What Can We Learn About Regulatory Agencies and Regulated Parties from the Empirical Study of Judicial Review of Regulatory Agencies’ Decisions? The Case of Croatia

: The purpose of this paper is to examine regulatory agencies and regulated parties in an empirical study of administrative disputes initiated against the decisions of regulatory agencies in Croatia. We first aim to provide an overview of the status and trend estimates regarding these disputes; second, to answer the question how well does the system work from the perspectives of both the plaintiffs and the regulatory agencies; third, to identify the problem areas and to compare these with problem areas identified by the authors studying the broader area of administrative judiciary in Croatia, and finally to compare efficiency level of regulatory agencies to other public authorities in confirming the legality of their decisions and actions. Data on all administrative disputes against 12 Croatian regulatory agencies’ decisions in the 17-year period between 1995 and 2011 are used to identify the main characteristics and trends relating to these disputes. Data for 2012 to 2013 was also examined to identify initial changes and emerging trends in the new administrative judiciary system resulting from fundamental legal reform as part of Croatia’s process of accession to the European Union in 2013. The results show these administrative disputes to be often costly and timely with modest outcome for the plaintiff and impressive success rate for the most of regulatory agencies. METHOD The paper deals with the empirical analysis of a narrow segment of administrative judiciary, i.e. the processing and analysis of the available quantitative data on initiated ADADRAs in a 17-year period from 1995 to 2011.2 They account for only 0.36% of the total running of administrative disputes (ADs) in the specified period. The analysed sample includes ADs initiated against decisions of 12 regulatory agencies, according to the classification made by Musa (2013)3: (1) Croatian Post and Electronic Communications Agency (CPECA), (2) Croatian Financial Services Supervisory Agency (CFSSA), (3) Croatian Competition Agency (CCA), (4) Croatian National Bank (CNB), (5) Croatian Agency for Electronic Media (CAEM), (6) Croatian Energy Regulatory Agency (CERA), (7) Croatian Civil Aviation Agency (CCAA), (8) Croatian Agency for Medicinal Products and Medical Devices (CAMPMD), (9) Croatian Agency for Science and Higher Education (CASHE), (10) Croatian Rail Market Regulatory Agency (CRMRA),4 (11) Croatian Railway Safety Agency (CRSA) and (12) Croatian Agency for Quality and Accreditation in Health Care and Social Welfare (CAQAHCSW), along with their predecessors.5 The sample consists of 902 initiated ADADRAs in the 17-year period from 1995 to 2011, of which 96% were resolved until November 2013. For the purpose of this paper, regulatory agencies are ‘organisations structurally separated from the public administration, formally separated from the ministries, for conducting public affairs at the national level on permanent basis, with usually employed civil servants and founded mainly by the state budget and subject to the public law’ (Musa 2014: 112–116, elaborated from: Pollitt et al. 2004, Pollitt & Talbot 2004, Christensen & Lægreid 2005, 2006, Thatcher & Stone Sweet 2002). The obtained data from the HACRC allowed, with the examination and processing of data,6 an analysis of: (1) the 300 cases at the Administrative Court’s divisions that solve cases in many different areas: financial and labour law, ownership restructuring rights, privatisation, public procurement, competition policy, property, residential, architectural, municipa, and intellectual property rights, access to information and general administration.


INTRODUCTION
Regulatory agencies (RAs) were formed as the need arose to regulate utility service providers, who evolved from the major technological innovations during the Second Industrial Revolution (Bernstein 1955;McCraw 1984). Over time, this model was adopted widely throughout the world, and extended to social regulation, including public health and environmental protection. Despite the many criticisms that their status and authority have provoked (Sunstein 1987), RAs have shown exceptional strength and staying power (Smerdel 2012).
