Education Demand and Conflict in Ukraine: Using Web Scraping to Understand the Relationship

(1)

Education Demand and Conflict in Ukraine: Using Web Scraping

to Understand the Relationship

By

Mykola Skrynnyk

Submitted to Leiden University

Institute of Public Administration and Department of Economics

In partial fulfilment of the requirements for the degree of Master of Science in Public Administration

Supervisor: Prof. Max van Lent Second Reader: Dr. Eduard Suari Andreu

The Hague, the Netherlands 2019

(2)

Abstract

This work examines the consequences of conflict for education. With Ukraine serving as a case, the analysis investigates how the start of the armed conflict in 2014 impacted on the regional distribution of the demand for higher education. Using a publicly available online information source, I create a dataset on millions of university applications in the country between 2010 and 2018. At the aggregated level, panel data analysis suggests that the conflict negatively impacted on the total number of applications to Kharkiv region that neighbours the affected oblasts as well as to Kyiv city, Ukraine’s capital. In contrast, Lviv oblast in the West is estimated to have attracted extra applications in both 2014 and 2015. A closer examination using logistic regression confirms these findings at the disaggregated level. In the treatment period applications were 8% less likely to be submitted to Kharkiv region in the East but 17% more likely to end up in Lviv oblast in western Ukraine. The study finds a differentiated impact on applications to high-prestige programmes as well as male and female applications.

Keywords: armed conflict, higher education, university admissions, web scraping, panel data, logistic regression

(3)

Figure 8. Share of Undergraduate and Postgraduate Applications per Geographic Region (2012-2018) 39 Figure 9. Share of Undergraduate and Postgraduate Applications per Region of Interest (2012-2018) 40 Figure 10. Average Standardised Admission Scores of Undergaduate Applications per Region and Gender

pooled across 2012-2015 45

Figure 11. Undergraduate Applications to Prestigious Programmes in Kyiv city, Kharkiv and Lviv oblasts

(Share from the National Total in 2012-2018) 46

Table 1. Fixed Effects Regression to Predict the Number of Undergraduate Applications (in Thous,) 43

Table 2. Actual and Expected Numbers of Undergraduate Applications in 2015 44

Table 3. Application-level Logistic Regressions to Estimate the Probability of Applying to One of the Three

Regions in 2012-2015 47

Table 4. Application-level Logistic Regressions to Estimate the Probability of Applying to One of the Three

(5)

Introduction

The rise of higher education as an integral part of human live is remarkable. While at the beginning of 20th century less than 1% of what is now considered ‘college-age’ population were enrolled at universities, the figure achieved a twenty-fold increase at the turn of the millennium (Schofer & Meyer, 2005). Available data for 1970-2010 show that the developed world enjoyed a particularly strong expansion. In Germany and the United States, for example, the numbers swelled from 1.5% to 13% and from 11% to 27% correspondingly (Barro & Lee, 2017; Roser & Ortiz-Ospina, 2019).

Naturally, education came into focus of scientific research. Already at the outset of education research, sociologists studied status attainment processes and social class implications for school choice (Karabel & Astin, 1975; Sewell & Shah, 1968), while economists - inspired by Mincer’s earning equation – analysed investment decision-making behaviour of prospective students and estimated return on education (Psacharopoulos, 1994).

The incidence or rather devastating effects of armed conflicts led a different group of scholars to study the consequences of violence for education (Akresh & de Walque, 2008; Lai & Thyne, 2007; Shemyakina, 2011). Being primarily concerned with enrolment rates and average years of schooling in conflict-ridden countries, this strand of research has been limited in geographic scope and neglected the consequences for higher education.

Since armed conflicts are of rare occurrence in Europe and North America, an increased focus on African and Latin American states is reasonable. Yet, it leaves a significant knowledge gap in our understanding of how armed conflict could impact upon education in generally more stable countries. Besides, armed conflicts have been repeatedly shown to negatively impact on education in and increase migration from directly-affected areas, but very little is known about collateral effects on non-affected regions or the direction of within-country migration flows. With Ukraine serving as a case, this work aims at bridging the knowledge gap by examining the relationship between a recent military conflict in the East of the country and the regional demand for higher education. In so doing, the thesis draws on the research in (i) conflict and education, (ii) migration and education and (iii) economics of education. Equipped with a rich data source on millions of university applications in the country, the study estimates the collateral damage to the demand for higher education in the country’s regions.

(6)

Due to the fact that the demand side is often overlooked, this work has little overlap with the existing literature. The most closely related research is an article by De Groot and Göksel (2011), which was a first major attempt to link the conflict to the demand for and not the supply of education. This article along with other relevant literature is reviewed in detail in Chapter I, which sets a basic theoretical framework.

Chapter II provides a detailed description of the case. It starts by situating Ukraine’s education system in the historical context and continues with an overview of key events related to the conflict. The last section of the chapter presents hypotheses that guide the analysis.

Empirical results are reported in Chapter III. Along with a brief explanation of the data collection process, the chapter presents an exploratory analysis and provides first evidence of the effect. Two separate sections in Chapter III deal with the analysis. The first one uses panel regression to estimate an aggregated regional impact of the conflict, while the second one employs logistic regression to quantify the effect at the disaggregated level.

The concluding part of the thesis summarises the findings while linking them to the literature. An extensive list of annexes provides additional information in the form of complementary figures and alternative individual-level statistical models. A detailed description of the data acquisition and cleaning process together with some guidance as to the replicability of the analysis can also be found in the annex.

(7)

Chapter I: Literature Review

Research on education has come a long way. From a vaguely delineated subfield of economics and sociology in the middle of the previous century, it has grown to a stand-alone branch with many subfields of its own. Due to the lack of specific literature on the demand for education in conflict-ridden states, this thesis adopts an eclectic approach. In so doing, it draws on insights from economics of education, education and migration as well as conflict and education. This section explains key empirical findings in each of these subfields as well as points out to their relative strength and weaknesses.

Economics and Education

A recent study of articles in top economic journals revealed that economics of education is on the rise now with a growing number of non-American – still mostly Western – researchers who look at education through the lens of economic theories (Machin, 2014). Long before this rise, however, economics and education were brought together by scholars such as Schultz (1961), who suggested that differences in national outputs can largely be explained by differences in investment in human capital. Ever since, a substantial share of economics of education is concerned with estimating both private (Psacharopoulos, 1994; Psacharopoulos & Patrinos, 2004) and social return on education (Jæger, 2007).

