University autonomy under democratic backsliding: a case study of a plagiarism investigation against Serbian Minister of Finance (2014–2019)

Scholars have documented a tendency of (semi-)authoritarian regimes to undermine university autonomy, mainly through organizational (de jure) changes. This paper presents a case study of a publicly triggered plagiarism investigation by the University of Belgrade into the doctoral thesis of the Serbian Minister of Finance, one of the key members of the increasingly authoritarian regime. The analysis finds a proceduralized and delayed response of the university’s leadership, which indicates lowered de facto autonomy from politics, despite the university’s continually high de jure autonomy. The investigation was closed only after a mobilization within the academic community which resulted in a university’s blockade that forced its leadership to retract the contentious thesis. The case study shows that, in contexts of democratic backsliding, political capture can extend farther than usually thought, impacting even the implementation of internal university standards. On the other hand, the analysis also shows that political capture is not necessarily irreversible and that academic community can mobilize to ‘undo’ it. This reinforces the notion of academic communities as value-driven groups capable of exerting peer pressure to override even authoritarian pressures. In order to understand the dynamic of the plagiarism inquiry in its entirety, we apply insights from theory of power to complement and overcome the limitations of the conventional theoretical frameworks on democratic backsliding and academic autonomy.


Introduction
Autonomy is a fundamental value for universities (Aberbach & Christensen, 2018;Enders et al., 2013;Scott, 2021). However, in societies undergoing democratic backsliding, university autonomy is far from guaranteed. As recent examples from Central and Eastern Europe show, (semi-)authoritarian regimes are content putting universities into their function (Corbett & Gordon, 2018;Vlk et al., 2021).
Prior analyses of the de-autonomization of university in contexts of democratic backsliding have primarily focused on the de jure route (Enyedi, 2018;Kováts et al., 2017). Governments that seek to reduce university autonomy do so through legislative changes that alter the rules for the governing and control of universities (for a typology of these rules, see Verhoest et al., 2004). University autonomy is thus compromised through 'autocratic legalism', which is pursued through dismantlement of democratic norms without the violation of legal provisions (Labanino & Dobbins, 2021;Levitsky & Ziblatt, 2019).
There are sometimes struggles over autonomous university conduct that do not pertain to the university's structural autonomy per se. In such cases, it may be too late for an illiberal government to resort to 'autocratic legalism' to prevent an unwelcome university decision. Is it possible, however, that the university's leadership reduces their autonomy to privilege political interests over the institutional norms? Given the proverbial image of academic leaders as ones cherishing their autonomy and integrity standards (Leveille, 2006), especially in cases where foundational regulations are affected, such a scenario seems unlikely. However, this paper will provide a case study from a context of democratic backsliding that demonstrates that university autonomy can be voluntarily subdued to politics, even without autocratic legalism.
The case study concerns an investigation into plagiarism at the University of Belgrade (UB), Serbia's and Western Balkan's largest university, regarding the doctorate of the current Minister of Finance and former Mayor of Belgrade, Siniša Mali. The investigation began in 2014 and lasted for 5 years, during which time the regime's authoritarianism continuously increased (Freedom House, 2021;V-DEM, 2020). Plagiarism inquiries are a routine matter, usually resolved in a few days or weeks. This is the norm even when highranking officials are implicated, as shown by the examples of the past plagiarism probes of former German Ministers for Defence and Education (Kulish & Cottrell, 2013;Pidd, 2011) and the former President of Hungary (BBC, 2012), which were all closed quickly. Our case, however, lasted much longer and involved multiple, perennial reviews. Only after a series of protests by the academic community, culminating in a blockade of the Rectorate, did the UB leadership close the case by declaring the contentious doctorate plagiarized.
Drawing on press reports, regulatory and statutory acts of UB, its official documents and reports, and discourses developed through traditional and social media, we analyze how the investigation unfolded, focusing particularly on how the behaviour of the two key stakeholders in this process, namely representatives of the political establishment and the wider academic community, was influencing the course of the investigation. The analysis will produce evidence that the UB leadership demonstrated, for the majority of the process, 'mobilization of bias' (Schattschneider, 1960) by perpetuating 'non-decision-making' in its agenda setting (Bachrach & Baratz, 1970). While the mobilization of bias is common to political inquiries (Stark, 2020), it has hitherto seemed unthinkable for university-led plagiarism inquiries; the conventional notion implies that the latter should be mere technicist processes. The analysis further documents a process of counter-mobilization within the academic community, which led to a radical student-led protest that forced the UB leadership to retract the minister's thesis. The 5-year-long dynamic of the investigation, with all its 'twists', will initially be analyzed with reference to two countervailing theoretical frameworks: the traditional theory of academic leaders as fierce autonomy-defenders and the theory of democratic backsliding. However, we also argue that these two frameworks, either individually or together, cannot explain the entire dynamic of the observed investigation without including theory of power (Foucault, 1978(Foucault, , 1991(Foucault, , 1998Hayward, 1998;Hayward & Lukes, 2008;Sawicki, 1991) as an interpretative lens.
We contribute to broader theoretical debates in three main respects. First, our account contradicts the notion of a priori politics-defying academic leaders who are able and willing to ignore political tensions and take fully autonomous decisions driven by academic norms. While being part of a profession that is value-driven (Labanino & Dobbins, 2021;Leveille, 2006), academic leaders are at the same time subject to wider socio-political power relations, which might impact their freedom of decision-making. Our case analysis shows that theory of power, including insights about 'reshaping of the social boundaries' (Hayward, 1998: 20;Grzymala-Busse, 2008) of appropriate behaviour, offers an explanation for the politics-subduing behaviour of the UB leadership. In contexts of democratic backsliding, the power in socio-political relations tends to shift towards the political establishment, and while some academic leaders remain unrestrained in their exercise of autonomy, others could be relenting to those power relations.
This speaks to our next contribution, which relates to the nature of political capture in contexts of democratic backsliding. In such contexts, as we demonstrate, political capture might extend beyond public law and 'intrude' even on private, i.e. internal institutional regulation. The notion that political capture spreads profusely with the consolidation of (semi) authoritarian regimes is widely familiar. Prior reports have documented political capture 1 mainly within the sphere of public law (Bermeo, 2016: 10-11). However, in the Mali case, as it turned out, it was professional self-standards that governed the plagiarism inquiry, not public law, so their obstruction highlights a more malignant form of capture.
Third, the case study yields another, more 'optimistic' finding, namely that, in circumstances of democratic backsliding, political capture is not necessarily irreversible. The UB leadership continuously delayed the case closure, but bottom-up action by a wider academic community eventually forced them to make its overdue decision. One critical question in backsliding contexts has been whether political capture, as it spreads and gets reinforced across the public sector as the predominant modus operandi, substantively becomes irreversible (Tudoroiu, 2015: 671-672). A growing perception has formed that political capture could be reversed only after a macro-political change, which is seen as the determinant of (subsequent) institutional liberation. In our case study, academic communities 'uncaptured' the process without a regime change. This analysis supports the notion of academic leadership being subject to peer pressures (Leveille, 2006) that can countervail even authoritarian pressures. The observed 'uncapturing' of UB, through bottom-up mobilization by the academic community, illustrates how power relations among stakeholders can change over time-they are in flux rather than being fixed (Sawicki, 1991). Power is contestable, and, even in (semi-)authoritarian settings, 'contestants' can produce and expand their power to influence 'hesitant actors' through joint action and association (Hayward, 1998: 20).