The European Union imported RAs as a convenient model for the efficient harmonisation of European regulation and supervision between the EU institutions and the Member States (Geradin 2006;Geradin & Petit 2004). The establishment and growth of a large number of RAs in Croatia since the 1990s is related to the process of European integration. Although they have certain common characteristics, RAs in Croatia are not regulated as an institutional form, instead they have been established by laws that govern particular area (Musa 2013: 120-121).
Generally speaking, the main tasks of regulatory agencies are regulation of a specific sector and its implementation, supervision and sanctions. They are allowed a higher degree of organisational flexibility compared to conventional public authorities with a certain level of judicial powers on the grounds that the RAs must be able to quickly and effectively sanction illegal behaviour of the supervised entities in order to protect the public interest and ensure the efficient functioning of markets (OECD 2002: 85-97, Majone 1999: 3-6, 1994. This "efficient-model" comes with a price of a constant complex balancing between RA's autonomy and political control (Christensen & Laegreid 2004). Regulatory agencies may not be directly controlled, however they are held accountable, i.e. their acts and actions are subject to legal, financial and political control processes (Geradin 2006: 30-36;Bovens, 2007).
Legal control of RAs is implemented through the administrative judiciary system (AJS). The role of the AJS is to protect the rights and legal interests of citizens and legal persons against illegal acts and actions by public authorities. Geradin and Petit (2011:3-8) observe that the aforementioned legal doctrine is practiced in dogmatic spirit in continental Europe, in comparison with the United States, where academic discussion blooms with debates about the politicisation, effectiveness and bare purposefulness of the judicial review (also Tushnet 2009, Pierce 2010. Similarly, there are numerous empirical studies of judicial review of the decisions of the U.S. regulatory agencies conducted by the American scholars (Pierce 2010, Pierce andWeiss 2011), schoolwork not followed by the European peers. Pierce (2010) makes comparative analysis of 10 such empirical studies in search for factors that could explain patterns of decisions made by the courts, completing it with practical implications for lawyers, courts and academics. He finds that panel effect and ideological preferences have the most significant impact factor, institutional disadvantage only to a certain extent, while the choice of administrative law doctrines and agency's decision types and position consistency over time have insignificant impact factor on the pattern of court's judgements regarding RA's cases.
At the opening of a scientific conference at the Academy of Science and Arts dedicated to the reform of the administrative judiciary and administrative procedures, Barbic (2006: 9) expresses concern about the lack of assessment and analysis of the situation in that area: "Such complete analyses are still lacking, so we do not know the exact number of initiated administrative procedures and areas of public administration in which they are initiated, the number and types of decisions reached, as well as the share and type of lodged legal remedies and finally, the success rate. The only reliable data is that the plaintiffs succeed in 44% of initiated administrative disputes." EU's technical assistance specialists produced a significant amount of written expertise during the Croatian AJS reform, including a strategy paper and impact assessment, both lacking in quantitative analyses (CARDS 2004(CARDS , 2004a. The Croatian legal tradition of the administrative judiciary dates back to the time of the Austro-Hungarian Empire in the last quarter of the nineteenth century and has undergone numerous changes during the politically turbulent twentieth century (Medvedovic 2003). Judicial review of administrative acts adopted by public authorities is considered to be one of the basic principles of freedom and human rights (the Constitution of the Republic of Croatia (RoC), Art. 19). After almost six decades, Croatian AJS underwent a fundamental reform as part of Croatia's process of accession to the European Union (July 2013). On January 1, 2012, a new Administrative Disputes Act [ADA] (Official Gazette (OG) 20/10, 143/12, 152/14) came into force, introducing a two tier administrative judiciary model. This paper provides empirical analysis of judicial review of regulatory agencies' administrative acts, i.e. initiated administrative disputes against the decisions of the regulatory agencies (ADADRAs). Data on all ADADRAs during the 17-year period between 1995 and 2011 under the old AJS have been analysed to identify the main characteristics and trends relating to these disputes.1 Data for 2012 to 2013 was also examined to identify initial changes and emerging trends in the new AJS resulting from changes in legal framework due to Croatia's accession to the EU.
