also control for the stock of previous migrants in a reasonable way. In fact, as

recently remarked by Beine et al. (2013: 26) ‘‘at the annual frequency, migration

stocks are quite stable over time’’. Besides, I control for the existence of bilateral

agreements, which is likely to be correlated with the migration stocks in Australia,

as highlighted by Beine et al. (2013). So that part of the effect associated to the

‘migrants networks’ is controlled for by that variable.

use of countries of origin’s fixed effects controls for the so-called‘‘unobserved heterogeneity’’.

The equation also includes gt that controls for time specificity(time fixed effects) which are common for all countries, but differsfrom one year to another. This is quite important giving the datacycle taken into consideration for this research. The time spanconsidered actually includes the so-called ‘‘Great Recession’’, i.e.the global economic downturn. The inclusion of time fixed effectsallows the equation to control for its influence. As the analysisconcerns only one country of destination, time fixed effects alsocontrol for changes in Australian immigration policies, such as, forexample, changes in the total number of skilled workers admittedinto the country.

Finally, I also provide the results of the estimations given byEq. (2), because in the literature on international migration, thisapproach has longer represented the standard way of estimation:

lnðskillimmitÞ ¼ b0 þ b1 lnðO f fshoreenrolit�tÞ

þ b2 lnðGDP per ca pitaitÞ þ b3 lnðDemogitÞ

þ b4ðBilagreeitÞ þ b5 lnðUnem plitÞ þ ai þ g t

þ eit (2)

However, it is important to reiterate how recent research bySantos Silva and Tenreyro (2006, 2011) raises serious concernsabout biases originated by this specification due to the logarithmictransformation of the dependent variable, the way zero values aretreated and the failure of the assumption of homoscedasticity.

5. Data

The panel used in this research covers a 10-year period (2002–2011). From the countries of origin in the dataset, I explicitlyexclude New Zealand, because this country and Australia have hadspecial arrangements in place since the twenties to facilitate a freeflow of people between them. In this way, New Zealand citizens arenot counted in the data on visas granted.

Data on skilled immigration to Australia are provided by theAustralian Department of Immigration and Citizenship (DIAC) andconcern all skilled visas granted offshore.8 Skilled immigrants canenter Australia through two different schemes: the ‘General Skilled

Migration (GSM)’ scheme and the ‘Employer Nomination (ENS)’scheme. Both include a wide range of permanent and temporaryvisa subclasses to be granted to individuals with an educationaldegree. The other two small categories of skilled visa (Business

Skilled Visa and Distinguished Talent Visa) are explicitly excludedfrom the estimations because they are granted to particularcategories of individuals and do not necessarily require anytertiary education.

Data on offshore and onshore enrolment in Australian highereducation institutions have been provided for each country oforigin of the students by the Australian DIIRSTE. Enrolment in non-award courses, i.e. courses which do not lead to a qualification, isexplicitly excluded.

8 Data on visa granted offshore are collected by country of citizenship whereas

the main independent variable is collected by country of origin. An alternative to

using data on visas would be to use Census data. However, this is also far from

perfect because ‘‘the information about entry year is often based on answers to an

ambiguous question’’ (Rosenzweig, 2005: 9). From DIAC it is possible to get data on

skilled visas granted by country of birth for the considered period. Nonetheless, the

information by country of birth concerns the total number of visas conceded

(offshore and onshore). I compared the two datasets (by country of birth and by

country of citizenship), and the correlation between the total number of skilled

visas granted by country of birth and citizenship is remarkably high (0.9441),

demonstrating that the use of the information on skilled visas granted offshore by

country of citizenship is not problematic. The information available gives the

number of skilled visas granted from June of one year to June of the following one.

0

20.000

40.000

60.000

80.000

100.000

120.000

140.000

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Fig. 1. Total number of skilled visa grants, Australia (2002/03 to 2011/12).

Source: Data from DIAC, December 2012.

Table 1List of variables.

Variable Source Definition

Total offshore skilled visa Australian Department of Immigration and Citizenship,

December 2012

Total offshore visas granted trough the skilled migration stream

(Skilled Independent, Employer Sponsored, State Sponsored,

and Skilled Australian Sponsored) by country of citizenship. It

excludes Business Skilled and Distinguished Talent Visas.

Observations with no clear identification of the country are

dropped. In the original dataset, counts less than 5 are indicated

as <5. This is why these counts are substituted with 3

Total offshore enrolment Selected Higher Education Statistics, Department of

Industry, Innovation, Science, Research and Tertiary

Education (Australia)

Total number of students enrolled in Australian higher

education institutions and/or programmes offshore by country

of birth, excluding first-year students, in t � 1 and in

t � 2. Dependent territories are considered jointly with the

controlling state. Observations with no clear identification of

the country are dropped. In the original dataset, counts less

than 5 are indicated as <5. This is why these counts are

substituted with 3

GDP per capita World Bank World Development Indicators Gross domestic product divided by midyear population. Data

are in constant U.S. dollars

Enrolment in tertiary education UNESCO/UIS Stock of people enrolled in total tertiary education. Missing

values are linearly interpolated

Unemployment World Bank Unemployment refers to the share of the labour force that is

without work but available for and seeking employment.