Institutional autonomy and democratic backsliding
Institutional autonomy from politics is the ability of institutions to translate their preferences into action without being constrained or directed by political influences or by considerations of political interests (Hanretty & Koop, 2013: 199). The analysis of autonomy in university settings to date has predominantly concentrated on de jure autonomy (De Boer & Enders, 2017: 59-62;De Boer & File, 2009;Estermann & Nokkala, 2009;Fumasoli et al., 2014), which is conceptualized through structural lenses, i.e. the legal/statutory provisions that define how universities are run. According to this approach, the more power that legal provisions give to a university, the more likely it will achieve high de facto autonomy. A typical example of the de jure perspective is provided by the European University Association, which ranks universities by using legal provisions that define various aspects of institutional autonomy, from staffing and budgeting to operational matters (Verhoest et al., 2004). Recent accounts of university deautonomization under hybrid regimes have applied de jure perspective, finding a wider trend of the weakening of legal safeguards of universities' autonomy (Dobbins, 2017: 688;Enyedi, 2018).

The disciplining mechanism in democratic backsliding
An institution's de facto autonomy may be low even when its de jure autonomy is high, especially in states ruled by (semi-)authoritarian regimes, where many institutions 'preventatively' avoid taking measures that might 'anger' the regime. In their early stages, (semi-)authoritarian regimes often resort to non-institutional forms of retaliation against 'recalcitrant' actors through smear campaigns, direct threats, and foreclosure of future career opportunities (VonDoepp & Ellett, 2011). This spreads fear, and self-censorship quickly establishes itself as the 'new normal' among observers (Ong, 2021), even when they are not directly pressed or threatened.
This mechanism resembles Foucault's disciplining mechanism (Foucault, 1991), where individuals are conditioned into obeying an authority's will, even when they are (no longer) explicitly asked to do so. Thus, actors' autonomy is reduced without direct, hierarchical control. Boundaries of acceptable behaviour are redrawn (Hayward, 1998: 20) as the new 'logic of appropriateness' (March & Olsen, 2009) mandates not following own institutional mission but matching the regime's interests.
Power does not necessarily materialize by decree or through hierarchical coercion (Foucault, 1998: 63). Autonomy can be suppressed through social narratives, whose forcefulness can redefine the boundaries of acceptable behaviour. Institutional freedom is not restrained only by current pressures and events; actors' memories of past events can also have a restraining effect on their current freedom in decision-making (Hayward, 1998: 16). In backsliding societies, such memories are often associated with early 'exemplary' cases in which the regime persecuted those contradicting its interests.