The aims of the paper are: first, to provide an overview of the status and trend estimates of ADADRAs; second, to identify problem areas and to compare these with the problem areas identified by the authors who are engaged in a broader study of Croatian administrative judiciary; third, to discuss how well the system works from the perspectives of both the plaintiffs and the regulatory agencies, and finally, to compare the efficiency level of RAs to other public authorities in judicial review of legality of their decisions and actions. The aim of the authors was also to compare this research to other CEE and Western European countries, however, our database search found no such research for the European countries published in English in the international journals. As an introduction, the authors provide a brief overview of: 'agencification' process, issues concerning public control of the RAs and other empirical studies of ADADRAs, as well as Croatian AJS. The following section introduces the reader to the methodology of the empirical study. The results include general synopsis of structure and trends regarding ADADRAs and pros and cons of the AJS from the perspective of the plaintiffs and defendants (RAs). The final section provides a summary of the main conclusions.

METHOD
The paper deals with the empirical analysis of a narrow segment of administrative judiciary, i.e. the processing and analysis of the available quantitative data on initiated ADADRAs in a 17-year period from 1995 to 2011.2 They account for only 0.36% of the total running of administrative disputes (ADs) in the specified period.
The analysed sample includes ADs initiated against decisions of 12 regulatory agencies, according to the classification made by Musa (2013) For the purpose of this paper, regulatory agencies are 'organisations structurally separated from the public administration, formally separated from the ministries, for conducting public affairs at the national level on permanent basis, with usually employed civil servants and founded mainly by the state budget and subject to the public law' (Musa 2014: 112-116, elaborated from: Pollitt et al. 2004, Pollitt & Talbot 2004, Christensen & Laegreid 2005, Thatcher & Stone Sweet 2002. The obtained data from the HACRC allowed, with the examination and processing of data,6 an analysis of: (1) the structure of ADADRAs (according to the agencies, as a share of total ADs and associated trends), (2) the structure of the plaintiffs and defendants, (3) duration of ADADRAs, (4) the structure of decisions of the ACRC and the structure of judgements according to the parties to the proceedings, and (5) the structure of initiated extraordinary legal remedies and constitutional complaints against decisions of the ACRC. Additionally, the authors have estimated approximate costs for conducting an ADADRA.
For the purpose of exploratory data analyses, the authors used descriptive statistics and linear and exponential trend models, additionally elaborated in the Appendix.

General overview of structure and trends
The observed 17-year period from 1995 to 2011 saw a total of 902 ADADRAs, which represents a negligible share (0.36%) of all the initiated ADs. Since the beginning of the analysed period, there was a trend of increase in the number of RAs in Croatia, as well as 2 The analysed period 1995-2011 was chosen for several reasons. First, the computerisation of the ACRC was introduced in 1995, enabling data search of ADADRAs by a number of variables. Second, most of the 'older' generation agencies were established around 1995, so with the exception of a few cases related to decisions of the CNB, ADADRAs started from 1995 and onwards. Thirdly, authors analysed the disputes initiated until 31 December 2011 because after that date a whole new system of administrative judiciary was installed. For the time-being, new data on ADADRAs will be fragmented at four regional first instance courts and the HACRC and consolidation of data will have to be conducted in the future. the number of ADADRAs, growing at an annual average of about 20% (Graph 1, Appendix). The small share of ADADRAs could point to the legacy of socialism, the general restraint of raising disputes against the public authorities and/or gradual development of the private sector in the areas of lobbying for legislation to their advantage.
Graph. 1: Changes in the number of ADs at the ACRC, 1995-2011.
Source: The authors, according to the data of the HACRC. Note: There is a striking increase in the number of ADs during 1998 and 1999, caused by 25,685 initiated ADs relating to pensions as a result of Constitutional Court's decision. There were treated during the coming years and contributed to a further backlog of cases (app. 75% were declared inadmissible and 22% were granted).