Return on Education

Academics working in this subfield try to explain wage differentials by the level of education while controlling for important covariates, such as academic ability and family background. These works are usually characterised by methodological rigorousness and availability of high-quality administrative data that enables researchers to causally link education levels to persons’ earnings. Hence, it is common in this research to use regression discontinuity (e.g. Khoo & Ost, 2018; Öckert, 2010) and resort to instrumental variables approach that reduces bias in the estimates (e.g. Heinesen, 2018; Kirkeboen, Leuven, & Mogstad, 2016). Doing so allows for a greater degree of certainty that the established relationship is indeed causal.

Thus, Öckert (2010) uses administrative data on university admissions in 1982 in Sweden to estimate the earnings of that year’s entrants in 1990s. In early 1980s college admissions were extremely selective in Sweden and applicants with very similar academic merits occasionally received different admission decisions. Using admission status as an instrument, Öckert finds small or sometimes even negative impact of years of college on earnings and somewhat

(8)

paradoxically concludes: “If taking the earnings penalty while in college into account, most applicants would probably have done better in the labor market if they were not admitted in 1982” (Öckert, 2010, p. 516). Although interesting of itself, his results run counter to most empirical evidence found elsewhere. This might be attributed to a year-specific effect or a peculiarity of Swedish society or perhaps even Berkson’s paradox at play (Berkson, 2014). More recently, Kirkeboen et al. (2016) have not only found a significant effect of education on earnings but also succeeded in quantifying the differences in return on education by study fields. A valuable empirical finding of their paper is that the payoffs are less different between selective and non-selective universities than between different programmes. For instance, by enrolling in an engineering programme instead of humanities individuals would essentially triple their earnings in the short-run (Kirkeboen et al., 2016, p. 1060). The research was conducted in the context of Norway and again relied on a rich source of administrative data. Notwithstanding empirical rigorousness and the scope of the research undertaken in these papers, they all share several methodological limitations. For instance, regression discontinuity approach allows one to estimate only local average treatment effect and therefore limits the applicability of estimates to the cases far away from the cut-off point (Angrist & Pischke, 2015). Another shortcoming is a rather limited geographic score, for the scholars predominantly focus on developed countries, in particular on Scandinavian states. These are known for their generous welfare systems with high levels of redistribution (Bradley, Huber, Moller, Nielsen, & Stephens, 2003; Esping-Andersen, 1990). The results obtained in the Social Democratic Welfare States might not be effortlessly generalised to other countries, especially to those in the developing world.

Education and Economic Growth

Such limitations are of much lesser concern to scholars who approach economics of education from the macro-level. Schofer and Meyer (2005), for example, examined the factors that contribute to the growth in university enrolment rates worldwide. Utilising panel dataset with seven hundred year-observations, they estimate that the expanded secondary education system is one of the strongest predictors of enrolments. Major determinants also include the level of economic development, ethno-linguistic fractionalisation and Communist legacy.

In a similar vein, Valero and Van Reen (2019) study the effect of the expansion of universities on economic growth in 78 countries. Even after controlling for some regional trends and conducting in-sample validation, their evidence suggests that not only are increases in the

(9)

number of universities positively related to economic growth, but that there is also a spill-over effect among countries (Valero & Van Reenen, 2019, p. 66).

Of course, these macro-studies do not purport to establish a causal relationship, though several papers on the micro-level lend additional credibility to the link between universities and economic growth. In the study conducted using national survey data in South Africa, Borjas et al. (1994) estimate a two-stage regression on Cobb-Douglas production function and discover that it is actually only those with a higher education degree who, on average as a cohort, contributed to the economic growth in that country. This suggests that the expansion of universities is indeed likely to stimulate growth through increases in labour productivity. Relatedly, whether education is a productivity-increasing tool or a simple signalling device (Layard & Psacharopoulos, 1974; Stiglitz, 1975) is another popular topic in the discussion on the role education in the generation of earnings. But long before one can actually reap any benefits of a higher education degree, one needs to enrol in a college. A separate subfield of economics of education studies the determinants of educational choices that individuals make.

School Choice

Studies on enrolment behaviour have been a big growth area since 1990s. Early literature on the topic mostly focused on the aggregate response of student enrolments to changes in federal financial aid or looked at the enrolment process from university’s standpoint (Chapman, 1981), but recent research is more diverse.

In an extensive report on educational behaviour in the US, Paulsen (1990) brings together evidence from several dozens of scholarly works on the topic to summarise key determinants of school choice. He comes up with three main categories of variables: environmental (in fact, national), institutional and student. Family income, from the last category, was the often-cited predictor positively related to enrolment, while academic ability and parental education level were cited less frequently. Among institutional determinants, tuition fees were deemed central. Of course, higher fees were associated with a lower probability of enrolling.

Other notable factors included university’s location, especially urban location, admission selectivity and the extent of the federal financial aid. Several studies reviewed by Paulsen also found distinct group differences. Thus, some suggested that female applicants attached less importance to the cost factor than male applicants. Besides, urban location and residential life carried more weight for females too (Paulsen, 1990, p. 62).

(10)

Mangan and Baron (2003) extended of the list of relevant factors with university reputation and course content. The authors point out that these factors have a differential impact on male and female applicants. The former value university reputation more than course content whereas the latter prefer vice versa. The results are based on student survey data from England. A follow up research conducted in Scotland (Briggs & Wilson, 2007) confirmed these findings. This time, however, the scholars used a two-year survey and a much larger sample. While the list of relevant factors did not change per se, the relative weights of factors were slightly different in Briggs and Wilson’s sample. Academic reputation, location, graduate employment opportunities and social life were cited as the key ones. Their survey data also show that the role of costs – not only in terms of tuition fees but also living, travel and accommodation costs – significantly increased in the second year of the survey, implying that the cost factor started to be a more important consideration than it used to.

Still, the evidence as to the role of tuition fees remains contradictory. In the study on the impact of higher tuition fees on educational choices, Dunnett et al. (2012) demonstrate that fees play a rather secondary role in college choice. Using mixed methods, the authors establish that course reputation and location are major determinants. Concerning the location factor, it has been shown that high-performing school leavers are less likely to apply to a prestigious college if there is no such college in the vicinity (Mangan, Hughes, Davies, & Slack, 2010). This leads to the undermatch between universities and students. In other words, applicants with higher academic abilities end up in lower-quality universities and vice versa1_.

While closely related to college choice, information-seeking behaviour of applicants received little coverage in scholarly works. Drawing on survey data from Australia, Brennan (2001) analysed applicants subjective knowledge relates to objective expertise regarding prospective universities. She found no clear relationship between the two, concluding that many students are ill-informed about the university of their choice at the time of application. Her results also suggest that applicants with higher levels of self-perceived knowledge about prospective university are less likely to engage in information-searching activity.