Is de-autonomization locked in?
Is this process of political alignment across the institutional and societal landscape selfreinforcing? Based on prior studies, it would seem so. In the literature on democratic backsliding, political capture is often cast as irreversible once it starts spreading. As hybrid regimes consolidate power through fear and intimidation that lead to massive self-censorship, levels of autonomy across the socio-institutional landscape drastically go down. Then the regime proceeds to structural de-autonomization, by adoping legislative changes that curb the de jure autonomy of institutions. Such changes enable subsquently the appointment of pro-regime allies to lead those institutions. This locks in the 'politically friendly' mode of institutional operation (Levitsky & Ziblatt, 2019). This entrenches power relations, and the logic of subjugation to politics become self-reinforcing.
Yet, even in (semi-)authoritarian settings, one should not ignore the old Foucauldian notion that power can constantly shift; it exists everywhere, and it circulates throughout society (Sawicki, 1991). Power can be created and enhanced through joint action and mobilization (Hayward, 1998: 20), through which new narratives can be imposed to override the extant 'us and them' narratives. They might lead to discourse changes that can alter power relationships and affect how the involved actors, i.e. stakeholders behave.

Serbia and democratic backsliding
Since 2012, when the Serbian Progressive Party (SPP) came to power, Serbia has seen major democratic backsliding, despite its declarative plea for EU accession. From the early phase of its reign, SPP established control over the media and started persecuting critical voices and opponents through smear campaigns, arrests, and violence. This allowed the regime to tighten its grip on power and efficiently expand its control over the public sector (Castaldo, 2020). Today, international reports, such as Freedom House's, place Serbia in the category of hybrid, i.e. semi-authoritarian regimes (Freedom House, 2020. The V-Dem index indicates that Serbia has experienced the fourth largest fall in democratic standards in the world in the last decade (V-Dem, 2020).
This provides unfavourable circumstances for exercising institutional autonomy, particularly when such an exercise would contradict the regime's interests. However, our case poses a theoretical puzzle, as two logics compete, one relating to the political context and another to the value-driven operation of the higher education sector. Specifically, the 'deautonomization logic', which is associated with democratic backsliding, is pitched against the proverbial logic of academic resistance to political interference. The latter is exemplified in cases like the Central European University in Hungary (Corbett & Gordon, 2018) and is particularly likely when internal regulation and fundamental academic principles are at stake.

Two opposing logics of investigation, their hypotheses and observable implications
In light of these two theoretical postulates, two different hypotheses can be derived regarding the level of de facto autonomy from politics that UB is expected to demonstrate. A swift resolution of the case with a damning verdict (for Mali) would indicate that UB's de facto autonomy from politics is high. This outcome would suggest that the 'academic matters first' logic has prevailed, in line with the notion of academic leaders as autonomy preservers. The observable implication of this approach would be a 'zealous investigative style' (Tomic, 2019: 40) of the UB leadership, characterized by technicist and efficient decision-making, devoid of 'proceduralism' and investigation delays.
However, the logic of political capture, which thrives during democratic backsliding, provides some reason to suspect that the ongoing autocratization might have lessened the space for autonomous university action. The growing atmosphere of intimidation might sway the university leadership to compromise some of its traditional values, including those related to academic integrity. When political interests are at stake, a 'captured' university's leadership might try to 'balance' between confronting the regime and uncompromisingly carrying out its mandated procedures, although one might expect that any such balancing that undermines the professional standards and integrity norms would be severely contested by members of the academic community (Leveille, 2006).
If the latter, 'backsliding logic' prevails, the observable implication would be that either UB clears Mali of plagiarism charges or that it exhibits behaviour that relies on so-called non-decision-making: a delayed inquiry, led with a view to prolonging or preventing its conclusion. The literature recognizes this as the 'mobilization of bias' (Schattschneider, 1960;Bachrach & Baratz, 1970: 8) in which the institution in question demonstrates the 'second face of power' (Lukes, 2005), namely the ability to prevent a certain outcome from being realized. In our case, the outcome that is being prevented is the retraction of the minister's thesis. The mobilization of bias is a common, if not an almost universal, modus operandi in political integrity investigations (Stark, 2020). However, it is less expected in academic settings. If it is observed in our case, it would indicate a UB's shift towards a politicized modus operandi, meaning that the 'democratic backsliding logic' transformed the UB autonomy into a function of wider socio-political power relations in Serbia.
The theories of democratic backsliding and academic defiance to political influences provide an overarching framework for hypothesizing what could happen in the Mali case. Yet, to nuance the analysis of the dynamics of the plagiarism investigation, we will need to bring in theory of power as well. Theory of power is particularly helpful for explaining situations where events are not unidirectional and where the underlying logic is influenced by the actions and interactions among multiple actors. The Mali case is characterized by a plurality of interests (Dahl, 1957), held by three main groups of actors: (a) the regime; (b) the university; and (c) the academic community. This plurality of actors produces interactions that over time can push the process in various directions.