Definitions of labour force and unemployment can differ by

country

Bilateral agreement Australian Treaties Database (ATD) of the Department of

Foreign Affairs and Trade of the Australian Government

Dummy variable = 1 if there is a bilateral Social Security

Agreement between the country of origin and Australia

A. Levatino / International Journal of Educational Development 40 (2015) 106–116 111

Concerning the other variables, I take data on GDP per capitaand on unemployment rates from the World Bank DevelopmentIndicators. Information on tertiary enrolment is taken from theUNESCO/UIS database. The variable that controls for bilateralagreement is a dummy variable (=1) if there is a bilateral SocialSecurity Agreement between the country of origin and Australia.Information comes from the Australian Treaties Database (ATD) ofthe Department of Foreign Affairs and Trade of the Australian

2002 2003 2004 2005 2006

Off shore 50428 57034 64864 66347 66476

Onshore 844.880 872.820 880.030 892.688 914.081

Fig. 2. Total enrolment in Australian higher education, 2002–2011.

Source: Data from Selected Higher Education Statistics, Australia DIISRTE.

Government. All variables, their precise definition and data sourcesare shown in Table 1.

6. Results

6.1. Descriptive analysis

The annual inflow of skilled migrants into Australia has notablyincreased during the last years. The total number of skilled visasgranted (both offshore and onshore) has nearly doubled from2002/2003 to 2011/2012 (Fig. 1).

In ten years, only 16 countries of origin of skilled migrants arepresent in the top-ten lists, suggesting a high impact of time-invariant characteristics in shaping skilled migration flows. Thisconfirms that an analysis on changes within countries is morepertinent than an analysis of differences between countries.

Checking if the increase in offshore enrolment corresponds to adecrease in onshore enrolment is noteworthy. As Fig. 2 shows, inthe period considered, both offshore and onshore enrolment inAustralian institutions grew.

Fig. 3 shows the geographical distribution of offshore enrol-ment in Australian higher education in 2002 by country of birth. Inthat year, students enrolled in Australian TNE are mainly from Asia,most likely enrolled directly in their countries of origin.

The total enrolment in Australian TNE grew from 50.428enrolments in 2002 to 80.962 in 2011. Fig. 4 shows thegeographical distribution of offshore enrolment in Australian

2007 2008 2009 2010 2011% Growth

2002-2011

68453 68843 73846 76892 80962 60.55%

956.997 992.986 1.056.701 1.114.794 1.139.830 34,91%

Fig. 3. Total enrolment in Australian TNE by country of birth, 2002.

Source: Data from Selected Higher Education Statistics, Australia DIISRTE.

Fig. 4. Total enrolment in Australian TNE by country of birth, 2011.

Source: Data from Selected Higher Education Statistics, Australia DIISRTE.

A. Levatino / International Journal of Educational Development 40 (2015) 106–116112

higher education by country of birth in 2011. Asia remains thecontinent with the most enrolment by country of birth, but thenumber of students from Africa, Middle East and North Americabecame larger.

Comparing this map with that for 2002, it is possible toremark how the number of countries with more than 1000students enrolled offshore in Australian higher educationconsiderably increased and how the origin’s composition ofstudents enrolled became more varied in the period considered.If from one side, this reflects the enlargement of the presence ofAustralian TNE worldwide, then the number of students comingfrom countries where no Australian TNE is available is similarly

Table 2Descriptive statistics.

Variable Min. Max.

Total offshore skilled visa 0 19934

Log of total offshore skilled visa 1.098612 9.90018

Log of total offshore skilled visa +1 0 9.90023

Log of total offshore enrolment, t � 1 0 9.37881

Log of total offshore enrolment, t � 2 0 9.33397

Log of GDP per capita 4.652165 12.1348

Log of enrolment in tertiary education .6931472 17.2594

Log of unemployment �2.302585 4.08597

Bilateral agreement 0 1

growing. Actually, enrolment does not exclusively concernpeople coming from countries where Australian higher educa-tion is offered.

If the majority of people enrol in their countries of origin, theshare of people enrolled in a country other than their country oforigin is increasing as well. TNE seems to entail new patterns ofstudent mobility towards countries where the opportunity tostudy TNE is available. TNE can actually constitute a less costlytype of student migration for people who do not have thepossibility to study directly in Australia.

The descriptive statistics of all the variables included in theanalyses is provided in Table 2.

Mean Std. Dev. (overall) No. of Obs.

211.0314 1163.696 2450

2 3.674214 2.009216 1250

2 1.915321 2.330401 2450

7 2.783036 2.008 1076

3 2.755733 2.001477 947

1 8.254849 1.651877 1918

11.52491 2.422913 1489

6 1.9786 .689847 1037

.0469388 .2115509 2450

Table 3PPML specification.