The majority of ADADRAs occurred in telecommunications, financial markets and competition policy, while other sectors have very few or no cases initiated (e.g. energy, civil aviation, pharmaceuticals, railroads etc.; see Graph 3). The research shows this trend is a result of the process of European integration and agencification, i.e. increasing number of RAs and the date of their establishment, great increase in EU regulation pertaining to those areas and consequently the amount of new acts and actions. The majority of the plaintiffs are legal persons, some 87% of the total, while 13% are natural persons.
Regarding decisions and judgements of ADADRAs by the ACRC, 13% were granted (in favour of the plaintiffs), 58% of the cases were dismissed, 16% declared inadmissible and 13% terminated (Graph 2). There was no right of appeal against the ACRC's decisions, but certain other actions were possible (e.g. extraordinary legal remedies and appeals to the Constitutional Court, the old ADA, OG 53/91, 9/92, 77/92, Art. 52-59; the Constitutional Act on the Constitutional Court of the RoC, OG 49/02, Art. 62).
ADADRAs last on average for two years, and 36% are resolved within a year, but half of them last for two to six years. The basic costs of ADADRAs range from €270 to €400, but may increase if extraordinary legal remedies or constitutional complaints are lodged. Each party bears its own costs (the old ADA, Art. 61). The new ADA came into force on January 1, 2012, and brought many significant changes, such as the two-tier model (the decentralisation of first-tier administrative judiciary and the right to appeal), mandatory oral and public hearings, mandatory resolution of administrative issues on substance, etc.
Available data on the new system are fragmentary, but there are signs of new trends. The analysed two-year period (2012-2013) shows that the new AJS is in an adaptation and learning phase. For now, there is a decrease in the number of ADADRAs resolved and a large number of referrals, i.e. disputes that are reassigned to other courts due to frequent legislative changes regarding the jurisdiction of the courts. Because of this, almost 30% of the "resolved" ADADRAs at first-tier courts and almost 80% at the HACRC are stuck in this "ping-pong space", which will surely increase the duration of ADADRAs in the medium-term.
This situation had the most negative effect in CPECA's cases, resulting in the resolution of the extremely low percentage of disputed (excluding reassigned cases): 4% at the regional first-tier courts and 10% at the HACRC, which is a large decline from the rates in the old system, in which these cases had the shortest average duration.
Productivity rate (measured by the percentage of settled ADADRAs, excluding reassignments) of the four regional first-tier courts differ significantly, however it will require larger timeframe before the announcement of champions. Finally, the new AJS is still in an adjustment phase and there have been few regulatory turnarounds, regarding admissibility of the appeals to the HACRC andhfrequent changes on the subject of covering costs for the parties (elaborated below).
In the new AJS, the chances of winning the ADADRA fell by half and the share of inadmissible cases is also worrisome (Graph 2). A high percentage of inadmissible cases indicates a lower level of legal protection due to restrictive interpretation of the administrative matter by the administrative court (Kopric 2006: 62), which was to a certain extent a drawback of the old system (Medvedovic 2011: 256). An additional setback was duration of inadmissible cases, mostly due to procedural inefficiencies.
Under the new ADA, the scope of administrative disputes was extended, from current assessment of the lawfulness of administrative acts, to assessment of: the lawfulness of general acts, the activities or lack of activities of public authorities and the legality of conclusion, termination and execution of administrative contracts (Art. 3).
Initial data analysis of the new AJS (2012-2013) reveals significant increase of the inadmissible ADADRA cases, reaching 43% at the first-tier courts and 25% at the HACRC (Graph 2). On one hand, increased share could partly be a positive sign of better organisational efficiency, i.e. speedy detection of inadmissible cases as a result of the reform carried out. On the other hand, it indicates lack of positive change in achieving main objectives of the reform -increasing the level of legal protection by decreasing the restrictive interpretations of administrative acts.
