Keywords

1 Demographic Transformations and Higher Education Recruitment Patterns

Romania is part of a number of countries in Central and Eastern Europe that experienced a sharp change in demographic trends in the 1990s. Fertility decline in the wake of the collapse of the communist regime kickstarted a process of rapid ageing in the general population, with the size of each annual cohort of new births shrinking sharply. For example, while annual births in 1990 were still as high as 315,000, they had fallen to just under 200,000 by 2019.Footnote 1

At the same time, the elimination of restrictions on higher education enrolment saw a boom in student numbers, with both traditional students (i.e. upper secondary graduates) and mature learners enrolling in numbers that would have been impossible under the communist regime’s strict numerus clausus.

Given that enrolment in higher education in Romania occurs at age 18 or 19, the fertility decline began to be felt in the higher education system after the 2008/2009 academic year. This initiated a two-stage demographic transition in higher education. The first stage was one of rapid contraction in student numbers, happening exactly 19 years after the abrupt decline in birth numbers which occurred in Romania in 1989–1991. After around 2013–14, a plateau emerged, with Romanian Statistics reporting a stable number of around 540.000 students throughout the following five years.Footnote 2 This can be described as a second stage, though in itself it represents merely a stop-gap at constant participation rates: demographic decline resumed at a slower pace after the mid-1990s.

The first stage of this transition, which essentially halved student numbers, represents one of the more seismic contextual shifts in contemporary European higher education. After going through massification starting in the 1990s, the dramatic collapse in the number of students attending universities was seen in almost all categories, but there were also dramatic changes in the shares of students by type of provider and funding (CNFIS 2013; Curaj et al. 2013).

In particular, the rapid transition of 2009–2014 upended funding patterns. By 2008, roughly 25% of the income of public universities came from tuition fees (CHEPS, 2010 as cited in Kwiek 2014). As student numbers fell after that year, the share of income derived from regular fees declined at public universities, with fee-paying student numbers halving over the next five years (CNFIS 2013; CNFIS 2014). Private institutions, where almost all places were fee-paying, lost some 80% of their students. State-funded study places, which were not reduced over the period, came to represent a majority of university study places. The once large distance-learning programmes offered by certain private and even public universities have seen a fall in enrolment. The number of long-distance and part-time students enrolled in private universities fell by 90% between 2008 and 2013 (CNFIS 2013, p. 9). This abrupt trend was bolstered by legal and reputational issues surrounding many of these programmes.

Coupled with stricter supervision of baccalaureate examinations, student numbers did not rebound over the coming decade. The private university sector never recovered its pre-2008 size, though the public sector also saw declines in student numbers. Overall, the public sector now dominates, though smaller universities struggle to attract sufficient students. In fact, some of the smaller universities have ended up being funded from non-regular channels. An example is the use of the Ministry of Education’s emergency funding stream to finance regular activities in some of the smaller, more vulnerable universities (Santa 2018). Nevertheless, after 2013, the number of students has broadly stabilised, partially influenced by a relative stability in both cohort size and access levels (Table 1).

Table 1 Decline in student numbers during the rapid 2008–2013 demographic transition
Full size table

As mentioned before, the demographic transition is ongoing. The demographic decline has continued after the 1990s, and while rising fertility rates limited the year-on-year (y-o-y) declines in cohort sizes, there is growing evidence of regional divergence in general demographic trends. These already became apparent at the 2011 census, with some regions that hitherto had a lower average age and relative population stability seeing steep falls in population numbers. And while the expected 2021 census has been postponed due to the Covid-19 global pandemic, there are growing signs that Romania’s demographic landscape is changing, with the population increasingly concentrating in a few regions, while others are seeing a steeper demographic decline. Unlike in the past, these trends seem to be largely influenced by economic considerations. For example, many North-Eastern counties are losing population via emigration (despite their higher fertility rates), while more developed regions seem to maintain demographic stability and even growth.Footnote 3

Universities, which see their funding broadly tied to student numbers, have recruitment basins that differ in nature and coverage. Most of the universities that have a broad geographic appeal tend to be those that already existed before the 1989 Revolution, though some older universities also struggle to appeal beyond the local counties. This paper identifies recruitment typologies among Romanian universities and discusses vulnerability and opportunity in the context of ongoing demographic transformation. In particular, it explores this topic in an environment in which the dominant funding model is one that links financial allocations to student numbers.Footnote 4 Given the structural constraints of these financing models in the current context of change, we emphasise the opportunity for universities to understand their local and global competition environments, to be able to better assess alternative ways to attract scarce students, retain talent, and emerge as unique educational and resource pillars in Romanian society.