(1) (2) (3) (4)

Total offshore enrolment,

t � 1 (log)

0.164***

(0.04)

0.154***

(0.04)

0.078

(0.06)

0.060

(0.06)

Total offshore enrolment,

t � 2 (log)

0.149*

(0.06)

0.156**

(0.06)

GDP per capita (log) 0.223

(0.17)

0.255

(0.17)

0.192

(0.16)

0.224

(0.16)

Enrolment in tertiary

education (log)

0.236

(0.20)

0.289

(0.21)

0.532**

(0.21)

0.607**

(0.20)

Bilateral agreement 0.352**

(0.12)

0.384**

(0.14)

Constant 1.012

(3.72)

0.557

(2.58)

�4.442

(3.53)

�8.343**

(2.70)

Country FE YES YES YES YES

Time FE YES YES YES YES

# Observations 662 662 514 514

Pseudo-R2 0.977 0.978 0.984 0.986

Dependent variable: total skilled visas granted offshore. Robust standard errors are

indicated in parentheses.* p � 0.05.** p � 0.01.*** p � 0.001.

Table 4PPML specification, augmented equation.

(5) (6)

Total offshore enrolment, t � 1 (log) 0.282***

(0.05)

0.152**

(0.06)

Total offshore enrolment, t � 2 (log) 0.253***

(0.06)

GDP per capita (log) 0.750***

(0.18)

0.744***

(0.17)

Enrolment in tertiary education (log) �0.508

(0.33)

�0.023

(0.25)

Unemployment (log) 0.424**

(0.15)

0.514**

(0.16)

Bilateral agreement 0.272**

(0.11)

0.296*

(0.13)

Constant 9.490

(5.58)

0.510

(4.33)

Country FE YES YES

Time FE YES YES

# Observations 446 340

Pseudo-R2 0.984 0.991

Dependent variable: total skilled visas granted offshore. Robust Standard errors are

indicated in parentheses.* p � 0.05.** p � 0.01.*** p � 0.001.

9 I also estimated the model substituting this variable with the stock of

educated people and the results are very similar. Data on stocks of educated

people come from the dataset by Barro and Lee (2010) on educational attainment

for members of the population aged 15 and over. The dataset provides

information only every 5 years, thus the values for the remaining years have

been interpolated. The variable on tertiary enrolment is here preferred because

its use requires less interpolation.

A. Levatino / International Journal of Educational Development 40 (2015) 106–116 113

6.2. Panel data analysis

This section presents a panel data analysis focused on therelationship between skilled immigration into Australia andenrolment in Australian TNE one and two years before.

Table 3 presents the results obtained using the PPML estimator.In the first two columns, only one lag of the main independentvariable is added, whereas columns 3 and 4 presents the resultsobtained with the introduction of two lagged values of theindependent variable of main interest. The introduction of thevariable controlling for the stock of students enrolled in AustralianTNE in t – 2 implies the drop of many observations, thus the resultsconcern two different subsamples and are therefore not totallycomparable.

Columns 2 and 4 show the results of an augmented version ofthe equation estimated in columns 1 and 3, adding a variablecontrolling for the presence of Social Security bilateral Agree-ments.

The stock of students enrolled in Australian higher education inprevious years is found to have a positive and significantassociation with subsequent skilled immigration into Australiain all the estimations. Although the dependent variable here isspecified as the total number of skilled visas granted offshore (andnot as a logarithm), the coefficients of the independent variablesentered into logarithms (such as the variable of my interest) canstill be interpreted as elasticities. The results in columns 1 and 2show that a 10% increase in the number of enrolments inAustralian TNE from individuals from country i at time t – 1 isassociated with a positive increase (between 1.5% and 1.6%) in theskilled visas granted to individuals from the same country to enterAustralia in the subsequent year.

As expected, the introduction of social security bilateralagreements is positively related with skilled migration with astatistically significant coefficient. The variable that controls forGDP per capita seems also to be positively associated with thedependent variable. The result suggests that outmigrationincreases with the income per capita in the origin countries. Thisis quite understandable as the level of wealth in one countryinfluences both the capability of people to acquire skills and toafford emigration. The result is however not significant. Thenumber of people enrolled in tertiary education, introduced as a

proxy of the stock of young skilled individuals in the country, ispositively related with skilled migration.9

I further test for the robustness of the results with respect to theinclusion of an additional variable which controls for the rate ofunemployment in the country of origin. An increase in theunemployment rate in one country is likely to be related to moremigration. The results obtained with the introduction of thisadditional variable with the PPML estimator are shown in Table 4.

An increase in the rates of unemployment in the country oforigin is significantly associated with a growth in skilled migration.Most surprisingly, the inclusion of this variable has no impact onthe sign of the relationship between enrolment in Australian TNEand subsequent skilled immigration into Australia; rather, indeedthe coefficients here are quantitatively higher and statisticallymore significant.