The previous Croatian one-tier model of administrative judiciary meant that appeals against the decisions of the ACRC were not allowed. However, one could have taken the following actions: a proposal for reopening of proceedings and constitutional complaint to the Constitutional Court of the RoC for violation of constitutional rights. Data analysis shows that during the 17-year period these actions brought extremely little success to plaintiffs.7 Regarding duration of ADADRAs, the average is approximately 2 year.8 36% of all ADADRAs were resolved within a year, while 25% of them stretch from 3 to 6 years. The shortest disputes were against decisions of CPECA, lasting on average about 17 months.9 The longest average duration of disputes was recorded at CFSSA's and CAMPMD's cases, two and a half years or longer. Another troubling factor is the percentage of cases with duration between 2-4 years (CFSSA 69%, CCA 53%, CNB 54%). These cases included financial sector and competition policy, areas of high significance to the economy and economic growth where time represents a crucial asset. In other words, justice delayed is a justice denied in these cases since it can cause incalculable and potentially irreversible damage to market participants and consequently have negative impacts on the economy.
Unreasonable duration of court proceedings represents a violation of human rights (The Convention for the Protection of Human Rights and Fundamental Freedoms, Art. 6) and can result in financial compensation for damages. The experience of the Republic of Slovenia is an interesting case study, which also had a significant impact on Croatian administrative judiciary reform. In 1997, Slovenia adopted a new legislative framework with a two-tier administrative judiciary model applicable to almost all cases, but it eventually brought the system into a crisis due to the inability to process the backlog in a reasonable time (Jerovsek 2006). In addition, in mid-2006, the European Court of Human Rights has awarded compensation for exceeding a reasonable trial time. The Slovenian Ministry of Justice prepared an estimate of the total amount of damages that could be collected -approximate €0.24 billion (ibid. p. 65). Slovenia then proposed a new ADA which would retain a two-tier model, but with a significant narrowing of the possibilities of appeal.
Subsequently, more restrictive approach to the definition of the admissibility of the appeal was advocated during the Croatian reform process (e.g. Omejec 2006: 85-86) and the Croatian legislator took the same approach when drafting the new ADA "Explanation of certain provisions", Final proposal of the ADA, Art. 66).10 Consequently, since 2012, the number of cases brought to the HACRC fell sharply and abruptly and a large number of backlog cases were resolved. A substantial fall in the number of judges and court advisors correlated with a decreasd in the number of cases (Table 1). Based on the dynamics of case resolution until 2013, simulation implies acutely reduced flow of cases and probable organisational changes at the HACRC.11 7 From a total of 863 resolved ADADRAs (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011), in 8% of cases the constitutional complaint was filed on account of the violation of human rights and fundamental freedoms. Only one judgement was adopted in favour of the plaintiff by the Constitutional Court, who overturned the judgement of the ACRC and resolved the matter himself.
Also, 90% of resolved constitutional complaints were declared inadmissible, raising concerns about unnecessary rejection of complaints due to backlog of cases at the Constitutional Court (Giunio, 2013: 43). Regarding reopening of procedures, there were only five cases (related to decisions of CPECA, CFSSA, CCA and CAEM), all with a negative outcome for the claimants.  However, such a scenario is unlikely to actually occur, because of the new amendments of ADA that just came into force on December 30, 2014, which allowed removal of 'filters' for appeals against first-tier Administrative Court rulings (Act on Amendments to the ADA, OG 152/14, Art. 19,20).12 Due to this regulatory turnaround, we are now forecasting completely opposite scenario -an abrupt and significant increase in appeals to the HACRC from 2015 onwards, which will require more judges and court advisors in the future and will likely lead to Slovenian scenario in the long term. Concluding, lessons (not) learned?