Broadly, the paper will try to achieve three major objectives. First, it aims to analyse trends in student admissions and map university recruitment flows. Secondly, it aims to analyse student recruitment networks of key Romanian universities using RMU data. Lastly, it aims to critically discuss the implications for Romanian universities taking into account demographic changes taking place at the regional level.

2 Data and Methodology

The article derives its conclusions based on the use of RMU (Unique Matriculation Register) registration data for students, which have previously been employed in the development of a UEFISCDI demographic forecasting paper (see Santa and Fierăscu 2020). For recruitment patterns, first-year student registrations were used. The students registered in the database might not be a completely accurate reflection of actual matriculations as, upon database processing, certain errors were uncovered. Nevertheless, the RMU database remains the best measure to date of enrolment patterns across universities, using a single methodological approach to data collection. The RMU data we could access covers the period 2015–2019 and refers to information about recruitment destinations and high school information for more than 1 million Romanian and international students.

It is important to note that the following discussion is bounded by the context of domestic enrolment to Romanian universities, excluding the impact of degree mobility abroad, i.e., when a county’s higher education participation patterns are analysed, the information refers strictly to those students that have applied to study in Romania. The breakdown of RMU student registrations is detailed in Annex 1—Distribution of first-year students in our RMU data sample (Table 2).

Table 2 Distribution of first-year students and differences in participation in higher education, 2015–2018
Full size table

The methodology used to analyse the data samples is Social Network Analysis (SNA) (Scott 1988), an approach often found in Applied Network Science (Barabasi 2016). In the network framework, universities are nodes, and the shared student recruitment flows are edges. Two universities are connected if they share student recruitment basins. The thickness of the edges represent proportions of shared recruitment basins. Social Network Analysis is a theoretical, analytical and visual framework to explore, analyse and visualise this complex ecosystem of recruitment (Wasserman and Faust 1994) in order to statistically characterise the ecosystem, the clusters of universities with similar recruitment patterns, and the key players in this space. SNA is a methodology that does not depend as much on the size of the ecosystem influencing inferences from different levels of analysis. It is, however, sensitive to missing data. Take, for example, the missing link between the maritime universities in Constanta. It is likely that we missed other smaller edges among universities and even some universities due to the random cut-off point of the data sample we worked with. However, the sample does include the main Romanian universities, if different sizes, profiles and geographical locations. The missing edges and nodes are not likely to alter the main results of the analysis nor invalidate the conceptual framework that focuses on the mechanism at work that generates such network structures.

Based on university enrolment patterns, we construct a network of recruitment flows among Romanian universities in a bipartite fashion, operationalising nodes of two types: node type 1—universities/university centres; node types 2—recruitment basins at the level of counties. The bipartite network was projected on inter-university linkages based on common recruitment basins in counties. The resulting network of interest makes explicit the national level network of competition among universities in their recruitment patterns. Furthermore, we categorise key universities based on their positions in this complex network of recruitment flows, allowing us to zoom into the specific profiles of key universities. We thus employ the method to better understand the structure of the recruitment landscape and the emerging network communities but also identify potential vulnerabilities and strengths of the overall higher education system as it is.

RMU data offers a snapshot of the student recruitment patterns captured in the past few years. However, university recruitment basins are likely to be influenced by changes to regional demographics in the near future. When analysing the demographic vitality of the recruitment basins themselves, proxy data was broadly employed, as the intercensal period is approaching its end, with many formal databases being calculated in virtue of 2011 results. While Romania does keep a formal account of the resident population that is distinct from nominal domiciles (which often ignore migration), this is still unlikely to be accurate. Indeed, the effect of migration flows and faulty registrations was visible in a few “improbable” y-o-y variations identified in INS databases. The use of proxy data enables us to identify a series of recruitment basins that are particularly vulnerable to future demographic change.