I also provide the results of the logarithmic specification. Asexplained above, this approach longer represented the standardway of dealing with gravity equations. I run two set of regressionswith the inclusion of country and time dummies (FE) andcalculating robust standard errors. The dependent variable is thenatural logarithm of total offshore visas granted in the first set ofregressions (Table 5; columns 7 and 8) and of total offshore visasgranted +1, as a common way to deal with zero values on thedependent variable, in the second one (Table 5; columns 9 and 10).

The results of the logarithmic specification broadly confirm thefindings of the PPML for the independent variable of interest. Thepositive relationship found between enrolment in Australian TNEand subsequent skilled immigration into Australia is robustthroughout the different estimations presented. Such outcomesuggests that if TNE is effectively expanding recruitment, it canrepresent a good way for countries that place their highereducation offshore to enlarge their ‘pool’ of skilled individualsto recruit.

Table 5Logarithmic specification.

(7) (8) (9) (10)

Total offshore enrolment,

t � 1 (log)

0.103***

(0.03)

0.079*

(0.04)

0.104***

(0.03)

0.077*

(0.04)

Total offshore enrolment,

t � 2 (log)

0.023

(0.04)

0.062

(0.03)

GDP per capita (log) 0.242

(0.15)

0.396*

(0.18)

0.416*

(0.18)

0.610**

(0.22)

Enrolment in tertiary

education (log)

0.174

(0.12)

0.032

(0.17)

0.173

(0.12)

0.139

(0.17)

Bilateral agreement 0.081

(0.11)

�0.023

(0.13)

0.097

(0.11)

0.043

(0.13)

Constant �2.526

(1.65)

�1.831

(2.78)

�6.223**

(2.26)

�5.766

(3.15)

Country FE YES YES YES YES

Time FE YES YES YES YES

# Observations 618 482 755 564

Pseudo-R2 0.971 0.976 0.968 0.972

Dependent variable:

Columns (7) and (8); Log of (total skilled visas granted offshore).

Columns (9) and (10); Log of (total skilled visas granted offshore +1).

Robust Standard errors are indicated in parentheses.* p � 0.05.** p � 0.01.*** p � 0.001.

A. Levatino / International Journal of Educational Development 40 (2015) 106–116114

6.3. Discussion and concluding remarks

This paper analyses the relationship between enrolment in TNEand migration of skilled individuals. Using aggregated panel datafrom Australia, the analysis has been pursued by estimating an‘‘augmented version10’’ of the gravity equation following previousliterature on international labour migration.

Several scholars have theoretically speculated on the natureand sign of the relationship of interest, but no empirical analysishas been ever conducted until now because of the scarcity of data.The theoretical debate on the topic has been divided into two mainpositions. From one side, it has been claimed that enrolment in TNEconstitutes a substitute for student mobility and can thereforelower the outflows of skilled individuals that result from thisphenomenon assuming that people studying in the framework ofTNE do no migrate into the country of the educational providerafterwards. From the other, it has also been argued that TNE acts asa magnet for more people to be attracted into the country of theeducational provider after having completed the degree. Theresults of this analysis are more in line with this second position.They show a positive and statistically significant associationbetween skilled immigration into Australia and offshore enrol-ment in Australian higher education in the previous years.

In order to better explore the relationship found, theexplanatory variable ‘‘enrolment in Australia TNE’’ was taken intoaccount at a period of time prior to that of the dependent variableso that its changes can result in changes of the dependent variable(the students have to finish their studies before to be able to enterinto Australia as skilled migrants). Furthermore, the factors whichcan influence skilled migration (population size, GDP per capita,unemployment, social rights portability) were introduced into theanalysis as suggested by previous research. The panel structure ofthe dataset used enabled the employment of a full set of fixedeffects. This makes possible to account for all the time-invariantcharacteristics of one country of origin which can motivate boththe facts to have enrolment in Australian TNE and skilled migrationinto Australia. These include, for example, distance, historical and

10 In econometrics, a gravity equation is defined as ‘‘augmented’’ when variables

are added into the model to test whether they are relevant in explaining the

dependent variable.

economic strong relationships with Australia, the Australianrecognition of degrees from a particular country, a particularfriendly attitude of Australia towards skilled immigrants from aspecific country and other possible unobserved factors. From theother side, time fixed effects control for the specificity of aparticular year, taking into account, for example, the globaleconomic situation. As the analysis exclusively concerns migrationinto one country (Australia), this particular type of fixed effectsalso controls for Australian changes in immigration policies andregulations towards skilled migrants. The results obtained indicatethat an increase in enrolment in Australian TNE is associated withan increase of skilled migration into Australia in the followingyears. This could be due to the fact that TNE facilitates the creationof links and ties with people from the country of the educationprovider, furthermore permitting an easier access to informationabout its job market and legal framework. The educationalinstitution may also directly provide links with employers.According to a paper by Bagchi (2001) examining skilled migrationinto the United States, skilled peoples’ weak ties, with particularregards to potential employers, play a key role in explaining theirmigration, even more than contact with relatives and friendsabroad. Moreover, the qualifications obtained through TNE arerecognized in the higher education provider’s country and thiscould pave the way to future migration.