Other notable works on college choice concern the impact of school quality on chances to be admitted to a selective university (Berkowitz & Hoekstra, 2011) and the effect of changes in

1_{Matching mechanisms are itself a large subfield of scholarly research inspired by theoretical works of}

Abdulkadiroğlu (2005; Abdulkadiroğlu & Sönmez, 2003) and the likes (Haeringer & Klijn, 2009). The issues of mismatching and matching however stay beyond the scope of this work, since they have been well studied using both large administrative datasets (Bo, Liu, Shiu, Song, & Zhou, 2019) as well as laboratory experiments (Chen & Sönmez, 2006; Lien, Zheng, & Zhong, 2016).

(11)

admission policies on entrants’ academic efforts in China (Grau, 2018). Compared to the research on return on education, school choice literature lacks in both scope and quality. Most works in this subfield heavily rely on small-scale surveys and predominantly focus on educational choices of school leavers in the developed world.

One notable exception to the latter trend is a rather dated study by Gerber (2000). He analysed the changes in educational transitions in post-Soviet Russia using multi-year data from a large-scale national survey coupled with the data from a more focused follow-up questionnaire. His findings largely conform to those of the mainstream literature. Parental education and occupation are shown to significantly impact on the probability of college entry. So does parental KPSS affiliation, which turns out to have an effect comparable to urban residence (Gerber, 2000, p. 236). Concerning gender, the work reveals no differences in the probability of university entry for males and females up until 1992. Following that year, it was actually female and not male school leavers who were more likely to transition to university studies. The research on college choice serves as a major theoretical underpinning for this study, yet it does not provide enough explanatory power on its own to construct a hypothesis for the case at hand. A discussion on migration, conflict and education is therefore in order.

Education, Migration and Conflict

On the one hand, there is a considerable amount of literature that deals with the relationship between the level of education and decision to migrate (Levy & Wadycki, 1974; Schwartz, 1976), ‘brain drain’ and economic growth (Mountford, 1997), university aspirations and migration (María Cubillo, Sánchez, & Cerviño, 2006). On the other hand, migration during civil wars has also been well studied (Bohra-Mishra & Massey, 2011; Shryock & Eldridge, 1947). Yet, while the proposition that education, migration and conflict can be intimately related is a common sense, there seems to have been no attempt to systematically study the three together. In the attempt to formulate a more fine-tuned hypothesis for the case of Ukraine, this section reviews some findings from the migration-related literature and conflict studies.

Migration and Education

When education is considered in the context of migration, it is usually seen as a purpose for migration (María Cubillo et al., 2006) or as a covariate that differentiates migrants from a source country (Schwartz, 1976). Early research on the topic showed that the choice of a destination for migration heavily hinges upon migration costs which are best proxied by the

(12)

distance from a source country to a prospective destination (Levy & Wadycki, 1974). For the case of Venezuela, it was also claimed that educated individuals are on average more mobile than non-educated (Levy & Wadycki, 1974, p. 387).

Cross-country analyses only confirm such claims. In their study of 56 countries, Bernard and Bell (2018) establish that migrants are a selective group of people in terms of educational attainment. In all but two of the countries in question – Nicaragua and Haiti being the exceptions – people with secondary and university degrees have higher odds of migrating. Importantly, their data demonstrate that the effect is independent of human development index, i.e. more educated people in the West are likewise more inclined to migrate than less educated. However, within-country migration remains rather overlooked in the literature. A rare exception is a study on the effects of the education reform in Finland in 1990s (Haapanen & Böckerman, 2013). The study showed that the reform increased the within-country mobility of those students whose vocational colleges were transformed to polytechnics as opposed to those whose colleges were not affected by the reform.

Education-related migration in conflict-ridden countries is even less studied. Because the driving force for migration in conflict areas is the desire to ensure one’s security, education has been deemed irrelevant either as a goal or as cause for migration (Browne, 2017). However, the incidence of migration was linked to the level of violence in the area, with less violent areas having smaller migration rates (Bohra-Mishra & Massey, 2011).

While education might not be the reason to migrate, it could matter for migrants once they have left a conflict area. This argument comes from in-depth interviews with 50 people who migrated from war-torn countries to Europe (Hagen-Zanker & Mallett, 2016). Some of the migrants cited educational prospect for their children as a significant factor for choosing big cities, especially in Germany. In stark contrast to the scarce evidence on the effects of conflicts on education migration, conflict and education literature has much more to offer.

Conflict and Education

In the early 20th_{century, the type of education people received was seen as both a factor that}

contributed to the start of the World War I (Bagley, 1918) and a remedy that could prevent the repetition of the war (Gerwig, 1920). The prescribed remedy was not used properly and the World War II did break out, but the relationship between education and conflict was again stressed by Brown (1943). He saw the benefits of education for career prospects of women and

(13)

noted that while the university enrolment rate in the US decreased on average over the war-time period, new university enrolments actually grew. The decrease, he argued, was driven by upper year students in graduate professional schools who left universities to undergo military service (Brown, 1943, p. 408). His work was therefore one of the first to suggest that conflict might in fact increase the investment in human capital.

Economic side of the relationship between conflict and education was studied by Ichino and Winter-Ebner (2004), who examined the effects of WW II on lifetime earnings. By comparing people in Germany, Austria, Switzerland and Sweden who were school-aged when the war started, the authors show that the population in two former states suffered a significant loss in years of schooling. When linked to earnings, this loss is estimated to account for 10 percentage points decrease for each year out of schooling (Ichino & Winter‐Ebmer, 2004, p. 76).

Generally, most of the scholarly articles published on the topic give disproportionally large weight to the supply side of education in the context of military conflicts and political instability. Thus, the pernicious consequences of wars for education were estimated on the macro-level by Lai and Thyne (2007). The scholars find no evidence to support a popular belief that educational expenditures suffer from increased military spending during civil wars (Lai & Thyne, 2007, p. 284).

The negative impact of conflicts, their argument goes, is primarily due to directs costs on education, such as physical destruction of buildings, and indirect costs, such as smaller investments or injuries to educators (Lai & Thyne, 2007). In their article, tertiary enrolment rates are estimated to decrease by 3 percentage points in countries engulfed in civil wars. Although the scholars suggest that civil wars cause education-related migration flows and drive out students from conflict areas, they do not elaborate on how this migration unfolds.