Regarding costs, they include court fees and lawyer costs (the ADA, the Act on Judicial Fees, the Tariff Schedule for Rewards and Costs for Lawyers' Work) depending on the outcome of the dispute and further actions.13 In Croatian AJS, each party covers its own costs, so in case of victory, the plaintiff must bear €270 cost for lawyers' fee and in case of loss, the cost rises to €400, which includes court fees. Renewal of proceedings would increase costs by an additional €360 and constitutional complaint costs €660.14 Since 87% of the plaintiffs are legal entities, the above figures represent only the initial costs. Larger companies will incur other costs, such as opportunity costs (e.g. employees who are engaged in gathering and analysing documents for the case, costs of management time spent), costs of external experts such as the costs of legal teams for preparation of the case, consultants and the like.
The new ADA changed the determination of costs in administrative disputes by providing for payment of costs to the winning party (the new ADA, Art. 79). However, in less than a year, the law was amended, returning to the old stipulation (Act on Amendments to the ADA, OG 143/12, Art. 17).
This represents a multi-fold increase in costs for plaintiffs: representation at hearings, court fees for appeal, composition of appeals, compilation of motions for extraordinary re-examination of the legality of confirmed judgements, compilation of motions for carrying out judgments, etc. It also evokes a decrease in legal protection since the cost of administrative dispute is an important factor in the decision whether to initiate a dispute or not. Giunio (2013: 43) and Sikic and Turudic (2013: 850-855) point out that the argumentation used by the legislator was illogical, claiming that these changes were, in fact, motivated by efforts to shore up the government budget and by distorting statistical indicators.
Wrapping up, from the perspective of the plaintiff, the chances of winning the ADADRA case are petite. The new AJS introduced a two-tier system, providing plaintiffs with one more instance of appeal. However, in practice, the outcomes for the plaintiffs have so far included higher costs and even lesser chance of success. In the future, one of the strategies for the plaintiffs might be to redirect their actions towards the alternative dispute resolutions. 12 In the opinion of the Legislator, the current system of administrative judiciary has led to "… non-uniformity of case law in equal or factually and legally similar cases, and therefore inconsistencies in the exercise of the same rights of citizens, which certainly leads to inequality of citizens and legal uncertainty." (Proposal of the Act on Amendments to the ADA, 2014:1). 13 The costs may also include the cost of expert testimony and may vary depending on the Value of the Subject of the Dispute (VSD -can be estimated or no estimated, determined by the ACRC). In ADADRA cases, expert testimony and estimated VSD represent exceptions, therefore are not included in subsequent calculations.

The regulatory agencies' (defendant's) perspective
As a general rule, the decisions of the Croatian regulatory agencies are final but may be subject to judicial review before the Administrative Court. On average, RAs win in 82% of ADADRAs initiated at the ACRC.
Graph 3 shows that 92% of initiated ADADRAs are related to the four oldest regulatory agencies in Croatia: CPECA (40%), CFSSA (23%), CCA (21%) and CNB (8%).15 The judgements of the ACRC were on average five and a half times more often in favour of aforementioned RAs, confirming the legality of their acts. Against CERA, CCAA and CAMPMD, the plaintiffs have not won a single case so far.16 The only exception is the CAEM, where there are more judgments in favour of the plaintiffs. Most of the initiated ADs against the decisions of CAEM are linked to (not)allowing concessions for television and radio services. Graph 4 presents the findings of trend movement of judgments in favour of CCA, CFSSA and CPECA.17 Until 2004, CCA succeeded in 70% of cases, but with significant variations in performance over the years, and from 2005 to 2011 it had a continuous success rate of 99%. In case of CFSSA, from 1995 to 2002 there were not enough cases needed to detect a trend, and during that period CFSSA mainly won the cases (9 out of 11). From 2003 to 2010, CFSSA recorded an upward trend of successful cases with an average annual growth rate of 3.43 percentage points.18 Until 2006, CPECA did not have enough data to identify trends. From 2006 onwards, CPECA has won 83% of the disputes, with minor variations that were not statistically significant (Appendix). Concerning ADADRAs, which were declared inadmissible or terminated, there were no observed trends.