Certainly, one may argue that people enrol in TNE in order torealize a previous desire to immigrate into the country of theeducational provider, constituting an intermediate step towardsthe realization of an emigration. Even if this is the case, claimsregarding which TNE is a good way to retain the potential skilledemigration resulting from student mobility seem misleading.Indeed, even if TNE is a substitute of student mobility, the results ofthis analysis suggest that a large proportion of people who enrolledin TNE tend to immigrate into the country of the educationalprovider. In this case, enrolment in TNE is merely delaying theemigration. All this should be seriously taken in consideration bythose countries which open their educational market to foreignproviders with the hope of retaining skilled migration outflowsand/or to self-attract skilled individuals from abroad. Concerningthis last point, the relationship found suggests that, even if theymanage to attract foreign students, it is predictable that afterpursuing the degree in the third country, they will end upimmigrating to the country of the educational provider. Certainly,further research is needed to corroborate these findings, especiallyconsidering other national settings and longer periods of time.Future inquiry should also be directed towards testing andidentifying the truly causal nature of the relationship found here.

A number of questions on the possible mechanisms at workremain. It is possible that some students who obtain undergradu-ate education offshore may then move to Australia to attendpostgraduate programmes onshore. This additional possiblemechanism at stake could not have been taken into account inthe present paper due to the unavailability of data. Future researchshould consider this additional way of ex-TNE students toimmigrate into Australia which could mean a bigger effect ofenrolment in TNE in future skilled migration flows. Furtherresearch is similarly needed in order to analyze the relationshipbetween enrolment in TNE and the mobility of students, and tobetter understand who the students enrolling in TNE are and whatmotivations and aspirations they have. It seems hence necessary,and relevant, to build micro level surveys focused especially on thecomparison between students enrolled offshore and traditionalmobile students in order to understand their characteristics,motivations and attitudes towards migration. A study by Pyvis andChapman (2007) has also recently shown how the motivations ofstudents enrolled offshore in their own country and those enrolledoffshore in a third country can be very different. Moreover, in this

A. Levatino / International Journal of Educational Development 40 (2015) 106–116 115

paper, the focus is transnational higher education, but futureresearch may want to consider also transnational vocationaleducation and training (VET).

Within the limits of a macro level approach, the resultsconcerning the positive association between enrolment in TNE andskilled migration into Australia, taking into account only thoseindividuals who were granted visas outside the country, have areasonable magnitude and are quite reliable throughout differentestimation techniques. Additionally, the quality of Australian dataallowed for the measuring of skilled immigration in an accurateway, excluding migrants’ children and ex-international studentswho stay in Australia after having completed their studies, thuschanging visa-type (to a skilled migrant visa) directly onshore.Moreover, compared with Census information, normally used inresearch on skilled migration, data on visas granted assures morecertainty regarding the year of entry of the individuals counted.

Of course, this research is not exempt of shortcomings, mainlydue to the remarkable scarcity of data available. First of all, data onoffshore enrolment are available only from 2002, so the timelength considered for the analyses cannot be longer. Likewise, eventhough the case of Australia as a receiving country is highlyrelevant, the consideration of only one case as a destinationcountry prevents generalizations to be made. Still, if we comparethe official strategies towards student mobility’s policies and TNEof the major TNE exporting countries (for instance, US, UK andAustralia), we can remark how they have reasonably similar goalsand rationale.11 This could suggest similar outcomes. As a majorexporting country, Australia is unique in providing data onstudents enrolled in its higher education offshore based on theircountry of birth. If more data will be produced in other nationalsettings in the future, it would be useful to test the results of thisstudy with other countries that are very active in offshoring theirhigher education.

Besides, the consideration of an English-speaking country as acountry of destination does not allow for the disentangling of afurther important issue related to the phenomenon of TNE: i.e. therole language plays in the decision-making of students and offuture skilled migrants. It is a fact that TNE has been almostexclusively Anglophone to date and studying in English is perhapsone of the attractions of enrolling in it. This is a reason why it is notyet feasible to disentangle the role language has in such dynamics.This surely constitutes a valuable avenue for future research,especially considering the recent growing interest of non-Englishspeaking countries, such as Germany and France, in offshoringtheir higher education abroad, often through the implementationof programmes in English but with the firm goal of promoting theirnational languages as well.

This study can be considered a first step in the exploration of anissue of particular importance for both developed and developingcountries. The introduction of a variable measuring enrolment inTNE into the debate on skilled migration makes this study uniqueand hopefully it will inspire future research and the gathering ofimproved data on this issue.