It should also be noted that Lai and Thyne’s (2007) estimates are an average impact and the real costs of war can be much dearer, especially for school enrolment. Rwanda is a notorious example. This African country was devastated by a genocide in 1994 and the most conservative estimates indicate that children in affected areas suffered on average 20 percentage points drop in school achievement (Akresh & de Walque, 2008).

Beside evidence from war-torn countries in Africa and Latin America, there are data from post-Soviet states too. Using national households survey data from Tajikistan , Shemyakina (2011) estimates the effect of a prolonged armed conflict on schooling in the country. Children in directly affected areas, she convincingly demonstrates, suffer a much greater loss from the

(14)

conflict in terms of schooling and are less likely to complete mandatory education. Looking at the group-specific effects, she estimates that female pupils were 12 percentage points less likely to attend a school if their household was directly affected by the unrest, but finds no significant effect for male pupils, concluding that it is likely to be a country-specific phenomenon rather than a general trend (Shemyakina, 2011, p. 194).

The Basque case

Before wrapping up this chapter and proceeding to case description, a substantial contribution from de Groot and Goksel (2011) should be mentioned. In their paper on the Basque region, the scholars focus on the demand-side for education, assuming that due to a low military intensity of the conflict, any changes in educational attainment are unlikely to be driven by the supply side. Premised on the insights from ‘brain gain’ literature, their augment is that the low-intensity conflict might incentivise people to invest in human capital, i.e. their own education, to increase odds of migration.

To test this hypothesis, De Groot and Goksel adopt a sophisticated methodological approach. First, they use longitudinal individual-level data to estimate major factors contributing to educational attainment. Then, using synthetic control trial – a method which has recently proved especially instrumental in comparative policy studies (Abadie, Diamond, & Hainmueller, 2010, 2015) – the authors construct a synthetic region that is assumed to be similar to the Basque region, had the conflict not started. This is followed by a matching procedure, in which the authors derive an educational distribution for this synthetic region and finally compare it to the actual distribution using difference-in-difference method.

The results largely confirm their theoretical prediction. They find that the level of education of Basque-born individuals rises more that of other Spaniards. Conversely, the level of education of migrants from the Basque region rises less than that of migrants from other communities in Spain. This, the authors conclude, shows that while the conflict increased the demand for education in the Basque region, it also lowered the education cut-off point at which Basque people decide to migrate.

The following chapter serves as a basis for adapting theoretical insights from this chapter to the case of Ukraine. It offers a detailed overview of the country’s education system, provides a basic description of conflict-related events and formulates a hypothesis that directs the analysis in the third chapter.

(15)

Chapter II: Case Description

It is a well-established fact that the way one selects a case for their study bears immediate implications for research outcomes (King, Keohane, & Verba, 1995). Selection on the dependent variable reduces variation and may lead to Type II error, while selection on both dependent and independent variables is an even more egregious mistake conducive to confirmation bias. Bias might creep into the study even when the selection is seemingly random. As shown by Geddes (1990), a prima facie innocuous selection of cases according to data availability can also lead to biased results.

Aware of these methodological pitfalls, both the author and the reader should understand the strengths and weaknesses of the case at hand. For that reason, this chapter starts by explaining the rationale behind and proving justification for the selected case while acknowledging its drawbacks. This is followed by a historical description of the higher education system in Ukraine and its more recent developments, including the introduction of External Independent Testing (EIT). A separate section deals with the conflict and broader socio-economic conditions. In the last section of the chapter, hypotheses are formulated.

Case Selection

There are several merits to selecting Ukraine as a case. Firstly, although conflicts, especially the ones with a non-state involvement, are still common in the modern world, they tend to be geographically constrained to several regions. As could be seen in the UCDP Conflict Encyclopedia2, African countries or war-torn states in the Middle East could well be chosen to study the consequences of armed conflicts and economic hardships for the demand for education. But it would be more difficult to tease out any effect there due to confounding variables, such as ethnic fractionalisation or religion, and a much larger incidence of conflicts in the first place. In contrast, Ukraine’s case offers a more favourable opportunity to estimate the effect of armed conflict as a shock event, producing more generalisable results.

Secondly, a few of the potential candidates, i.e. states involved in a conflict, have an education system comparable to Ukraine’s in terms of size and diversity. Due to its Soviet past and a popular belief that one just “must obtain a degree”, Ukraine can boast circa 25% of the

2_{A database maintained by the Department of Peace and Conflict Research at Uppsala university in Sweden. It}

provides powerful infographics with temporal and geographic dimensions of conflicts. The database is accessible at www.ucdp.uu.se.

(16)

population with completed tertiary education (Barro & Lee, 2017)3_{. Thirdly, the military}

conflict which started in 2014 and remains unresolved as of today has directly affected only part of the country. With the exception of several major cities in the conflict area, education infrastructure remained intact. This helps to isolate the impact by controlling for changes in the supply side of education.

Last but not least, Ukraine’s education system and IT infrastructure were mature enough to establish an online information service that enables university applicants to monitor their university admission. From the outset, the architects of the system were not as concerned about privacy issues as their counterparts in the EU and made all records in the system publicly available. In other words, anyone with Internet access – and of course some command of Ukrainian – can observe millions of university applications4. Given a sufficient know-how, one can use an invaluable source of information on school choices that spans almost a decade. On a critical note, it should be acknowledged that there are disadvantages to the case at hand. Firstly, the turbulent period of 2010s was not only characterised by a revolution and change of government, the annexation of Crimea and military conflict, but it was also full of policy changes in the domain of education. Importantly, the fundamental Law on Education had been substantially modified just before the war in the East broke out. Some other changes were made to the Law on Higher Education as well as to the process of university admissions.

The second downside concerns the generalisability of results which is still somewhat limited. Being a developing country, former Soviet outpost and today’s member of the Eastern Partnership, Ukraine has not much in common either with the EU member-states or with its former fellows from the Soviet Union. Its education system has got rid of much of its Soviet legacy, but still does not fully resemble the European model.

With regard to theory, however, Ukraine can be seen as a crucial case. That is the case in which the variable of interest is most-likely to produce a visible effect if there can be any effect at all (Blatter & Haverland, 2012). In other words, if any hybrid conflict is to indirectly impact upon university applicants’ preferences – be it because of a feeling of perceived threat or concomitant economic hardships – then the conflict in Ukraine is a perfect candidate. With these caveats in mind, next section sets the country’s education system in the historical context.

3_{An official estimate stands at 30%. However, the last census in Ukraine was carried out in 2001, so a large}

measurement error should be expected for any indicator based on population size. Still, even in a pessimistic scenario, Ukraine remains one of the top five countries by the share of the population with a university degree.

