Location : Home > Resource > Paper > Empirical Research
HE Hong-quan | The Effectiveness of Administrative Adjudication System Reform ——Research Based on 580, 000 Judicial Documents
2023-12-25 [author] HE Hong-quan preview:

[author]HE Hong-quan


HE Hong-quan: The Effectiveness of Administrative Adjudication System Reform ——Research Based on 580, 000 Judicial Documents

Author* HE Hong-quan

Doctoral Candidate, School of Law, Tsinghua University; Research Assistant, China Institute for Socio-Legal Studies, Shanghai Jiao Tong University

Abstract: Administrative cases are generally under the jurisdiction of the court where the defendant is located, so the court trial is easily affected by the defendant, leading to the malfunction of administrative litigation. Since 2015, The current reform idea is to adjust the system of regional jurisdiction. Two schemes have been piloted; first, the jurisdiction of ordinary courts in other places; second , using the transformed railway courts to hear administrative cases. However, the reform effectiveness lacks sufficient empirical e-valuation, and which scheme is better is not clear enough. Based on the 580, 000 administrative judgment documents of first instance published by the China Judgment Document Network from 2015 to 2019, the study found that the performance of railway courts is better than that of courts in other places at the level of grass-roots courts, but the opposite is true at the level of intermediate courts. The study also found that, the railway court is more likely to make a withdrawal ruling. Although both the court in other places and the railway court have got rid of the direct influence of the defendant , it is still impossible to get rid of the hierarchical nested and interconnected tiao-kuai system. The distance between the relative positions of defendant and court in the organizational system determines the strength of defendant's ability to influence the neutrality of trial. This distance in the organizational system reflects the common principles behind the two different reform schemes and explains the reasons for their different effects at different levels.
Key Words: Administrative Adjudication System; Railway Transport Court; Administrative Court; Remote Jurisdiction; Tiao-kuai Relationship; Organizational Distance

1. Presentation of the issue

China's courts have the duty to hear administrative cases and supervise administration in accordance with the law, but the fulfillment of their duties is subject to a delicate power structure. In the system of "combination of tiao-kuai and kuai-based", the courts are subject to the "tiao" management of the higher courts and to the "kuai" leadership of the local party committees at the same level (and, to a certain extent, of the government as well). Under the jurisdictional system set out in the The Administrative Litigation Law of the People's Republic of China (hereinafter referred to as the Administrative Litigation Law), the vast majority of administrative cases are heard by the court of the place where the administrative agency is located. This makes it difficult for the local courts to adjudicate administrative cases neutrally, and the functions of the Administrative Litigation Law in resolving disputes and supervising the administration are thus greatly undermined. The question of how to reform the administrative trial system and enhance trial neutrality was once considered the "No. 1 issue" in the reform of administrative litigation.

On the premise that the court establishment remains basically unchanged, the main way to enhance trial neutrality is to take administrative cases away from local courts and give them to other courts for trial. After a long period of discussion, there are now two main options: first, to have administrative cases heard by an ordinary court of the same level outside the administrative division in which the defendant's government is located, that is, the "court in other places option". The second is to rely on the railway courts, which were converted in 2010, to set up cross-administrative courts to hear administrative cases. Many scholars advocate the establishment of a Chinese administrative court on this basis. In this paper, the second option is referred to as the "railway court option"; in 2015, in accordance with the relevant provisions of the judicial reform, each province could determine, in the light of local circumstances, whether administrative cases should be brought under the centralized jurisdiction of a court in other places or a railway court.

The question is, after so many years of reforming the administrative trial system, which option has worked better? Theoretical debates predate the reforms on the ground, but so far there is still a lack of a convincing and clear conclusion. Some empirical studies have compared the effectiveness of administrative cases heard in courts in other places or railway courts compared to local courts. However, the current study still has regrets: Firstly, it does not answer which option is more effective. Secondly, the method of judging which reform program is better by the results of previous studies is also unreliable. Because their samples and benchmarks are different in time, scope and type, the comparison results may be as difficult as "Guan Gong versus Qin Qiong". Thirdly, the classification is too coarse and the accuracy of the conclusions needs to be improved. For example, in the dichotomy of "railway courts or non-railway courts", there is no distinction between the two categories of "non-railway courts", namely, courts in other places and local courts, which may lead to bias in the conclusions. It can be seen that the concrete effects and differences of the reform program still need to be clarified, and if the stone is not clearly felt, a hasty crossing of the river will lead to a fall.

This paper uses 580, 000 first-instance judgement documents made public by the China Judgement Documents Network for the period 2015-2019 to assess the effects of the two reform programs. The results show that the two reform programs perform differently at different levels of courts. In the basic courts, the role of the railway courts in ruling in favour of the plaintiff is stronger than in the courts in other places; in the intermediate courts, the opposite is true. In terms of decision and judgement tendencies, railway courts are more likely to make a dismissal decision, while courts in other places are not. This empirical result differs from the previous theory that one reform program is better than the other. This paper argues that courts in other places and railway courts remain in a hierarchical nested and interconnected tiao-kuai system, where the ability of defendants to influence the courts is not a matter of presence or absence, but rather weakens with organisational distance. This paper will use "organisational distance" to describe the "proximity" of the defendant and the court in the organisational system, and explain the empirical results of this paper on this basis.