Acknowledgements

The author would like to thank the Spanish Ministry ofEconomy and Competitiveness for funding her doctoral researchthrough a FPI (Formacion de Personal Investigador) grant. She wouldlike to thank Pau Baizan Munoz, Luis Ortız Gervasi, and theanonymous peer-reviewers for their constructive comments.Furthermore, she would like to express appreciation to all the

11 Such similarities are easy to find by comparing, for example, the UK’s

government strategy report (HM Government, 2013) with the discussion paper

elaborated by the Australian International Education Advisory Council (2012).

members of the International Migration Institute at the Universityof Oxford (especially to Mathias Czaika) for the feedback receivedduring the presentation of this research. She is especially gratefulto Bruno Arpino for valuable and stimulating support, and toChristopher Parsons for his valuable help with the elaboration ofthe maps and his much appreciated advice. Finally, she is thankfulto Natasha Cutts Andersen for the English editing and proofreadingthis paper.

References

Abella, M., 2006. Global Competition for skilled workers and consequences. In:Kuptsch, C., Pang, E. (Eds.), Competing for Global War for Talent. ILO, Geneva,pp. 11–32.

Agarwal, V.B., Winkler, D.R., 1985. Foreign demand for United States higher educa-tion: a study of developing countries in the eastern hemisphere. Econ. Dev. Cult.Change 33 (3), 623–644.

Arthur, N., 2003. Preparing international students for the re-entry transition. Can. J.Consel. 37 (3), 173–185.

Australian Bureau of Statistics, 2009. Perspectives on Migrants, 2009. Available at:http://www.abs.gov.au/AUSSTATS/abs@.nsf/Lookup/3416.0Main+Features32009 (last consulted 14.04.13).

Australian Education International, 2012. Transnational Education in the HigherEducation Sector. Available at: http://www.aei.gov.au/research/Research-Snapshots/Pages/default.aspx (last consulted 04.03.13).

Barro, R.J., Lee, J.W., 2010. A New Data Set of Educational Attainment in the World,1950–2010. NBER Working Paper No. 15902.

Bagchi, A.D., 2001. Migrant network and the immigrant professional: An analysis ofthe role of weak ties. Popul. Research and Policy Rev. 20 (1–2), 9–31, http://dx.doi.org/10.1023/A:1010608225985.

Bashir, S., 2007. Trends in International Higher Education: Implications and Optionsfor Developing Countries. Education Working Paper Series No. 6. World Bank, ,pp. 1–98.

Becker, S., 2012. Comments. In: Boeri, T., Brucker, H., Docquier, F., Rapoport, H.(Eds.), Brain Drain and Brain Gain: The Global Competition to Attract High-Skilled Migrants. Oxford University Press, Oxford, pp. 203–208.

Beine, M., Parsons, C., 2013. Climatic Factors as Determinants of InternationalMigration. IMI Working Papers Series No. 70. International Migration Institute,Oxford.

Beine, M., Docquier, F., Ozden, C., 2011. Diasporas. J. Dev. Econ. 95, 30–41.Beine, M., Burgeon, P., Bricongne, J.-C., 2013. Aggregate Fluctuations and Interna-

tional Migration. CESinfo Working Paper No. 4379.Bertoli, S., Fernandez-Huertas Moraga, J., 2013. Multilateral resistance to migration.

J. Dev. Econ. 102(C), 79–100.Bertoli, S., Fernandez-Huertas Moraga, J., Ortega, F., 2013. Crossing the border: self-

selection, earnings and individual migration decisions. J. Dev. Econ. 101(C),75–91.

Bessey, D., 2012. International student migration to Germany. Empir. Econ. 42 (1),345–361.

Brucker, H., Bertoli, S., Facchini, G., Mayda, A.M., Peri, G., 2012. Can the battle forbrains turn into a tragedy of the commons? In: Boeri, T., Brucker, H., Docquier,F., Rapoport, H. (Eds.), Brain Drain and Brain Gain: The Global Competition toAttract High-Skilled Migrants. Oxford University Press, Oxford, pp. 169–183.

Burger, M.J., Van Oort, F.G., Linders, G.-J.M., 2009. On the Specification of theGravity Model of Trade: Zeros, Excess Zeros and Zero-Inflated Estimation.Spatial Economic Analysis 4 (2), 167–190, http://dx.doi.org/10.1080/17421770902834327.

Chiang, L.-C., 2012. Trading on the west’s strength: the dilemmas of transnationalhigher education in East Asia. High. Educ. Policy 25, 171–189.

Coulon, A., Paivandi, A., 2003. Les etudiants etrangers en France: l’etat des savoirs.Rapport pour l’Observatoire de la Vie Etudiante. CRES, Universite de Paris 8,Paris.

De Haas, H., 2011. The Determinants of International Migration: ConceptualisingPolicy, Origin and Destination Effects. IMI Working Paper Series No. 32.