The case study
The autonomy of the University of Belgrade UB is the largest and most reputable Serbian university. Between 2012 and 2018, it ranked between the 301st and 400th place on the Shanghai Ranking list and, since, 2018 between the 401st and 500th place. 2 UB has a long tradition of resistance to authoritarian tendencies, including the 1968 rebellion against the communist regime and, in the 1990s, multiple waves of mobilization against the authoritarian Milošević regime and its attempts to abolish the autonomy of the university.
Like the other six public universities in Serbia, UB enjoys a high degree of legal autonomy, guaranteed by the Constitution of the Republic of Serbia (Article 72), functionally defined by legislation (Higher Education Law, 2021, Article 4), and further reaffirmed by their statutes. According to the Higher Education Law, universities have full autonomy when appointing their senior management, whose members are recruited from the ranks of the university's academic staff. There are no external appointments or nominations by non-UB actors. UB sets its own policy and regulates how internal policies and student-related matters are dealt with. The only area outside UB's full control is budgeting: a small part of its funding comes from tuition fees, while the rest is provided by the state. Overall, in de jure terms, UB and the other six public universities are among the most autonomous public institutions in Serbia.
When the Mali case broke out, the term 'plagiarism' did not exist in the Higher Education Law; it was down to individual universities to regulate such matters internally. UB's Statute did recognize plagiarism as a form of academic misconduct but did not provide guidelines for such investigations. This only changed in 2016 with the adoption of the UB Code for Professional Ethics, which specified a procedure for dealing with plagiarism allegations. Until then, it was the individual faculties of UB who ensured that submitted theses had met its standards of academic integrity.

The investigation
In July 2014, Professor Raša Karapandža of the European Business School in Wiesbaden (Germany) published an article on the Serbian web portal Peščanik in which he alleged that Siniša Mali, then Mayor of Belgrade and one of the closest aides of the regime's strongman Aleksandar Vučić, had plagiarized his doctoral dissertation. The thesis had just been defended at the UB's Faculty of Organizational Sciences (FON), pending formal promulgation at the university level-a routine annual procedure. Karapandža's article (2014) consisted of a report obtained through the plagiarism-detecting software Turnitin and a manual analysis of Mali's thesis. The Turnitin report revealed a large number of unreferenced parts that were identical to text from previously published sources (academic papers, professional outlets, internet portals including Wikipedia, etc.). One-third of the thesis' pages contained 33% or more of plagiarized text, and some of them featured as much as 70% or more of plagiarized material.
Karapandža's manual analysis further revealed that some sections in Mali's thesis were translated verbatim from sources in foreign languages, including a chapter of another doctoral dissertation defended more than a decade ago at the University of Groningen by an Eritrean graduate. Anecdotally, that chapter discussed cotton production, which, unlike in Eritrea, does not exist in Serbia but was nonetheless featured in Mali's Chapter 2, the focus of which was the privatization and economic restructuring in Serbia in the 2000s. Further, most graphs and tables in Mali's dissertation were copied directly from other sources without acknowledgement. It was not only the content of those figures that was identical to the original sources but also their font and colors. Karapandža (2014), therefore, concluded that Mali 'did not just pinch a bit, but committed a massive plagiarism' and he invited UB to revoke Mali's dissertation (Fig. 1).

The first review
Following Karapandža's revelations, the then UB Rector made a brief statement announcing that, if necessary, the university would support the formation of a commission to formally review the allegations (Večernje Novosti, 2014). Shortly afterward, UB decided to leave it to FON to examine Karapandža's allegations, even though the FON Dean and other members of Mali's viva panel had immediately stated that there was no need for a formal review as 'the plagiarism allegations are baseless' (RTS, 2014;Večernje Novosti, 2014). However, it did not take long before FON altered this initial position and formed a commission to look into the matter.
Within several weeks, FON produced a report concluding that Mali's thesis was in order despite acknowledging that 'some inadequate use of literature was found' (Mondo, 2014). The report was passed, for final confirmation, to the UB Senate, which only placed it on its agenda after a year. Once this happened, the Senate members then decided not to consider the report, suggesting that -for the inquiry to continue -it was necessary to have precise misconduct regulations at UB (Novi Magazin, 2015). Some critics were of the opinion that instead of returning the case to FON and thus restarting the review process, the university should have taken over and made a final judgment on whether the thesis was plagiarized. The academic movement Save the Science, which had just been formed with the mission to protect the integrity of the academic community, suggested that FON's own acknowledgment of the thesis' referencing omissions was sufficient for the university to declare it plagiarized (Spasimo nauku, 2015). Still, the Senate returned the case to FON for a second review, announcing that, in the meantime, UB would be working on a new Code for Professional Ethics to set out a precise academic misconduct procedure.