3 Recruitment Typologies and Network Organization

Instead of operationalising recruitment patterns in terms of similarity and diversity of student characteristics (which assumes independence of observations), we choose to reveal university profiles using an often-hidden aspect of recruitment—similarity and diversity of recruitment flows (which assumes universities are interdependent in their competition for recruitment because they target similar potential student pools). We redefine recruitment in terms of flows to highlight the interdependency of universities in the Romanian higher education landscape—when it comes to student recruitment.

The profiles identified in this setup reveal the main vulnerabilities and strengths of universities in terms of their resilience to demographic changes, as well as in the context of competition. We thus conceptualise two dimensions across which we assess universities’ positions in the national recruitment network: (a) diversity of recruitment basins for each university and (b) structure of the universities’ local recruitment environments.

3.1 Diversity of Recruitment Basins: Narrow Verses Diverse Recruitment Networks

This perspective reveals university centres that depend on narrow recruitment (i.e., single-basin recruitment) versus diverse recruitment (multi-basin recruitment) when it comes to a shrinkage of the supply, the potential student populations.

Universities with single-basin or narrow-basin recruitment are expected to suffer more from the ongoing demographic transformations, though this remains partially dependant on demographic trends in the counties that dominate each basin. When the basin shrinks, the number of students recruited by these universities shrinks as well; therefore their financial budget shrinks too. In the network position, they are expected to be peripheral to the system and dependent on only one or two shared recruitment basins with a single central entity. Starting new recruitment bases in other counties further require resources, and the dependence on the university’s actions towards adapting the recruitment strategy is higher.

Universities that attract students from multiple counties are expected to adapt better to changing recruitment flows. When one or a few channels of recruitment shrink, the university can attempt to compensate by increasing recruitment from other counties. Also, nearby universities in decline can be targeted for mergers, thus expanding recruitment basins. Good examples include mergers between Cluj’s UTCN and UBB universities and smaller public institutions in other counties.

3.2 Structure of Local Recruitment Environments: Open Verses Closed Recruitment Networks

Assessing the network of higher-education institutions and their recruitment patterns also allows us to identify the embeddedness of universities in open or closed co-dependence networks on similar recruitment basins.

Universities that have an open local network recruit exclusively from various counties (star configurations). Their advantage is in their exclusive recruitment flows from smaller regions, with the regions themselves not competing with each other. Universities that have a closed local network co-recruit from shared basins (clique configurations). We often see shared recruitment patterns with other universities in the same or nearby university centres.

Network theory posits that open networks rely on exclusive ties to diverse recruitment basins, which give them visibility and monopoly over those regional basins, while closed networks enter constructive competition among universities of shared potential student pools. Embeddedness of universities in these configurations links them to the resilience they have in the face of major demographic and mobility transformations.

3.3 Network Self-Organisation: Vulnerabilities and Resilience

The connection between network embeddedness and resilience has been demonstrated across a wide range of empirical networks and complex ecosystems, from studies of urban development to studies of cancer and cell re-organisation around systemic shocks (Callaway et al. 2000; Newman 2003; Boccaletti et al. 2006).

Network Theory informs us that the structure of the ecosystem encodes information about the vulnerability or resilience of the landscape itself. If the network is dense, cohesive and clustered, it is more resilient to random shocks (even if those shocks are endogenous or exogenous to the system). If the network is, on the other hand, fragmented, centralised and sparse, random shocks can substantively affect the connectivity of the system or its main characteristics, leading to an irreversible impact on the landscape (Table 3).

In other words, if universities have clustered and diverse recruitment networks, major changes that affect recruitment flows and the positions of universities in the landscape have minor repercussions on these communities for two reasons: (a) students have visible alternatives in their university choices and (2) universities have visible alternatives in their recruitment basins. If, on the other hand, universities have open and narrow recruitment networks, major changes that affect recruitment flows can irreversibly affect their ability to reinvent themselves after such systemic shocks (Figs. 1 and 2).