(17)

Historic Overview of the Education System

Higher education in Ukraine has its roots in the 17th century. The country’s two oldest universities were established in Kyiv, the capital, in 1615 and in Lviv, a major city in the West5_,

in 1661. Admittedly, at that time the two cities belonged to two different states: the Tsardom of Russia and the Polish-Lithuanian Commonwealth respectively. It was not until the first half of 18th_{century that higher educational institutions (henceforth – HEIs) were established in}

other parts of Ukraine. Kharkiv in the East and Odessa in the South emerged as centres of higher education at that time (Huisman, Smolentseva, & Froumin, 2018).

HEIs in different parts of the country differed in both quantitative and qualitative terms. Universities under the Russian rule mimicked the French model and therefore enjoyed a much narrower autonomy than their counterparts in Western Ukraine which were much less dependent upon the state (Kurbatov, 2014). The role of religion in education and the language of instruction also differed from one region to another (Huisman et al., 2018). Latin and later Polish and German were official languages at universities in the West, while Russian dominated academic life in central and eastern parts of the country.

In the run-up of the Bolshevik revolution, there were 24 HEIs on the modern territory of Ukraine, with the majority of universities located in the Russian Empire. The October revolution gave a powerful impetus for socio-economic transformations which defined the development path of the education system for decades.

‘Red’ Education

Despite generally negative depiction of the communist era in the discourse of post-Soviet Ukraine, the overall sentiment towards Soviet education policies is rather favourable (Kurbatov, 2014). The Soviets effectively eliminated illiteracy, including among rural population, built up a massive system of professional technical schools (PTSs), known abroad as vocational schools, and boosted enrolment rates at HIEs across all the republics.

An important feature of Soviet education was the provision of educational services free of charge at all levels. The state even awarded scholarships to high-achieving students to help them cover living expenses. This had a distinctly positive effect on student mobility (Huisman

5_{It is worth noting that administratively Ukraine is divided into 24 oblasts, the autonomous republic of Crimea}

and two cities with a special status: Kyiv and Sevastopol. Designations like ‘Western Ukraine’ are not officially used. Yet, the geographic division is a commonplace in popular parlance and has a long history of usage.

(18)

et al., 2018). Consequently, university cities started to attract students from other regionswho, owing to the state’s financial support, could now afford relocation and living away from home. In the interwar period educational institutions mushroomed in the Soviet republics. Ukrainian SSR saw a fivefold increase in the number of HIEs between 1917 and 1939. Just before the Second World War there were as many as 173 institutions and almost 200,000 students in the republic6. Naturally, technical and pedagogical universities constituted three fourth of this number, while comprehensive universities amounted to only 6 institutions.

Unlike the market-driven expansion of universities in European countries and the United States, the key determinants of higher education in the Soviet Union were the needs of the centrally planned economy (Huisman et al., 2018). That is why a particular emphasis was placed on professional technical schools. These schools fulfilled a dual purpose. On the one hand, serving as a lower-tier alternative to tertiary education, they provided some sort of qualifications to those who with 9 years of schooling. On the other hand, they established an essential link between the education system and command economy.

Together with formal credentials, students in PTSs gained professional skills and technical know-how that would later enable them to take a range of jobs. PTSs qualifications could be obtained in a much shorter period of time than higher education degree. Besides, graduates of professional technical schools were still eligible for university admission (Kurbatov, 2014). In 1950, more than 200 thousand students attended 584 PTSs in Ukrainian SSR. By the end of 1960s, the number of PTSs grew to 760 while the number of students swelled to 800 thousand. Shortly before the collapse of the Soviet Union the number of comprehensive universities in Ukrainian SSR hovered around a dozen, but technical and pedagogical universities still surpassed other types of HEIs. In the academic year of 1988/1989, there were almost 900 thousand university students in Ukrainian SSR. Every third student attended a technical university, while only one out of ten students studied at a comprehensive university (Huisman et al., 2018, p. 410).

Notwithstanding these remarkable achievements, there were quite a few problems that plagued the Soviet education. Needless to say, Russian language prevailed in educational institutions while local and national languages were effectively excluded from HEis in all the republics. Other notable issues include an allegedly widespread system of unofficial benefits that

6_{Education data for the Soviet period are taken from Narodnoje Khozyaystvo SSSR (Statistical Yearbook of the}

(19)

privileged groups enjoyed during the university admission process, disregard towards social sciences and omnipresent ideological indoctrination (Kurbatov, 2014).

As described by Janmaat (2008), each student – regardless of the type of institution attended – had to complete a series of mandatory subjects in humanities and social sciences such as History of the Communist Party of the USSR, Political Economy, Marxist-Leninist Philosophy, Scientific Atheism etc. These subjects were intended to mould a particular mindset of a Soviet citizen and instil communal values that would increase social cohesion in the USSR (Kuraev, 2016). Almost thirty years after the Soviet Union collapsed, its mark on education is still deeply felt in Ukraine and other former republics alike.

Education System in post-Soviet era

The downfall of the Soviet Union had disastrous consequences for Soviet republics. A massive system of economic ties that connected educational institutions, enterprises and individuals across the USSR ceased to exist. The system that administered the process of generating both demand and supply was no longer there. Instead, 15 national economies emerged.

It is fair to say that Ukraine’s newly gained political independence was coupled with a heavy economic dependence on former Soviet republics, most notably on Russia. The country underwent three major transformations in 1990s: from a Soviet republic to a nation-state, from an authoritarian regime to a democracy and from the planned economy to the market economy (Kutsyuruba, 2011). To adjust itself to a new modus vivendi, the state had to regain economic control over the production processes on its territory, abandon the planned economy system, introduce a new currency and undertake a large-scale privatisation (Cornelius & Lenain, 1997). In steering the reforms, Ukrainian leaders were faced with a set of domestic problems such as a sharp decline in fertility rates, runaway inflation and swelling unemployment (Kutsyuruba, 2011). Exacerbated by constant political crises and strenuous relationship with former allies, these challenges led to a dramatic drop in industrial output and significant decrease in living standards. The turbulent nature of reforms in 1990s had important educational ramifications. Together with the centrally planned economy, Ukraine inherited a part of a highly centralised education system. While the system could boast a strong school structure and extensive network of vocational and technical education, it was also characterised by a considerable reliance on state funding, Soviet ideological footprint, including deliberate neglect towards social sciences, and endemic corruption (Huisman et al., 2018).