The second review and the mobilization of an academic watchdog community
Like its first report, the second FON report, released in December 2016, concluded that 'while the contested PhD did contain omissions concerning the use and citation of other Fig. 1 The extent of the plagiarism in Mali's thesis (Karapandža 2014) sources, these were not too problematic, especially in the light of the alleged scientific contribution of the thesis' (Politika, 2016). Several months after its publication, this second report was rejected by the UB Senate, with the rationale that the FON Ethics Commission, which produced the report, had not been qualified to carry out the review, as three of its five members were not academics. Thus, the Senate decided to return the case again to FON, instructing it to form a new, expert commission, fully composed of scholars, which would produce -this time under the newly adopted UB Code for Professional Ethics (RTV, 2017) -another report on the thesis' alleged plagiarism.
The decision to return the case to FON for yet another review led to growing criticism among scholars, who warned that the process was turning into 'endless pingpong' between UB and FON. These developments catalyzed the rise of a small 'watchdog community,' which, from that point on, started to mobilize academic resistance in more organized ways than in the past, when criticism of UB's handling of the case had been occasional and spontaneous. One of the leading voices in this rising mobilization of resistance was Dušan Teodorović, a former professor at the University of Virginia and currently Professor Emeritus at UB's Faculty of Transport and Traffic Engineering. From early 2016 on, he made a series of public appearances repeating the claim that Mali's thesis was a prime example of plagiarism and that the university -had it genuinely wanted to address the issue -could have done it expediently (Fig. 2) (Vreme, 2016).
In 2016, Teodorović averaged 1.4 media appearances per month, which was a solid frequency given the lack of access to the mainstream media for commentators critical of the regime. He kept a somewhat lower public profile in 2017, but then in 2018 his public presence, increasing it further in 2019. Alongside Teodorović, Karapandža, who discovered the plagiarism, acted as an active plagiarism inquiry 'watcher' throughout the whole process. With limited access to the mainstream media, he turned to social media channels -mainly Twitter -to maintain interest in UB's handling of the Mali case.
During the first year, Karapandža posted around 120 tweets about the Mali case (one tweet every three days), but he posted less frequently in the subsequent three years. In 2017, when his tweeting about the inquiry was at a minimum, he posted 30 tweets in total (one mention/reminder every 12 days). From 2018 on, Karapandža's Twitter activity in relation to the Mali case spiked, culminating in 2019 with an average of 0.77 tweets per day (Fig. 3).
Over time, an increasing number of Serbian scholars, both within the country and abroad, joined the protest over the 'doctorate affair' (Danas, 2017(Danas, , 2018. Alongside public commentary, they contributed to other protest activities such as petitions and legal analyses of the handling of the Mali case.

The third review
FON's formation of another review commission coincided with the appointment of a new UB Rector, Professor Ivanka Popović (formerly UB Vice-Rector), who, upon her appointment in May 2018, pledged that, with the recently adopted Code for Professional Ethics in place, the Mali case would be resolved soon (N1, 2018). However, FON only began working on the matter in early 2019, following several unsuccessful attempts to recruit external members. Simultaneously, the academic watchdog community continued its activities and public critique of UB's handling of the case. Through columns and TV appearances in independent media and on social media, an increasing number of scholars criticized the length of the inquiry, sharing their concern that the case was deliberately being delayed because of Mali's political standing.
The third FON commission published its report in March 2019, producing, again, a positive evaluation of Mali's thesis. The commission suggested that the plagiarism software flagged 16% of the thesis as overlapping with material from other sources and that 'only 7% appeared plagiarized,' adding that 'there is no strict rule to determine the amount of plagiarized material that is needed to invalidate a thesis' (Milivojević, 2019). The commission ignored the rest of the unreferenced material highlighted in Karapandža's original revelations, claiming that some of those sections represented 'common knowledge' (Milivojević, 2019).
Once shared with the public, the report sparked a new wave of criticism. In May 2019 alone, dozens of scholars stepped up their criticism of the reluctance of UB to take the lead and close the case by declaring the thesis plagiarized (Danas, 2019a;N1, 2019a).

The Rector's ambiguous stance and further delays
UB's reaction to the third FON report was ambiguous. The Rector initially suggested that the process was 'still in progress', confirming that the UB Committee for Professional Ethics would have the final say on accepting or rejecting the report (N1, 2019b). At the same time, the Rector stated that the amount of text replicated from other sources could not be quantitatively prescribed, adding that this was subject to mentors' discretion. For some critics, this was another attempt to downplay the scope of Mali's plagiarism and the importance of the previously acknowledged instances of plagiarism. The Rector also pledged to restrict access to the appeal process with regard to the latest FON report to those holding formal positions at UB (Mondo, 2019a). This would exclude a large circle of stakeholders from submitting their reasons as to why the report deserved to be rejected. Yet, within a few days, the UB Senate overruled this proposition, deciding that any UB academic employee could file an appeal.
Ahead of the anticipated decision about the last FON report, UB received four appeals. The most prominent one was signed by a group of around 140 UB professors, which pointed to dozens of further examples of plagiarism, not stated in Karapandža's original article (Novi Magazin, 2019). Another appeal was submitted by a smaller group of professors from the Faculty of Law, whose analysis of the FON report provided legal arguments as to why the UB Committee for Professional Ethics could have already overridden it rather than send it back multiple times (Danas, 2019b). Providing additional support, smaller groups of students began staging performances, including one outside the Rectorate, during which fake PhD certificates were dished out to passers-by (Table 1).
In mid-July (2019), the Committee for Professional Ethics concluded that the latest FON report could not be accepted as a valid document because it gave an 'incomplete, unclear, and contradictory opinion' (Mondo, 2019b). However, instead of making a final decision itself, the Committee decided -by a small majority -to return the report to FON for further revisions. This move was met with fierce opposition. One (deputy) member of the Committee and another member of the UB's Council of Legal and Economic Sciences resigned, disagreeing with the 'prolongation of the final decision' and the 'never-ending ping-pong game' (N1, 2019c). The Rector herself, however, stated that it was 'wise' to ask FON to 'more coherently explain why they think Mali's thesis is valid' (N1, 2019d).