Table 3 Conceptual mapping of vulnerabilities and strengths in co-dependencies of universities on shared recruitment basins
Full size table
Fig. 1
figure 1

Network of Romanian universities’ recruitment flows. Each node is a university. The edge connecting two universities represents the share of total recruitments for each university that is from the same recruitment basin. The thicker the link, the higher the overlap in recruitment between two universities. Node colours reflect network communities, identified using the Louvain clustering algorithm. A network community is defined by stronger connectivity patterns among nodes from the same community then with the rest of the network. Circled nodes are key universities, defined as important based on network centrality measures (in this case, relative Betweenness Centrality scores)

Full size image
Fig. 2
figure 2

County recruitment flows in degree-awarding Romanian higher education institutions. The nodes are counties connected among each other based on the rate of students enrolled in local universities coming from different geographical areas (width of the links). The size of the node reflects the In-Degree score (the number of incoming links). The larger the node, the more attractive that county is for first-time students

Full size image

The following sections of the paper offer a descriptive account of the higher education recruitment flows in Romania, linking them to patterns of demographic transformation, to better frame the debate over the role of universities in effectively managing a rapidly changing environment. We thus propose to address questions about the motivation or the reasons why the landscape looks like this in further research and dedicate this paper to mapping and contextualising the embeddedness of universities into network structures.

4 Results

4.1 Patterns of Enrolment: The Recruitment Basins

A key dimension in the identification of future demographic health in various university recruitment basins is the mapping of said basins today. Of course, enrolment patterns can change in the long run—especially if Romania manages to overcome some of its deeply entrenched access issues—but existing data can only paint an overall picture of what the situation is today. The small time period of available RMU registrations (2015–19) does not permit any extensive mapping of variations in enrolment trends.

The data revealed a complex hierarchy when it came to the size of recruitment basins among universities. The first broad conclusion is that there are just four large university centres that manage to attract students from multiple counties, having diverse recruitment basins and displaying dense and cohesive networks. These are Bucharest, Cluj-Napoca, Timisoara and Iasi. Outside of the city proper, Bucharest universities are the main destinations for students from no fewer than 12 counties, Cluj universities attract students from 8 counties, Timişoara universities from five and Iaşi universities from four. Constanţa universities are the main destination for students from Tulcea county, but the dominance here is moderate (being a plurality only), with just slightly fewer students opting for Bucharest universities. It is important to note that study programmes undertaken in extensions (campuses situated in other cities/regions) count towards the student totals of the home university. In the case of Cluj-Napoca, both the generalist UBB university and the technical UTCN university have a number of notable extensions, chief among these being UTCN’s North Campus in Baia Mare, formerly an independent university.

There are several university centres that manage to attract a majority of students in their home country but only smaller numbers of students from other regions. In some cases (Sibiu, Braşov), even though the university is dominant in the local county only, the number of applicants attracted from other regions is sufficient to end up forming a majority of the entire student body. For example, Sibiu has a comparable share of out-of-county students to Timişoara, but it is the primary destination only in Sibiu County itself.

Another key pattern is the dominance of urban students among total intakes. This comes as no surprise as factors leading to low access have been documented in other studies. We know that rural students have lower success rates at the Baccalaureate examination needed for university admission.Footnote 5 We also know that this, in turn, is influenced by trends that include higher rates of material deprivation in rural areas. In some areas, the urban/rural gap is the widest in the EU. For example, severe housing deprivation in rural areas tops 26%, but is less than 5% in urban settings. In sharp contrast, the Bucureşti-Ilfov region, which is entirely urban and suburban.Footnote 6

A crucial finding is that there is no major university with a majority rural student population. The highest share of rural students among top 20 universities can be found at the Valahia University of Târgovişte, standing at 49%. This is, in fact, higher than the national weight of the rural population, but when analysing the recruitment basin of the university, we see that it largely rests on the Dâmboviţa County, which at the 2011 census was among the counties with the second-highest share of rural population, at 71%. So, even here, there is a clear situation of under-representation among students with a rural background (Table 4).

Table 4 Share of students from local county per university
Full size table

Al of these particularities influence the structure of the nationwide network of recruitment flows. Table 7 shows that a primary clusterisation among universities can be made based on local county recruitment dependence. That is, between universities that have geographically diverse recruitment basins and institutions that cater to local communities. The latter are uniquely vulnerable to demographic trends in their local counties. This means that universities such as those in Reşiţa or Târgu Jiu, situated in regions with poor demographics, are more vulnerable than those in Ploieşti, Oradea or Suceava, which are situated in counties with either stable demographics or higher overall populations. That said, having a diverse recruitment basin is not an automatic insulation from the effects of demographic decline, as most counties in Romania are currently estimated to be seeing population decline and ageing.