We can conclude that since mid-2000s, there are patterns in the proportion of disputes won by the analysed agencies. These RAs, including the Croatian National Bank, show the characteristics of a learning organisation and have evolved into superior public authorities when it comes to confirmation of the legality of their decisions before the ACRC. The main characteristics and trends regarding ADADRA cases differ from ADs initiated against acts of other public authorities (e.g. ministries, local and regional government), in the quality of the actions of the regulatory agencies prior to the administrative dispute (quality of administrative procedures and acts), and by the success rate of confirmation of the legality of their decisions at the ACRC.
First, RAs have superior success rate against the plaintiffs, winning on average five and half times more often versus other public authorities score of twice as often. This has been, to a certain extent, result of relatively high number of issued administrative acts with suspicious legality by other public authorities (Medvedovic 2011: 256).
Second, implementation of court's decisions was found problematic, so plaintiffs were often forced to initiate a new AD because the respondent public authorities would again issue illegal acts or would deliver them with delay (ibid.). Unfortunately, the new AJS does not seem to address this problem efficiently because it still lacks sanctions for non-compliance with the Court's rulings (Marusic-Babic 2014: 193-199). The analysis by the type of disputes and prosecutors determined that the implementation of the Court decisions is not problematic when it comes to RAs.
Third, only 0.5% of ADADRAs were initiated with regard to (non) recognition of a party to the proceeding, which points to high quality of administrative procedure management by the RAs. This has not been the case with other public authorities (Gagro & Juric-Knezevic 2006: 23-24). Finally, the ACRC judges reported that the reasons for the dismissal of significant amount of administrative acts by some state administrative bodies was because of the "lack of knowledge about some of the most basic things of administrative procedure" (ibid.: 34).
Data analysis of the new AJS shows approximately the same level of the success rate of the RAs in judicial review at the Administrative Courts. We can conclude that through time and experience, regulatory agencies have evolved and became superior public authorities in affirming the legality of their decisions at the ACRC. These findings support the frequently used thesis about greater efficiency of regulatory agencies in comparison with other public authorities because of their specific organisational structure (highly specialised employees, more flexible organisation, independence, etc.;OECD 2002, Majone 1999, 1994; however, they are limited only to explored area of judicial review. There is one more issue to be considered here -the problem concerning specialisation of courts regarding ADADRAs. In the course of the reform of administrative judiciary, experts from RAs (in this case, CPECA) proposed the direct jurisdiction of the HACRC, raising concerns about the pace of development of specialisation and uniformity of case-law at the newly established regional administrative courts of first instance and referring to the case-law of the European Court of Human Rights and the European Court of Justice and examples of case-law of EU Member States (e.g. Hungary, France, UK), according to which RAs should be seen as first instance administrative courts (Popovic and Maricic 2014). Ultimately, in the new AJS, ADs initiated against the decisions of CPECA fell within the jurisdiction of the HACRC (Electronic Communications Act,OG 73/08,90/11,133/12,80/13,Art. 18,116;Postal Services Act,OG 144/12,153/13,Art. 11,62)19 and its example was followed by the CCA a year and a half later (Competition Act,OG 79/09,80/13,Art. 50). It is possible that, in the future, the legislator may follow the examples of CPECA and CCA, and give jurisdiction for judicial review of ADADRAs directly to the HACRC, the court with highly specialised knowledge in dealing with this type of cases.
ADADRAs represent somewhat distinctive group of cases. Statistically, they are insignificant, accounting for only 0.36% of total ADs initiated at the ACRC (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011). However, the importance of individual sectors in the whole Croatian economy and the influence of particular subjects (major Croatian companies such as Agrokor, Croatian Telecom, Atlantic, commercial banks) in both political and economical sphere is by no means easy to overlook. Also, they fall under the domain of highly regulated areas for various reasons: finalisation of the single European market through harmonisation of the EU legislation, their importance for macroeconomic stability and further increased regulation as a result of the global financial crisis and governments' efforts to prevent the appearance of a new systemic risk, new technological innovations that change the way of life, with the simultaneous need to protect the rights and safety of users, etc. Additionally, most of them belong to technologically highly sophisticated areas that are often comprehensible only to a narrow circle of experts who are dealing with them on a daily operational basis (e.g. telecom, capital markets, energy), consequently multiplying information asymmetry problem in this area.