(20)

Besides, a decades-long emphasis on educating specialists in sciences and engineering during the Soviet era led to the concentration of educational institutions in major industrial cities. Semiv and Hvozdovych (2012) point out that 40% of vocational educational institutions and 60% of HEIs in 1991 were located in six big cities: Kharkiv (East), Donetsk (East), Dnipropetrovsk (East), Odessa (South), Kyiv (Centre) and Lviv (West)7.

A central education policy in 1990s, however, was neither to increase universities’ financial independence, nor to modernise outdated technical equipment, nor to improve teaching standards. Instead, driven by the fact that education system is one of the central pillars of any modern nation-state, the government of Ukraine focused upon nation-building. Having been a minority nation for centuries and having experienced a substantial degree of Russification during the Soviet era, Ukrainians were quick to initiate their nation-building programme already during perestroika (Janmaat, 2008). Just before the Soviet Union collapsed, the republic passed the law ‘On the languages in the Ukrainian SSR’.

The law stipulated that Ukrainian shall be the only language of instruction at HEIs except for the regions where national minorities constitute the majority of the population. The only region which met this criterion was Crimea, which meant that HEIs in all other regions would not be allowed to use any minority language, not even widely-used Russian. The document heralded the beginning of Ukrainisation of the public domain and ensuing disputes regarding the rights of ethnic minorities.

Nation-building policies continued into 1990s when already independent Ukraine entered a period of reorganisation of secondary schools, redrafting university curricula and training of a new generation of teachers and educators (Kutsyuruba, 2011). The last issue was especially pressing since the Soviet institutional culture became deeply rooted in Ukraine’s higher education, mainly due to the fact that university staff remained largely unchanged. As Janmaat aptly noted: “Many teachers of History of the Communist Party, for instance, had to change their orientation overnight and teach History of Ukraine” (Janmaat, 2008, p. 15).

Implemented along with decommunisation, nation-building policy was designed to ensure that neither Ukrainian language nor culture was supressed in the domain of education. In early 1990s, a presidential decree effectively forbade opening of first-grade Russian-taught classes, thereby aiming to transform Russian schools in the eastern part of the country into Ukrainian ones within an 11-year span (Janmaat, 2008). The constitution enacted in 1996 only solidified

(21)

the government’s commitment to Ukrainisation by reasserting the status of Ukrainian language. Contrary to what many political experts predicted, Kutchma’s administration did not put Russian language in any special place and treated it on par with other minority languages. Despite their slow pace, reforms did not leave school programmes and university curricula untouched. Old ideologically-laden disciplines were rapidly eliminated from educational programmes while several new subjects made their way to classrooms. Some of these, for instance Ukrainian and Foreign Culture or Ukrainian Business Language, had an explicit national flavour. In stark contrast to schools, which are still required to use centralised Ministry-approved textbooks, HEIs are allowed to determine the subject matter of compulsory subjects (Janmaat, 2008). Admittedly, private HEIs, which were first established in the country not so long ago, are not exempt from the requirement to teach these subjects.

In sum, nation-building remains on the agenda in Ukrainian politics, but it has lost much of its lustre since 1990s. Having examined 14 policy documents produced by Ukraine’s government between 1991 and 2008, Fimyar (2010) concludes that catching-up Europeanisation is one of the recurring leitmotifs in the legal discourse of independent Ukraine. In fact, the fear that the country’s diplomas would be deemed worthless in the European community made catching-up Europeanisation a new policy-driver in early 2000s, while nation-building slowly slipped into the background (Huisman et al., 2018).

A tipping point for higher education reforms in Ukraine was reached in 2005 when two landmark events occurred. Firstly, Ukraine ratified the Bologna declaration, which set the country’s course on Europeanisation. The ratification greatly facilitated the integration of Ukraine’s education system into European education space. It speeded up a set of positive changes in higher education, including but not limited to, the adoption of ECTS by all HEIs in 2006, a large-scale implementation of a three-cycle degree system (Kovtun & Stick, 2009) and issuance of diploma supplements that follow EU guidelines (Nikolaeva, 2017).

Secondly, Yushchenko – a newly elected pro-European president at that time who was a central figure during the Orange revolution – issued a decree “On Urgent Measures to Ensure the Functioning and Development of Education in Ukraine”. The decree was a follow-up to the ratification of the Bologna declaration and aimed at accelerating the reforms that would help bring Ukraine’s system in line with European standards.

Most importantly, the decree established the Ukrainian Centre for Educational Quality (UCEQ) which was tasked with the development, implementation and monitoring of External

(22)

Independent Testing (EIT), also known as External Independent Evaluation or External Independent Assessment8_{. EIT has drastically changed the process of university admissions in}

Ukraine. Its results bear immediate relevance for education choices of prospective applicants. To understand the place of the External Independent Testing and its role in university admissions, it is worth discussing the structure of study programmes in Ukraine which uniquely combines Soviet legacy and European aspirations.

Education Laws and Degrees

As of 1991, there was no separate law that regulated the higher education system in Ukraine. Until mid-1990s, the only fundamental legal act in the domain of education was the Law of the Ukrainian Soviet Socialist Republic on Education. According to the law, HEIs could issue 4 types of academic qualifications: Junior Specialist, Bachelor, Specialist and Master respectively. Candidate of Sciences and Doctor of Sciences were two postgraduate degrees that could be conferred by academic senates at certain universities. HEI types were only partially defined in the law and included professional technical schools, colleges, music schools, academies and universities.

Together with the adoption of the Constitution in 1996, the Parliament of Ukraine made a substantial amendment to the Law on Education9. Beside finally updating the title of the law by substituting the name of a non-existent political entity with the country’s official name, the amendment introduced a new classification of HEIs. This classification grouped HEIs into 4 levels of accreditation. Essentially, the level corresponded to the type of educational institution. Thus, HEIs accredited at the first level comprised professional technical schools (technikum as well as uchylyshche), whereas the second level of accreditation included colleges. The classification was most important for institutes, universities and academies which could be accredited as either third or fourth level HEIs. Academic institutions accredited at a higher level enjoyed more powers with regard to admission policies, programme content, degrees they could confer and so on.Consequently, professional technical schools and colleges were limited to issuing only junior specialist’s and bachelor’s degrees respectively. Conferring specialist’s and master’s diplomas remained an exclusive right of third and fourth-level accredited HEIs.

8_{There is no official translation of the original term Zovnishnie Nezalezhne Otsiniuvannia (ZNO). For consistency,}

I use External Independent Testing or shorthand EIT to refer to this test.