In practice, however, university centres with wider geographical appeal tend to attract students both from buoyant regions and from counties facing demographic decline. A notable example is Timişoara. It attracts most applicants from Timiş county, but the rest of its recruitment basin consists of counties with a recent history of steep demographic decline (Caraş-Severin, Hunedoara) or rapid ageing (Mehedinţi). Timiş itself is, conversely, one of the only two counties to have recorded population growth between the last two censuses.

In the case of universities’ appeal within their home settings (local counties), there are also wide variations. For example, there are universities that attract the overwhelming majority of secondary education graduates within a county. No fewer than 96% of Bucharest and Ilfov students opt to study in Bucharest. 91% of Timiş and Iaşi students, as well as 88% of Cluj students also study in local universities. There are counties where students overwhelmingly opt for a single out-of-county destination. For example, Giurgiu County sends 95% of its students to Bucharest, with Călăraşi sending 85% and Ialomiţa and Teleorman 81% each. Cluj attracts 82% of students from Bistriţa-Năsăud and 81% of those from Maramureş, though in this case, many study in extensions. In most of these cases, geographical proximity seems to play a major role (students opting for the closest major university centre), though the economic vitality of the nearby university centre seems to play a role as well. While prosperous cities such as Cluj and Bucharest attract overwhelming majorities of students from nearby counties, this pattern does not exist in counties close to Craiova or Galaţi, also legacy universities that existed before 1989.

The most intriguing cases are those of small public universities that have been designed to cater mainly to regional applicants or non-traditional students but fail at meeting significant recruitment benchmarks, even locally. The most extreme case at the time of data collection and processing (early 2020) was Caraş-Severin, where there is a local university, yet 72% of students opt for a university in Timiş county. Just 17% of those that enrol in a university do so locally. The consequence is that the local public university relies on a tiny recruitment pool, given that the county is already one of the least populous in Romania.Footnote 7

Table 5 Dominant destination for studies per county
Full size table

Looking at the numbers from Table 5, it becomes clear that there are, in fact, several cases of counties that house a non-religious, public university only to have the majority of students opt to study outside the county. This pattern exists in no fewer than six counties (Alba, Caraş-Severin, Dâmboviţa, Gorj, Hunedoara, Prahova). In two other cases (Bacău, Argeş) only a plurality of students opt for the local university, with significant numbers still leaving the counties for study. In the case of Bacău, this is almost a 3-way tie between students opting for the local institution (28%), Iaşi (25%) and Bucharest (23%).

The data, of course, refers to the post-2015, after student numbers have already declined versus their 2008–2009 peak, so enrolment patterns might reflect better access to state-funded study places in more prestigious university centres. Nevertheless, it does point to the fact that some public institutions cater to a very small number of students from a limited geographic region.

In fact, there is a legal framework that permits and encourages university mergers with the purpose of enticing smaller institutions to pool resources with institutions from large university centres. Two mergers have already taken place, both with Cluj-based institutions. The first was the merger of Baia Mare’s North University with Cluj’s Technical University UTCN, while the more recent is the merger of Resiţa’s Eftimie Murgu University with Cluj’s Babeş Bolyai University. These mergers transform smaller institutions into de facto extensions, allowing the new university to keep existing infrastructure while saving funding via better use of human and administrative resources.

4.2 Different Patterns of Depopulation Within University Recruitment Basins

The current demographic recruitment basins of Romanian universities offer us a static picture. It tells us where students hail from currently. It does not provide information on the nature of the demographic recruitment basins themselves and evolutions that occur within. Now, using the INS databaseFootnote 8 for residents and inter-census variations, we get a picture of demographic trends in recruitment basins from formal statistics. Measuring 2019Footnote 9 versus 2012Footnote 10 numbers, the INS estimates that only one county (Teleorman) saw double-digit population contraction, and that six actually saw a population increase. In annualised terms, this represents a slower rate of population decline when compared to the 2002–2011 inter-census period, despite a rise in the natural population decline of the population. With migratory flows difficult to track, but with various sources indicating high ongoing levels of emigration,Footnote 11 it is likely that—much like before the 2011 census—the impact of outflows on populations is underestimated. Furthermore, there are growing indicators that internal population dynamics have been changing.