Viewed from the operational point of view, a narrow specialisation in ADADRAs is extremely challenging task, taking into considerations all of the above presented and the fact that ADADRA comes once in 300 cases at the Administrative Court's divisions that solve cases in many different areas: financial and labour law, ownership restructuring rights, privatisation, public procurement, competition policy, property, residential, architectural, municipa, and intellectual property rights, access to information and general administration.
American scholars entitle this problem "institutional disadvantage" (Pierce 2010: 8), describing the disproportion in information, technical or economical expertise between judges and RAs experts. The empirical studies show correlation between economic or technical complexity of the ADADRA case and percentage of cases upheld by the courts in favour of the RAs (ibid).
In some EU Member States, there are specialised courts dealing with ADADRA cases (e.g. the U.K., the Netherlands), but AJSs differ greatly throughout the EU (Lavrijssen & de Visser 2006). Since these areas are becoming more and more European, being harmonised both in legal and economic sphere, in the future, the European policy-makers might even propose a concept of transferring national ADADRAs to the EU level by establishing specialised European Administrative Court dealing directly with disputes against decisions of national regulatory agencies, the idea that requires further theoretical elaboration and is outside the scope of this paper.

CONCLUSION
The establishment and growth of regulatory agencies in Croatia since the 1990s is related to the process of European integration. ADADRAs account for only 0.36% of total ADs initiated at the Administrative Court but at the same time these cases are very demanding because they belong to highly regulated and technically sophisticated areas, which are continuously becoming more complex. American scholars conducted numerous empirical research studies, which showed a correlation between demanding cases and percentage of judgements in which the courts have upheld the RAs' decisions. In our research capacity, we found no empirical research in this area neither on the EU level nor for the Western or CEE countries. Although there are significant amount of political and economical uncertainties facing the EU's future as such (e.g. Brexit, immigration, the Eurozone crisis), the ongoing tendency is still directional towards more harmonisation in economic and legal sphere (e.g. Banking and Capital Markets Union). Therefore, from the European public policy perspective, our data and trends suggest a possibility of future proposal for creation of specialised European Administrative Court for ADADRA cases coming from national RAs of the Member States.
From the perspective of the plaintiffs, they have approximately a 13% chance of success in ADADRAs that last for 2 years on average, with a basic cost totalling between €270 and €400. In recently reformed administrative judiciary system, preliminary trend results show multi-fold increase in costs for the plaintiffs and even lesser chance of success in winning the ADADRA case. The research results of this paper point the plaintiffs to turn their energy and resources towards other alternative dispute resolutions.
Regulatory agencies win in 82% of ADADRAs. Trend analysis shows RAs to have characteristics of learning organisation. Through time and experience, RAs have evolved in the areas of administrative procedures and acts and became superior to other public authorities in confirming legality of their decisions and actions at the Administrative court. These findings recommend policy improvements for other public authorities based on the RAs successful model. Finally, this paper represents just the first step in examining the judicial review of RAs' decisions in one of the EU Member States and we hope further research of other EU Member States will follow. year. Unit y= the number of judges and court advisors.). Based on the dynamics of case resolution to date, we estimate that 10,383 cases will be resolved in 2014, implying that the HACRC will carry less than 1,000 unresolved disputes into 2015. In 2015, around 14 judges would be needed to resolve current and outstanding cases. Graph 4 -CFSSA's upward trend -81.76% of the variation of share of cases won by CFSSA was explained by a linear trend model.