A radical turn: students' occupation of the UB Rectorate
Amidst the latest prolongation, the Mali case took a radical turn on 13 September 2019 when a group of about 20 students broke into the UB Rectorate, taking control of the building and starting a strike inside. Under the motto that 'the University belongs to its students, not to politicians,' they demanded that the Rector call for Mali's resignation from his position of Minister of Finance on state television (N1, 2019e). The protesters invited fellow students, professors, and other citizens to join them inside the Rectorate. Ruling officials, including the Prime Minister and President of Serbia, as well as the pro-regime media, branded the protesters as politically instrumentalized, calling on the UB senior management not to give in to the students' 'ultimatum' (Danas, 2019d;Republika, 2019). The UB Rector's Collegium condemned the protestors, stressing that UB was not going to 'allow any disruption of regular work' (Radio Slobodna Evropa, 2019a).
On the first evening of the blockade, a group of ruling party activists of the Serbian Progressive Party-who initially claimed to be 'concerned citizens' (without declaring their party affiliation)-barged into the Rectorate and tried to expel the protesters. The protesters, however, repelled the attack and managed to lock themselves inside the building (N1, 2019e). The Rectorate's security staff did not act to prevent the intrusion, nor did they intervene over the course of the night when the 'anti-student-protesters' remained inside and around the building. The day after, footage emerged in which one of the 'intruders' inadvertently revealed that she had been sent by the ruling Serbian Progressive Party (Ćosić, 2019), thus exposing a violation of the Higher Education Law, the provisions of which forbid party and political activism on university premises. The early days of the strike thus exposed two things: (a) the ruling party was willing to use violence to confront the protesters, and (b) the UB authorities were not capable of safeguarding the Rectorate's autonomy in physical terms.
When the blockade reached its third day, the Rector, together with several members of the UB senior management team, paid a visit to the protesters, urging them to leave the building. The protesters refused to do so; instead, they demanded that the Rector publicly condemn the previous attacks of the ruling party activists. Shortly afterward, the Rector's Collegium condemned 'the physical assaults on the protesters' (from the first night) (RTS, 2019a), but since it did not explicitly state that the attackers were activists of the ruling party, the protesters disapproved of the statement's wording (N1, 2019f). In the days after, the Rector's approach and language started to change, indicating that a solution would be reached through dialogue.

Growing media coverage and resolution
With the blockade of the Rectorate continuing, the media coverage started to expand with each new day (Fig. 4). Over the first ten days, the 'doctorate affair' drew 121 media reports-far more than when FON released its three plagiarism reviews. Public pressure on the UB leadership was growing.
As the blockade entered its second week, the protesters announced their intent to radicalize the protest if the UB senior management continued to ignore their requests (N1, 2019g). Left with no choice but to demonstrate that the university could solve the crisis, the Rector started negotiations with the protesters (N1, 2019h). Eventually, three weeks into the blockade, the two sides agreed on a solution, and a joint announcement was made outside the Rectorate by one of the protesters, accompanied by the Rector, informing the public that the blockade was ending as UB had guaranteed a swift conclusion to the plagiarism inquiry. Although the Rector herself did not prejudice any outcome, this development was widely interpreted as UB's determination to retract Mali's PhD. 3

The regime's disapproval and epilogue
Caught off guard by the Rector eventual 'siding' with the protesters, the regime started attacking her and other members of the UB senior management. In a live broadcast from New York, during his UN visit, President Vučić accused the Rector of 'politically embracing the opposition' (Blic, 2019). Ruling MPs and the mainstream media followed up with a barrage of ad hominem attacks, including ones on ethnic grounds. 4 Nonetheless, UB retracted the thesis, several weeks after the end of the blockade, after following the decision of the UB Committee for Professional Ethics to overturn the last FON report. The Rector publicly announced that the thesis was plagiarized and that it would be cancelled, subject to final approval by the Senate, which took place soon after (RTS, 2019b). Vučić described UB's decision as political in nature, but also expressed regret that Mali had not taken his earlier advice to withdraw the doctoral dissertation himself in order to foil its possible retraction (Danas, 2019d)-a step which Vučić obviously deemed a better strategy for mitigating reputational damage. Mali refused to resign, remaining Minister of Finance.

Discussion: two tales about university autonomy and political capture
The plagiarism investigation into Mali's PhD reveals an unusual dynamic. It lasted for five years, and it took a radical student protest to force the UB leadership to close the case in a manner unfavourable to the political establishment.
The way the UB leadership conducted the inquiry indicates that its de facto autonomy from politics was low. Although producing tangible evidence that they were swayed by political considerations is hard, particularly when 'hidden forms of power' (Lukes, 2005) may be used, for instance, through subtle manipulation of the investigation agenda, including reliance on so-called non-decision-making (Bachrach & Baratz, 1970), there are still strong indicators that the investigation was led through a mobilization of bias (Schattschneider, 1960). In the context of integrity inquiries, showing mobilization of bias indicates low de facto autonomy from politics.
The literature on operational styles of integrity bodies points to certain observable patterns of behaviour that can be taken as signs of mobilization of bias. One such pattern features a 'retreatist style', which is characterized by two features: (a) a lack of an investigator's zealotry in addressing claims of misconduct, and (b) a lack of the investigator's rhetorical assertion that the case will be taken up and addressed decisively (Tomic, 2018). The evidence summarised in Table 2 shows that the approach of the UB senior management throughout the inquiry exhibited characteristics of this the retreatist style.