While, before 2011, internal migration was modest—with most Romanians seeking to work outside their communities doing so abroad—there are numerous indicators that certain regions are benefiting from significant migratory inflows. This is reflected in sharp differences in employment recovery after the Global Financial Crisis, growing differences in birth numbers between counties and housing construction. Of course, each of these indicators can be influenced by factors other than population concentration: employment can rise due to higher labour force participation, birth rates due to changes in fertility rates, and construction booms can reflect falling household sizes. Nevertheless, the combination of multiple factors is likely reflective of a positive demographic environment. Conversely, the opposite is true: falling employment numbers, birth numbers and low housing construction activity are likely reflective of a social and demographic environment that has been negatively affected by out-migration, ageing or both.

In fact, there are signs that the depopulation of certain counties is accelerating. Regions that gravitate economically towards Bucharest tend to see some of the fastest population declines. Teleorman county saw around 2,600 births in both 2018 and 2019. Deaths, however, exceeded 6,000 in both years, and preliminary data from 2020 points to just one birth for every three deaths. This natural decline is both a symptom and a fortifier of strong trends towards long-term demographic ageing. And this pattern is visible in many other areas. Over 40% of new houses are being built in and around Bucharest, Cluj and Timişoara, the country’s most dynamic cities, even though their share of the overall population is under 20%. The share of these metropolitan areas in total employment has also been rising after the post-2008 global financial crisis, which contributed to a realignment of the Romanian economy. In fact, a majority of Romanian counties have not yet recovered their pre-crisis employment levels, despite the country as a whole surpassing pre-crisis employment levels in 2019.Footnote 12 But, while overall job numbers have grown by little over 2% from 2008 to 2019, that growth topped 20% in Cluj county, 10% in Timiş and nearly 50% in Bucharest’s suburban Ilfov county. The share of national employment in these top cities has been constantly growing.

This phenomenon is, to some degree, self-reinforcing. Existing theoretical and empirical work has already identified networks as being key facilitators for migration via the construction of safety nets or the sharing of information (Palloni et al. 2001; Schapendonk and van Moppes 2007). With networks built up during the past 30 years often leading Romanians to seek work abroad, the networks existing inside Romania are concentrated on a few larger cities. And as the population increasingly concentrates in these cities and their metropolitan environs, there is an emerging risk that students from depopulating areas end up accessing low quality services and struggling to make use of existing opportunities. This phenomenon is increasingly prevalent across Europe and has even started to take a political dimension. In Spain, the question of the España vacía and the claims of a right to minimal services for regions with low demographic viability has emerged. Indeed, the province of Teruel sent a party called Teruel Existe (Teruel exists) to signal the plight of depopulation in Madrid.Footnote 13 There is also a network of European regions that is devoted to fighting the impact of depopulation.

While not a prevalent political issue in Romania as of yet, universities are likely to be among the first to feel the impact of ageing and declining populations. There is a lag of 18–19 years between any change in cohort sizes at birth and university entry, and migration (either as part of families or upon graduating secondary education) tends to contribute to falling cohort sizes as well.

5 Discussion—Sustainable Universities?

Given the above findings, we can see a number of patterns emerge. One of these patterns is the growing concentration of population around cities that are—at the same time—major university centres. More people live in places such as Bucharest, Cluj-Napoca and Timişoara, as well as their environs. The exact extent of their absolute or relative growth will only be known after the 2022 census is processed, but data on employment, births and housing construction clearly point to a divergence between these metropolitan regions (and a few others) and most of the rest of the country.

Incidentally, universities located in these cities—as well as Iaşi, Braşov and Sibiu—tend to attract numerous out-of-county applicants. Other large universities (i.e. from among those established before 1990), including Craiova, Galaţi or Constanţa, have ended up mainly catering to local students. Employment in these areas was, as of 2019, still down on 2008 levels, and housing construction numbers lagged the three top cities and their environs.

These patterns could indicate either a role for universities in fixing population and attracting local investments (i.e. part of a virtuous circle of development) or a preference by students to opt for universities situated in cities that offer a better range of professional development opportunities. The universities that seem attractive to students have thus often given access to employment in the same town.