2. Research design

2.1 Research ideas

The idea of this paper is to examine the differences in the outcomes of the railway courts and courts in other places compared to the local courts over the same period of time:

Firstly, a comparison between reformed and non-reformed courts over the same period of time. There are two ways of comparing "reformed or not" and "before and after reform" in terms of assessment dimensions. Both have advantages and disadvantages, and the article uses the former. Although the latter is theoretically able to eliminate differences between courts, judges and other factors, it is in fact a comparison between two different groups of courts, as some courts did not hear the corresponding administrative cases before the reform. And, again, this makes it difficult to control for the impact of changes in other factors on adjudication before and after the reform. If this were to be done, either the sample would be narrowed by selecting only a small number of places, or a large number of variables would be added to the 580, 000 decisions. In the former case, the reliability of the findings would still be questionable, and in the latter case, the workload would be enormous and omissions would still be unavoidable. On the contrary, a comparison of reforms and non-reforms over the same period of time can exclude the effects of laws, policies, and so forth, before and after their implementation. The problem with this approach is how to control for differences between reformed and non-reformed courts, and again there is a risk of omitting some potential variables, but these differences can be controlled for as much as possible by adding variables.

Secondly, the local courts are used as a benchmark to measure the effectiveness of the decisions of the courts in other places and the railway courts. No matter what the reform program is, it will be compared with the local courts in hearing local cases. Therefore, this article will first measure the effectiveness of the courts in other places and the railway courts compared to the local courts. Afterwards, the results of the two will be compared using the local courts as a common reference system. If the opposite is true, then the empirical findings of the empirical study will not be the same if different reference objects are chosen, and if the courts in other places and the railway courts are compared directly with each other, it will lead to either better or worse results. This does not reflect the full picture of the reforms, as it is likely that both have had good results compared to the ordinary courts, but one has been slightly less successful than the other.

Thirdly, the effectiveness of the reform is measured by the outcome of the judgement. The core concern of the reform is whether the neutrality of the courts can be enhanced. The outcome of the judgement is the conclusion of the court's comprehensive review of the legality of the government's actions, which centrally reflects the trial of the case as a whole. The current interference of the sued administrative organ in the court trial is mainly manifested in the difficulty for the plaintiff to win the case and the low rate of winning the case. Therefore, if the reform programme improves the probability of plaintiffs winning their cases, this paper argues that the reform has had a positive effect.

Fourth, adjudication outcomes include first-instance judgements and rulings. In addition to judgements, about half of all administrative cases are concluded by adjudication. Previous studies have neglected adjudication. In addition to the result of government intervention in which the plaintiff loses the case irregularly, the case may also be excluded from substantive judicial review through irregular dismissal, thus dissipating the effect of the reform. For example, judicial policy has called for an increase in the number of cases filed, which has resulted in an increase in the number of cases filed, but more than half of the cases do not reach the courts substantively, but are withdrawn from the judicial process by a dismissal decision. Thus, if adjudication is omitted from the measurement of the effects of reform, the assessment may be incomplete.

2.2 Data processing

The data in this paper comes from 580,000 first-instance referee documents made public by China Judgment Document Network during 2015-2019. Through machine and manual identification, more than 40 features of the adjudication documents, such as trial courts, administrative acts, trial procedures, adjudication results, and litigation participants, were extracted. The sample pool excludes intellectual property administrative cases. These cases are generally under the jurisdiction of specialised courts, the defendant is mostly the State Intellectual Property Office or the Trademark Office, the threshold of case specialisation is high, and there are mainly civil disputes behind them. Although the defendant is the government, there is less government intervention. If intellectual property administrative cases are not excluded, the assessment results are rather not relevant. After that, the author collected the specific programmes of administrative trial system reform of 28 high courts, and based on this, marked on the variable table.

The manual comparison of 1‰ randomly selected original instruments with the variable table found that there was no systematic bias in data processing. The average plaintiffs' success rate in all years of this data is about 7 per cent higher than that of the Judicial Statistics Bulletin of the Supreme People's Court. This bias may be related to the rules for online disclosure of judgement documents, where certain rules prevent a portion of plaintiffs' unsuccessful documents from being uploaded to the China Judgment Document Network. This bias belongs to the external fixed bias, which indicates that the plaintiffs' winning situation is even less optimistic in practice. Since the article uses local court decisions as the reference object and adopts a relative effect criterion, this kind of fixed bias will not have a subversive effect on the study's conclusions.