Theoretical implications
Our study shows that the de-autonomization of a university can happen without major de jure changes. The UB's conduct indicated a co-produced model of de-autonomization, departing from the 'forced model' of de-autonomization, which predominates the prior literature of university de-autonomization under democratic backsliding centered around the mechanism of autocratic legalism (Labanino & Dobbins, 2021;Levitsky & Ziblatt, 2019). Co-produced de-autonomization is consensual and does not require 'forced' reduction in the de jure autonomy; the UB leadership's conduct throughout the inquiry deviated from the proverbial notion of university leadership as a priori autonomy preservers who will defy politics (Leveille, 2006).
To explain the co-produced de-autonomization of UB observed in our study, one needs to approach the question of autonomy in contextual terms. The de facto autonomy will not simply be a function of a university's structural insulation from government, or a matter of the traditional norms, like academic integrity, that will arguably prevail in the leadership's decision-making regardless of the context in which such decision-making is taking place. University's de facto autonomy will also be a function of the wider socio-political relations, which can impose discourses that socialize-and perhaps discipline-actors (the academic leaders in question), thus determining how much freedom they will have in deploying their institutional powers in critical situations.
The post-2012 period in Serbia was an era of ever-increasing authoritarian tendencies where the regime continuously amassed its power. In its early stage of power consolidation, the regime pursued 'exemplary' retaliation against a number of critical voices (Freedom House, 2020); as intimidation spread, self-censorship has taken root across the political and social landscape. As this process progressed, a 'regime's interests first' discourse has taken hold across the socio-political landscape, leading institutional actors to start aligning their decision-making with the regime's interests instead of following internal institutional norms and mission. Within a few years, widespread self-censorship and self-alignment with regime's interest have emerged as the predominant mode of institutional decision-making, even in cases where the In late 2015, Rector Vladimir Bumbaširević publicly suggested to the UB Council of the Legal and Economic Sciences to declare the thesis 'clean and valid', based on irrelevant developments related to a published journal article co-authored by Mali, which drew from his thesis. The article was at some point retracted from the journal, but thereafter it was re-published in the same journal with a corrigendum (Miletić, 2015). 4. Under a new Rector (2018Rector ( -2021, BU continued to 'drag its feet' Upon her election as Rector of UB in mid-2018, Ivanka Popović promised to complete the inquiry within a reasonable timeframe (N1, 2018). Still, the process took too long before FON formed a new, expert commission, which eventually produced the same conclusion, namely that Mali's thesis is valid. Instead of using the freshly adopted misconduct regulations to finalize the process, UB decided to return the case to FON for yet another review. In summer 2019, the Rectorate requested FON to revise its latest (third) report, suggesting that 'FON shall take the responsibility and complete the inquiry' (Danas, 2019c).
regime did not express interest and where an institution in question was not under pressure or threat. A shift has thus occurred where institutions preventatively renounce using their powers to pursue their mission and institutional norms if it is appraised that those could harm the regime. Thus, although it did not hold hierarchical control or coercive instruments over UB, the regime nonetheless exerted power over its conduct through the imposed discourse of 'regime's interests first'. Memories of early regime's retaliation against critical voices have reinforced this discourse (Hayward, 1998: 20), leading actors to continue obeying a higher authority's will even after explicit coercion or regime's requests are no longer present. One's loss of de facto autonomy does not necessarily come through direct control and coercion (Foucault, 1998: 63); its autonomy can be subdued through prevalent discourses that dictate the established power relations (Foucault, 1978). In other words, the UB leadership operated in an environment where the notion of the regime's disapproval of unfavorable decisions might have disciplined it in Foucauldian terms (Foucault, 1978(Foucault, , 1991. Hence, it co-produced its own de-autonomization without being coerced into it. Yet, the UB's de-autonomization constitutes one part of the observed story. As we have seen, UB was eventually re-autonomized, i.e. uncaptured, thanks to a counter-mobilization by the academic community. This suggests that power is not fixed but is rather in flux (Sawicki, 1991). The political source of power-even when the regime is authoritarian-is not the only source shaping societal power relations; power 'resides everywhere' (Foucault, 1978) and 'circulates through discourses between and among individuals and groups' (Sawicki, 1991). Power can be produced and enhanced through joint action and association, as demonstrated by the bottom-up mobilization of the academic community that eventually prevailed over the regime's disciplining discourse 'tipping' the UB leadership back towards the autonomous mode of conduct. Thus, power is not necessarily hierarchical. Even during democratic backsliding, when an institution has become subdued to political power, this power relation can be contested through bottom-up mobilization (Hayward, 1998), and this could have a 'restoring' effect on the de facto autonomy. Thus, power is not a finite resource; alongside being producible, it can be shared and increased through social interaction (Astin & Leland, 1991: 1). In summary, it could be said UB faced constraints on its freedom (to use its formally granted autonomy) that are social in origin, while, on the other hand, such constraints turned out remediable (Hayward & Lukes, 2008) through agency.
The above observations indicate two theoretical implications about the nature of political capture in contexts of democratic backsliding. First, institutional capture can extend beyond areas of public law; it can 'mar' the operation of private regulation too. This indicates that capture can go farther than conventionally thought, even when the area in question relates to academic autonomy and academic integrity, which is traditionally seen as a 'no-go' area for politics. The disciplinary logic of power (Foucault, 1991) makes capture possible even when coercive instruments are not present, and this logic is not constrained to public law only.
Second, as UB's 'uncapturing' demonstrates, institutional capture could be reversed during democratic backsliding. So far, the prevalent perception has been that in contexts of democratic backsliding, political capture is irreversible. The thinking has been that, before a regime turnover happens, capture cannot be reversed; the institutions will continue operating with low de facto independence from politics until a macro-political change. However, as we have seen, power is not 'fixed'; once established, power relations could be changed. Despite the growing power concentration in the regime's hands, it turned out that power could also be present, produced, and expanded in other sections of the society too, and this can precipitate change in the wider power relations which can lead to the reversal of institutional capture, as demonstrated by the UB's eventual U-turn that was forced by the academic community's mobilization.