Docquier, F., Rapoport, H., 2012. Quantifying the impact of highly skilled emigrationon developing countries. In: Boeri, T., Brucker, H., Docquier, F., Rapoport, H.(Eds.), Brain Drain and Brain Gain: The Global Competition to Attract High-Skilled Migrants. Oxford University Press, Oxford, pp. 209–296.

Dreher, A., Poutvaara, P., 2011. Foreign students and migration to the United States.World Dev. 39 (8), 1294–1307.

Eichengreen, B., Irwin, D.A., 1995. Trade blocs, currency blocs and reorientation ofworld trade in the 1930. J. Int. Econ. 38, 1–24.

Felbermayr, G.J., Reczkowski, I., 2012. International Student Mobility and High-Skilled Migration: The Evidence. Ifo Working Paper Series No. 132, Munich.

Hall, G., Hooper, K., 2008. Australia’s Exports of Education Services. Reserve Bank ofAustralia, Bulletin June 12–17.

Harris, J.R., Todaro, M.P., 1970. Migration, unemployment, and development: a two-sector analysis. Am. Econ. Rev. 60, 126–142.

Hashim, I., 2007. Independent child migration and education in Ghana. Dev. Change38 (5), 911–931.

Hatton, T.J., 1995. A model of UK emigration, 1870–1913. Rev. Econ. Stat. 77,407–415.

A. Levatino / International Journal of Educational Development 40 (2015) 106–116116

HM Government, 2013. International Education: Global Growth and Prosperity.Department for Business, Innovation and Skills, London. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/340600/bis-13-1081-international-education-global-growth-and-prosperity-revised.pdf (last consulted 14.07.14).

IDP Education Australia, 2002. International Students Enrolled in Australian Uni-versities, 1994–2002. IDP Education Australia, Canberra.

IDP Education Australia, 2008. International Students at Australian Universities,Semester 1, 2008, National Overview. IDP Education Australia, Sydney.

International Education Advisory Council, 2012. Discussion Paper for the Develop-ment of an International Education Strategy for Australia. Available at: https://aei.gov.au/IEAC2/Consultation(IEAC)/Documents/InternationalEducationAdvisoryCouncilDiscussionPaper.pdf (last consulted02.06.14).

Kapur, D., Crowley, M., 2008. Beyond the ABCs: Higher Education and DevelopingCountries, Working Paper n. 139. Center for Global Development. Available at:http://media.sas.upenn.edu/casi/docs/research/papers/Kapur_Crowley_Higher_Ed.pdf (last consulted 07.03.13).

Karemera, D., Oguledo, V.I., Davis, B., 2000. A gravity model analysis of internationalmigration to North America. Appl. Econ. 32 (13), 1745–1755.

King, R., Ruiz-Gelices, E., 2003. International student migration and the European‘Year Abroad’: effects on European identity and subsequent migration behav-iour. Int. J. Popul. Geogr. 9 (39), 229–252.

Kuptsch, C., 2006. Students and talent flow – the case of Europe: from castle toharbour? In: Kuptsch, C., Pang, E.F. (Eds.), Competing for Global Talent.International Labour Office (ILO), Geneva, pp. 33–55.

Larsen, K., Vincent-Lancrin, S., 2002. International trade in educational services:good or bad? High. Educ. Manage. Policy 14 (3), 9–45.

Lewer, J.J., Van den Berg, H., 2008. A gravity model of immigration. Econ. Lett. 99 (1),164–167.

Lien, D., Wang, Y., 2012. The effect of a branch campus. Educ. Econ. 20 (4), 386–401.Massey, D.S., Arango, J., Hugo, G., Kouaouci, A., Pellegrino, A., Edward Taylor, J., 1993.

Theories of international migration: a review and appraisal. Popul. Dev. Rev. 19(3), 431–466.

Mayda, A.M., May 2005. International Migration: A Panel Data Analysis ofEconomic and Non-Economic Determinants, IZA DP No. 1590, DiscussionPaper No. 159.

Mazzarol, T., Soutar, G.N., 2002. Push–pull’ factors influencing international studentdestination choice. Int. J. Educ. Manage. 16 (2), 82–90.

McMahon, M.E., 1992. Higher education in a world market: an historical look at theglobal context of international study. High. Educ. 24 (4), 465–482.

Nelson, B., 2003. Higher Education. Report for 2003 to 2005 Triennium. Departmentof Education, Science and Training (DEST), Canberra. Available at: http://www.voced.edu.au/content/ngv35436 (last consulted 14.05.2014).