Conversely, the large number of public universities that struggle to attract even local students reflects poorly on the current structure of the university network. Public money is generally allotted per capita, though capacity is, of course, also adjusted to take into account differing numbers of students. Nevertheless, smaller universities have increasingly resorted to applying for emergency funding or other compensatory mechanisms (e.g., enrolment of Moldovan students on scholarships). However, these strategies are detrimental to the rationality of using public money, and many of these smaller universities are unlikely to be able to sustain economies of scale should local populations of students decline further. Already existing legal channels that enable mergers offer a partial solution, though extensions themselves do maintain higher infrastructure costs.

In the long run, a picture emerges of universities that seem particularly vulnerable to negative demographic trends. The network approach analytically supports the link between universities’ embeddedness into local networks of exclusivity and competition over potential student recruitment basins with their ability to survive major shocks to their recruitment flows. In practical terms, this means that universities at the periphery of the recruitment flow network will have to find alternative strategies to remain financially viable, particularly if their recruitment basin faces accelerated demographic decline.

Universities embedded into dense, cohesive and clustered structures, such as the universities in Bucharest, Cluj and Iaşi, are more resilient to random shocks to the ecosystem. Their location in these structures allows for faster re-organisation, reshuffling of recruitment patterns because they offer visible alternatives for both universities to reach target groups, as well as students to access alternative institutions. For universities that are, on the other hand, embedded into fragmented, centralised and sparse network structures, such as the universities in Oradea, Galaţi or Craiova, random shocks to the recruitment flows can substantively affect their connectivity to the system, leading to an irreversible impact on their recruitment landscape.

In other words, if universities have clustered and diverse recruitment networks, major changes that affect recruitment flows and the positions of universities in the landscape have more limited repercussions on these communities for two reasons: (a) students have visible alternatives in their university choices and (2) universities have visible alternatives in their recruitment basins. If, on the other hand, universities have open and narrow recruitment networks, major changes that affect recruitment flows can irreversibly affect their ability to reinvent themselves after systemic shocks. Such changes are particularly likely if recruitment basins are situated in demographically vulnerable regions.

One possible strategy for vulnerable universities is to design and implement new, attractive study programs that rethink their financial dependency on local recruitment basins and compete on attracting students based on the relevance, quality, and competencies new study programs offer. This strategy is in line with emerging national higher education directions towards internationalisation, engaging diaspora in teaching and research programs back home, designing study curricula that strengthen the link between education and employability and strengthening links to partner organizations (in business, civil society and government) in designing and delivering these new study programs.

It is important to note that there are certain factors that influence admissions in Romanian higher education that are independent of demographic trends or university efforts to improve attractivity. Chief among these is the low rate of access for the overall population. Romania is last in the EU in terms of its share of 30–34-year-old students who have completed tertiary education. There are mounting calls to attempt an expansion of access, often linked to calls for reforms in secondary education.Footnote 14 If tertiary education access and retention do rise, it is likely that a new layer of demand—possibly with unique socio-demographic characteristics—will arise. Wether these expanded admissions opt for metropolitan or regional universities does not change the fact that overall numbers of potential front-loaded students continue to decline. A slow rise in access would be—for many counties—completely offset by a more rapid decline in average annual cohort sizes. The pressure to rationalize the university networks—at least in the public sector—is unlikely to disappear.

The fact that recruitment basins are naturally urbanising and concentrating on large cities is likely to bring certain benefits. The existing social infrastructure in universities (such as dormitories) caters largely to out-of-town students and is likely to see less demand-driven pressure. A larger proportion of newly admitted students would have likely attended top quality schooling and had access to informal and nonformal learning opportunities that are more likely to exist in large cities.

This, of course, does not offset the likely problems associated with depopulation. Romania already operates a network of “simultaneous” education schools in which multiple grades are taught at the same time in sparsely populated communities. Commuting is also problematic over longer distances, and the network of pre-tertiary dormitories was dismantled during the 1990s. All of these are salient issues, given that depopulation will require significant efforts to maximize the number of children in isolated, ageing communities receiving quality schooling and having a chance at reaching tertiary education. Access to better social infrastructure in tertiary education has no impact on early school leavers, and access to quality and supportive secondary education should remain a key policy priority.

Lastly, Romania needs—and currently lacks—an adaptation strategy for ageing, depopulating regions. This should, among others, include provisions on facilitating access from early education all the way to university learning, if it is to prevent the consolidation of an opportunities gap.