Conclusion
Mali's doctorate plagiarism investigation featured an unusual dynamic of non-decisionmaking (Bachrach & Baratz, 1970), which was ceased after a long delay when academic counter-mobilization forced the retraction of the doctorate. For the most part, the UB leadership led the process through a mobilization of bias (Schattschneider, 1960), showing low de facto autonomy from politics. This was despite the fact that UB enjoyed high de jure autonomy without being hit by 'autocratic legalism', the mechanism that regimes usually resort to in contexts of democratic backsliding to lower a university's de jure autonomy (Labanino & Dobbins, 2021). The UB leadership co-produced its deautonomization on voluntary terms, showing that academic leaders will not always act as politics-defying autonomy preservers (Enyedi, 2018).
The analysis points to two main findings relating to the nature of political capture in contexts of democratic backsliding, both findings going against the usual conception of political capture. Firstly, political capture, i.e. the subordination of institutional autonomy to political interests, can extend beyond public law and reach into areas of private institutional regulation too, and this could be achieved non-coercively. Secondly, in contexts of democratic backsliding, political capture is not necessarily irreversible; an institution can be 'freed' even without a regime change. Overall, it is possible that the pendulum of capture 'swings' farther than conventionally thought, but it can also be 'swung' back.
For explaining why and when capture could go in one direction or another, theory of power is needed (Foucault, 1978(Foucault, , 1991Hayward, 1998;Lukes, 1986;Sawicki, 1991) to nuance the overarching framework comprising of the neo-institutional theory of structural institutional insulation, the theory of democratic backsliding, and the theory of autonomy-preserving university leadership. The structural setup of institutions and their professional values and traditions can provide them with more or less potential to pursue autonomous conduct; however, it is ongoing power struggles in the wider socio-political landscape that can eventually sway such institutions-including universities-towards more or less autonomous conduct. DissertaƟon defended.

October 2014
Commission for ExaminaƟon of the Doctorate FON Same as above.
Minor omissions in the use of literature, but the PhD is valid.

December 2014
Council for Legal and Economic Sciences UB 23 members elected from the ranks of academic staff of UB.
Rejects the report of FON (11 out of 23 votes for the report).

UB adopts a Code for Professional Ethics, and returns the review process back to FON.
December 2016 Ethical Commission FON Two professors, two administraƟve staff, and one student representaƟve, all from FON.
Minor omissions in the use of literature, but the PhD is valid.

January 2017
Senate UB Rector, Vice-Rectors, Deans, RepresentaƟves of the specialized councils and insƟtutes of UB.
Rejects the latest report of FON-a because of procedural fallacies.
March 2019 Expert Commission FON Two external professors from Serbia, one from Bosnia, and one from Slovenia. 6.97% plagiarized, but the PhD is valid.

April 2019
Ethical Commission FON Five academic staff from FON. Adopts the above report of the Expert Commission.

May 2019
Prof. Karapandža published further examples of plagiarism in Mali's thesis; four appeals submiƩed to UB against the latest FON report.

July 2019
Council for Professional Ethics UB 23 members elected from the ranks of academic staff of UB.
The report of FON unclear, incomplete, featuring contradictory opinions.
The case returned again to FON.

September 2019
Blockade: a group of 20 students occupied the Rectorate; aŌer two weeks, an agreement was reached with the Rector that UB will solve the case by the 4 th November 2019.

October 2019
Expert Commission FON Two non-FON professors from Serbia and two professors from the region (Bosnia; Slovenia).
Reiterates the conclusion from the prior report: "minor citaƟon omissions, but the PhD is valid".

October 2019
Ethical Commission FON Two non-FON professors from Serbia and two professors from the region (Bosnia; Slovenia).
Recommends a public warning to Mali because of unethical conduct, which was adopted by the Board of Studies of FON.