9_{Amended 47 times before being replaced, the Law was legally regarded as one document with the same number}

№ 1060-XII, even though the name of the Law was slightly changed in 1996. The Law on Education (1991) refers to the original legal act as well as all its subsequent versions.

(23)

It was not until late 2014 that another major revision of the Law on Education (1991) was undertaken10_{. Not only did a new version of the Law abolish the practice of distinguishing HEIs}

by the level of accreditation, but it also strictly defined types of HEIs, which now included only universities, academies, institutes and colleges. This did not mean however that other HEIs would lose their education licenses but rather that they would be rebranded and subsumed under one of the new categories. In particular, professional technical schools, first-level accredited HEIs inherited from the USSR, were reclassified as colleges.

Most importantly, the amendment transformed academic qualifications11. In stark contrast to Junior Specialist, which was kept on the list of degrees, specialist’s diploma was discontinued, leaving Ukraine’s students with no alternative to master’s degrees. Likewise, Candidate of Science and Doctor of Sciences were substituted for a single Doctor of Philosophy degree which is more familiar to Western scientific community. The amendment also introduced a new academic qualification, Junior Bachelor. Similar to Junior Specialist on the surface, this new degree was awarded by colleges and deemed incomplete higher education degree. Its holder was nevertheless entitled to continue studies in a bachelor’s programme and, in some cases, made it possible to follow a fast-track programme.

Despite the abovementioned changes, academic qualifications that each HEI type could confer were not altered by the revision of the law. Junior Bachelor, Junior Specialist and Bachelor were degrees conferred by colleges, while Bachelor, Master and Doctor of Philosophy were awarded by universities, institutes and academies. When contrasted with a type of academic degree, levels of accreditation seem to have played little role in applicants’ choice of HEIs. Hence, the change had significant long-term implications for the development of the education system of Ukraine but made little difference from university applicants’ perspective.

The Law on Education (1991) was eventually repealed in autumn 2017, when a new act was passed in the Parliament. Targeting primarily Ukraine’s school system, this law increased the duration of schooling from 11 to 12 years, granted greater autonomy to schools, outlined the powers of school principals and modified the certification of schoolteachers. Figure 1 illustrates the relationship between these legal documents throughout the time.

10_{Other revisions of the Law also made important changes to the way the education system operates. Yet, they}

dealt with issues which are of much lesser importance to the topic of this work, e.g. ownership rights, academic representation, teachers’ salaries etc.

(24)

Figure 1. Timeline of Education Laws in Ukraine (1991-2018)

As seen in the figure, the Law on Education was complemented by a more specific Law on Higher Education starting from 2002. That specific act regulated the higher education domain until 2014. Beside outlining the HE system of Ukraine, the law described the levels of accreditation and precisely defined the degree types. Shortly before the Law on Education (1991) was revised in 2014, the Parliament passed a new Law on Higher Education, which pinned down the definition of Junior Bachelor and established the National Electronic Education System (EDBO).

It is worth noting that most programmes in the country are offered by public universities. While private HEIs have been slightly growing in numbers, they remain small in size and only a fraction of students attends them. This might be attributed to the fact that a majority of private HEIs have an economic or business profile (Janmaat, 2008). Unlike public universities, their private counterparts do not provide state-funded places, i.e. the state does not cover tuition fees for students at these universities.

EIT and University Admissions

Piloted from 2004 to 2007 and initially accepted by selected universities, EIT became a national university entrance test in 2008. The test was designed to make university admissions fairer and more transparent. The centralised testing system put an end to most university-based entrance exams thereby allegedly eradicating rampant corruption (Osipian, 2009).

In some respects, EIT is similar to many other standardised tests used for student ranking during the university admission process. But unlike SAT and ACT in the US, External Independent Testing is a not a single exam but a series of subject-specific tests. While the exact list of subjects tends to slightly vary from year to year, the admission process in a generalised manner could be described as follows.

In their final year, i.e. in 11th_{grade, high school students register for EIT. The deadline for}

(25)

Literature has been mandatory for all EIT-takers since 2008. School graduates were also required to sit EIT in either Mathematics or History of Ukraine in 2010 and are required to do so again since 2016. Otherwise, the choice of EIT subjects is to the liking of the applicant. EIT exams take place within three to five weeks usually between May and June. Results are released approximately three weeks following the exams. Participants receive test certificates for each subject they have passed. EIT scores are standardised and reported on a 200-point scale with 100 being a minimum score.

Once the results for all EIT subjects have been made available, university admission period starts. Each study programme has its own requirements as to which EIT certificates should be submitted. The overwhelming majority of the study programmes require applicants to submit EIT certificates in three subjects with one of them being in Ukrainian Language and Literature. Some programmes like Arts or Architecture are exceptions and admit students based on one or two EITs and a special university-based professional exam12. Together with the certificates, university applicants are required to submit their high school diploma and supplement.

The total admission score is a sum of EIT scores and an average grade of the high school diploma. Before 2014, this average grade was converted to a 200-point scale, yielding a maximum obtainable admission score of 80013_{In 2014, it was decided that the school grade}

would yield only 60 points, lowering the maximum admission score to 660. Later on, the high school diploma was further depreciated to only 20 and further to 12 points.

However, starting from 2015 the admission process was modified in two important respects. Firstly, EIT and high school diploma scores were now weighted, with coefficients being set at the programme level14. Secondly, the preference ranking system was introduced, in which applicants choose up to a dozen of universities and rank them according to their preferences. In sum, the university admission process has undergone marked changes since 2008. While these had important policy implications, there is no valid reason to suspect that any of the changes significantly intervenes with a possible effect of the conflict. Since EIT weights are defined at the programme level, the changes in how the admission score is calculated could impact on the probability of applying to a particular programme, not to a particular region. Similarly, given that a newly introduced ranking system permits applicants to choose quite a

12_{Such programmes are referred to as programmes with special entry in Annex II that describes data cleaning.} 13_{School grades in Ukraine are originally measured on a 12-point scale. In fact, applicants could receive up to 50}

extra points to their admission score for winning an academic contest at a national level.

14_{Apparently, the contribution of a school average grade varies more within years than within programmes with}

(26)

few universities, it is unlikely to affect the probability of applying to any particular region15_{. It}

might be worth knowing if an application to some region is someone’s first or fifth preference, but it is more important to know if someone makes any application to that region at all.