2.3 Variables and Methodology

The dependent variable in this paper is the outcome of the judgement. There are two types of judgements: those in favour of the plaintiff or those against the plaintiff. The judgements in favour of the plaintiff are: annulment, alteration, performance of statutory duties, confirmation of illegality or invalidity, and compensation. If the court makes one of the aforementioned judgements, even if it does not support all of the plaintiff's claims, it is marked as a win. Because administrative litigation is the review of the legality of the administrative act, regardless of whether the plaintiff is satisfied, the aforementioned judgement denied the legality of the administrative act. The plaintiff lost the judgement has maintained the administrative act, dismissed the claim. The rulings are slightly more complex: dismissal of the lawsuit, inadmissibility is regarded as the plaintiff's defeat; other rulings such as withdrawal of the lawsuit, which do not involve the victory or defeat of any party, are regarded as neutral rulings. Details are shown in Table 1:

The independent variable in this paper is the trial court. There are three types of trial courts: local courts, courts in other places, and railway courts. The number of cases heard in each type of court and the average success rate of plaintiffs are shown in Table 2. Due to the extremely large number of classifications of administrative domains and types of administrative behaviour, this paper recodes them into two variables: scope of administration and type of behaviour. Their correspondence with specific administrative management fields and administrative behaviours is shown in Table 3. If a case involves more than one administrative field, the title of the adjudication document prevails.

The diagnosis of multicollinearity revealed that the variables type of court, location of the court and the defendant, year of filing and decision, and identity of the third party were subject to the problem of covariance. After eliminating the variables causing the covariance problem, the problem was solved and the variables retained are shown in Table 4. The reform also involves changes in the number of judicial districts. The courts in other places and the railway court programme may concentrate on cases from multiple administrative districts. The pressure of disruption on the courts from different numbers of judicial districts may be different. This was easily overlooked in the past. In addition, local socio-economic variables have been included to control for the impact of other factors on adjudication outcomes.

As the dependent variable adjudication outcome is a categorical variable, not a continuous and measurable value, based on statistical theory, this paper uses a Logit model fit in two parts: firstly regressing on the first instance judgement, and later adding the adjudication to it. With the addition of the adjudication, the dependent variable will change from a binary categorical variable to a ternary categorical variable, hence the use of Logit model extended regression. Equation (1) is the binary Logit model and Equation (2) is the multivariate Logit model used with the addition of adjudication to the dependent variable.

βxCourt is the regression coefficient for the independent variable trial court, which indicates the effect of court on the probability of occurrence of the event that the plaintiff wins the case. For example, β1 > 0, Court = Railway Court, which indicates that when the trial court is a railway court, it increases the probability of the occurrence of the plaintiff winning the case; and vice versa, it leads to a decrease in the probability of the occurrence of the event. Note that the Logit regression coefficients reflect the relative magnitude of the effect of the independent variable, not the absolute magnitude. The effect of a railway court or courts in other places on the probability of occurrence of a plaintiff's win is relative to local courts.

3. Empirical findings

3.1 Key findings

This section reports the empirical findings from the grassroots to intermediate courts and from judgements to decisions. Table 5 presents the regression results for judgements in grass-roots courts, using local courts as the reference category, and observing the effect of cases heard in other places or railway courts on judgement outcomes. In column (1) of Table 5, the regression coefficients are positive and significant for both foreign courts and railway courts. This suggests that cases heard in either field courts or railway courts are more likely to be decided in favour of the plaintiffs as compared to local grass-roots courts. Column (2) adds fixed effects to (1) and the results are the same as in column (1), with increased significance for railway courts.

[Note: (1) The reference category is local courts. (2) * stands for p < 0.1, indicating that the results are relatively significant; ** stands for p < 0.05, indicating that the results are quite significant; *** stands for p < 0.01, indicating that the results are highly significant. (3) The closer the R2 is to 1, the better the model fits, but the upper limit of the value in the logit model is uncertain and of little significance; when more than one model exists, reference can be made to compare the size of the goodness of fit. Same as below].

In order to further measure the difference in the strength of the impact of the railway courts and courts in other places on the outcome of the judgement, this paper standardises the regression coefficients of Column (2) using a formula calculated by Professor Guo Zhigang of Peking University through the PROC LOGISTIC program of SAS. The reason for using this formula is that fully standardised regression coefficients vary with the variance estimate of logit, which makes the standardisation of the same variable change in different models, but the formula does not have the above problem. The standardised regression coefficients for the grass-roots courts in other places and the railway courts were calculated to be 0.040 and 0.052, respectively.This shows that the relative role of the railway grass-roots courts in the plaintiff's success is greater than that of the grass-roots courts in other places.

Table 6 examines the differences in judgements between provincial and non-provincial capital railway grass-roots courts. There are 10 provinces where both provincial and non-provincial capital railway courts exist: Guangxi, Henan, Hubei, Hunan, Jilin, Jiangsu, Liaoning, Shandong, Shaanxi, and Yunnan. In Column (3), the regression coefficient for whether or not there are provincial capital railway grass-roots courts is not significant, which suggests that whether or not there is a provincial capital railway grass-roots court has no statistical effect on judgements.