OECD, 2004. Education at a Glance. OECD Publishing, Paris.OECD, 2009. Higher Education to 2030: Globalisation, vol. 2. OECD Publishing, Paris.OECD and World Bank, 2007. Cross-Border Tertiary Education: A Way Towards

Capacity Development. OECD, Paris.Pasadilla, G.O., 2011. Social Security and Labour Migration in ASEAN. Asian Devel-

opment Bank Institute, Research Policy Brief No. 34.Pedersen, P.J., Pytlikova, M., Smith, N., 2008. Selection and network effects – migration

flows into OECD countries 1990–2000. Eur. Econ. Rev. 52 (7), 1160–1186.Piore, M., 1979. Birds of Passage: Migrant Labor in Industrial Societies. Cambridge

University Press, Cambridge.Pyvis, D., Chapman, A., 2007. Why university students choose an international

education: a case study in Malaysia. Int. J. Educ. Dev. 27, 235–346.Rosenzweig, M.R., 2005. Consequences of migration for developing countries. In:

Paper Prepared for the United Nations Expert Group Meeting on InternationalMigration and Development, Population Division, New York, July 6–8.

Rosenzweig, M., 2008. Higher education and international migration in Asia: braincirculation. In: Yifu Lin, J., Pleskovic, B. (Eds.), Annual World Bank Conference onDevelopment Economics Regional. World Bank, Washington, DC, pp. 59–100.

Santos Silva, J.M.C., Tenreyro, S., 2006. The log of gravity. Rev. Econ. Stat. 88 (4),641–658.

Santos Silva, J.M.C., Tenreyro, S., 2011. Further simulation evidence on the perfor-mance of the Poisson pseudo-maximum likelihood estimator. Econ. Lett. 112(2), 220–222.

Sassen, S., 1988. The Mobility of Labor and Capital: A Study in InternationalInvestment and Labor Flow. Cambridge University Press, Cambridge.

Sjaastad, L.A., 1962. The costs and returns of human migration. J. Polit. Econ. 70 (5),80–93.

Stark, O., Taylor, J.E., 1989. Relative deprivation and international migration.Demography 26, 1–14.

Stein, J.A., 1996. International Education and Training of Scientists and Engineers andtheir Employment in European Industry. PREST Report, European Commission.

Taran, P.A., 2007. Globalizacion y migraciones: desafıos para el desarrollo y lacohesion social. Revista Futuros, 20(IV). Quaestor, Madrid.

Todaro, M.P., 1976. International Migration in Developing Countries. InternationalLabor Office, Geneva.

Tremblay, K., 2001. Student mobility between and towards OECD countries in 2001:a comparative analysis. In: OECD (Eds.), International Mobility of the HighlySkilled. OECD-SOPEMI, Paris, pp. 39–67.

Van Hear, N., Bakewell, O., Long, K., 2012. Drivers of Migration. Migrating Out ofPoverty Research Programme Consortium. University of Sussex, Working PaperNo. 1.

Venturini, A., 2012. Comments. In: Boeri, T., Brucker, H., Docquier, F., Rapoport, H.(Eds.), Brain Drain and Brain Gain: The Global Competition to Attract High-Skilled Migrants. Oxford University Press, Oxford, pp. 302–307.

Vidal, J.P., 1998. The effect of Emigration on human capital formation. J. Popul. Econ.11 (4), 589–600.

Viguier, M.-C., 1966. La vie sociale des etudiants etrangers a Toulouse. (Unpub-lished doctoral dissertation)Universite de Toulouse, Toulouse.

Viguier, M.-C., 1968. La vie sociale des etudiants etrangers a Toulouse. Population 23(2), 370–371.

Vincent-Lancrin, S., 2005. Building Capacity Through Cross-Border Higher Educa-tion. Observatory on Borderless Higher Education, London.

Vincent-Lancrin, S., 2008. Student Mobility, Internationalization of Higher Educa-tion and Skilled Migration. World Migration 2008. International Organisationfor Migration (Chapter 4).

Wilkins, S., Huisman, J., 2011. Student recruitment at international branch cam-puses: can they compete in the global market? J. Stud. Int. Educ. 15 (3),3299–3316.

Wooldridge, J.M., 1999. Distribution-free estimation of some nonlinear panel datamodels. J. Econom. 90 (1), 77–97.

World Trade Organisation, 2001. GATS – Fact and Fiction. WTO, Geneva.Zeugin, B., Van Dok, G., 2007. Brain migration-Entwicklungspotenzial fur arme

Lander? In: Schweizerisches Rotes Kreuz (Eds.), Migration-ein Beitrag zurEntwicklung? Seismo Verlag, Zurich, pp. 71–103.

Zhang, C., 2003. Transnational Higher Education in China: Why has the Stateencouraged its Development? (Master’s dissertation)University of Stanford,School of Education, Stanford. Available at: http://searchworks.stanford.edu/view/7932593 (last consulted 02.02.14).

Ziguras, C., 2008. How many international students can Australia accommodate?Unpublished conference communication. In: ELT Management Conference,Swissotel, Sydney, 15 May.

Ziguras, C., Law, S.-F., 2006. Recruiting international students as skilled migrants:the global skills race as viewed from Australia and Malaysia. Glob. Soc. Educ. 4(1), 59–76.