The Conflict

A growing discontent among the population over government policies, high-level corruption and sluggish economic growth were further fuelled by the decision of Viktor Yanukovych, the then President, to suspend preparation for signing a long-awaited Association Agreement between Ukraine and the EU. The decision sparked mass demonstration in Ukraine’s capital in November 2013. What was initially a peaceful demonstration of pro-European citizens – mostly students – grew into violent riots following police brutal crackdowns that attempted to suppress the demonstration (Diuk, 2014). Continuing pro-European and anti-government protests eventually led to the erosion of political support for the then ruling Party of Regions. As the consequence, Ukrainian government stepped down in January 2014 while Yanukovych himself fled the country by the end of February (Lapshyna, 2015). After Yanukovych’s hasty departure, the Party of Region eventually disbanded. Evidently, the retreat of the party created some sort of a political vacuum in eastern and southern Ukraine where the biggest political base was that of the Party of Regions. But before a snap election could be held, the political vacuum was rapidly filled by Russian-backed separatists (Kuzio, 2015).

In March 2014, Russian special forces, known as ‘green men’, started to take over official buildings in Crimea, eventually leading to a complete loss of control over the peninsula on the part of the Ukrainian government. Similarly, Russian-backed military groups occupied government buildings in Donetsk and Luhansk oblasts, two regions in the East that border the Russian Federation. Both regions declared independence from Ukraine in April 2014, establishing Donetsk and Luhansk People's Republics respectively.

The government of Ukraine soon launched a so-called Counter-terrorist Operation in Eastern Ukraine. The operation was de jure supervised by the State Security Service of Ukraine until April 2018 when it was succeeded by the Joint Forces Operation, delegating the supervision to the Armed Forces of Ukraine. Since mid-2014 the government has maintained only partial control over Donetsk and Luhansk regions16.

15_{But of course, it does have enormous impact on the likelihood of being admitted to some region.}

16_{The Center for Strategic and International Studies created a detailed timeline of events in Ukraine from the}

(27)

While Russian high-level officials have repeatedly denied any involvement in military actions in Ukraine, international community explicitly condemned both Russia’s annexation of Crimea (UN, 2016) and aggression against Ukraine in the East (Congress of USA, 2017; European Parliament, 2015). The ongoing conflict is viewed as an example of hybrid warfare which combines the use of conventional military actions, guerrilla operations, information warfare, cyber-attacks, political pressure, economic intimidation etc. (Thiele, 2015; Wither, 2016). Due to the conflict’s hybrid nature, the damages Ukraine has suffered are enormous. It is estimated that there are around 1,5 million internally displaced persons (IDPs) in the country, whereas some 5,3 million people are directly affected by the conflict (UNHCR, 2018). According to the UCDP Conflict Encyclopedia, more than six thousand people have died in the East since 2014, while UNHCR (2019) estimates that the death toll reaches 10 thousand people, including 3.5 thousand civilians.

In economic terms, the conflict was a disaster. Yearly inflation rates were double digits between 2014 and 2017; Ukrainian currency, Hryvnia, plummeted by more than 200 percent against the U.S. dollar by 2016; and military expenditure increased by 50% compared to the pre-conflict figures. Figure 2 and 3 highlight some these developments. Figure 2 illustrates a trend that runs counter to empirical evidence of Lai and Thyne (2007).

Data Source: World Bank

(28)

As seen from Figure 2, military expenditure indeed increases at the expense education expenditure. Contrary to what some might expect, this had no immediate effect on state funding provided to universities in the short-run. Importantly, the regional distribution of state-funded places, i.e. tuition waivers and stipends, remained stable over the years up until 2016 when the admission procedure was modified.

It is also worth mentioning that economic changes were rather uniform across the regions in terms of real income decreased as well as risings inflation rates. Economic indicators therefore exhibit relatively little variation among regions but substantial variation between years.

Similarly to the Basque case analysed by De Groot and Goksel (2011), the conflict in Ukraine could be regarded as a long-term and low-intensity conflict. But unlike it, the conflict in Ukraine has had dire socio-economic consequences. It is not unreasonable to suggest that the conflict and concomitant socio-economic instability have affected the behaviour of university entrants. The goal now is precisely to hypothesise possible directions of this effect and later test the hypotheses. Next section aims at reaching this goal.

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 Year 0 5 10 15 20 25 30 U S D a g ia n s t U A H 0% 20% 40% 60% 80% 100% 120% R e a l In c o m e (% fr o m P re c e d in g P e ri o d ) A d m is s io n A d m is s io n A d m is s io n A d m is s io n No Change

Data Source: National Bank of Ukraine

(29)

Hypothesis

Bringing together the insights from the literature reviewed in the first chapter and case specific characteristics reviewed above, I make a claim that the conflict has affected the behaviour of university entrants by making them less willing to apply to universities in eastern Ukraine and more likely to consider universities in the West. The argument is three-pronged.

Firstly, universities located in the East are in direct proximity to the conflict zone. Due to security concerns or a feeling of perceived threat, this might make eastern regions a less attractive destination for those who can effortlessly choose to study elsewhere. Moreover, a general perception of instability of the geographic region – which in its worst form causes an anxiety over Russia’ possible military invasion – could induce potential applicants to favour HEIs in other parts of the country.

Secondly, Kyiv, albeit still the most attractive destination for many applicants, could also lose out due to several factors, including political upheavals and revolutionary events that took place in the city in early 2014, just before the start of the conflict. Besides, since the capital was and still remains Ukraine’s most expensive city, exacerbating economic conditions might have been inducive to choosing a cheaper place to study in terms of both tuition fees and living costs. It is likely that high-scoring applicants were less affected as they would have been provided state assistance in the form of tuition waivers and academic stipends. Yet economic factors could lead a significant number of applicants to prioritise HEIs outside the capital.

Finally, one of the regions which could become much more attractive to Ukrainian applicants is Lviv oblast in the West. Not only does this region have a successful track record of providing academic services for centuries, but it also hosts some of the very top universities in the country. Located far away from the conflict area, it might have benefited from its image of a safer place as opposed to university cities in Kharkiv or Dnipropetrovsk in the East. Besides, Lviv city has a reputation of being Ukraine’s capital of culture. Due to its close proximity to the European Union and historical ties with Poland, the city could also attract those who are considering eventually moving abroad.

Several reasons suggest that undergraduate applicants are, ceteris paribus, more susceptible to the effect. Consider domestic, in regional terms, applicants in general. Post-graduate applicants, i.e. those who just graduated from undergraduate programmes, are already familiar with local universities, academic staff; they are likely to have created a network of friends and some might have already combined studies with work. This entails that the opportunity cost of

Education Demand and Conflict in Ukraine: Using Web Scraping to Understand the Relationship