Introduction
China's Unexpected Turn to Court Mediation
Few Chinese legal scholars in the 1990s and early 2000s would have predicted the immense surge of mediation in Chinese courts that has occurred since 2004. In this period, the mediation rate (Tiaochelü)Footnote 1 surged from around 30 percent in 2002–2004 to 67.3 percent in 2011, 68.2 percent in 2012, and 63.1 percent in 2013 (Ma Reference Ma2015). In different locations rates were even higher. For example, in Henan Province the mediation rate in 2010 increased by 48.7 percent compared with 2009, and reached 75.6 percent. In Jiangsu Province, the mediation and case dropping rate of labor cases increased by 37.71 percent compared with 2006, reaching 76.21 percent in 2010 (Jiangsu Higher People's Court 2011). During the first quarter of 2010, lower level courts that published their experience on mediation all reported a mediation rate over 80 percent (Pan and Liu Reference Pan and Liu2010). Some lower level courts reported a mediation rate exceeding 90 percent (Zhang Reference Zhang2010).
Although mediation had been the guiding principle in civil judicial work from the 1950s to the 1980s (Cohen Reference Cohen1966; Lubman Reference Lubman1967; Lubman Reference Lubman1997; Diamant Reference Diamant2000), since 1988 the Supreme People's Court (hereafter “the SPC”) officially initiated a civil trial reform. Judicial reformers hoped to build a modern and formal legal system characterized by formal trials and strict evidence rules to handle the booming number of civil cases (Fu and Cullen Reference Fu, Cullen, Margaret and Gallagher2011). A key component of this reform was legislation intended to limit the role of mediation and to prevent the abuse of mediation common to the judicial process. After the civil trial reform took place, the court mediation rate dropped sharply in both actual numbers and in percentage terms between 1990 to 2002.Footnote 2 Since 2004, mediation has grown rapidly, spurred by national level policies from the SPC and falling in line with national political ideology preaching harmonious relationships in society and emphasizing less the formal rationality that the party promoted under its governance based on law (yifa zhiguo) of the 1990s and early 2000s.
Many scholars have strongly criticized the return of court mediation (Su and He Reference Su and He2010; Fu and Cullen Reference Fu, Cullen, Margaret and Gallagher2011; Liebman Reference Liebman, Heilmann and Perry2011b; Minzner Reference Minzner2011). Authors fear that the turn toward mediation marks a return to a popularized justice in which formal rational thinking and sound legal decisions take a back seat. Scholars also warn that the strong growth in mediation rates evidence a new political interference in basic court processes. According to these authors, the change is part of a top-down authoritarian trend motivated by social stability concerns meant to prevent disputes from escalating (Minzner Reference Minzner2011), and thus used as evidence of the Chinese legal system moving “against law.” In this way, the existing literature attributes the rise of mediation to the purpose of courts keeping social stability.
Most existing studies have taken the published mediation rates at face value. First of all they largely do not question whether the reported higher mediation rates reflect actual mediation work in the courts. Second, they do not sufficiently differentiate between the different types of judicial activities in which the mediation rate has grown. Many existing studies simply have not examined the courts to see what mediation occurs, or where and how the mediation rate numbers are produced in the court.
Thus there is a strong need to understand the actual reality of mediation practices and understand how the mediation rates are produced. First we need to understand exactly in which types of judicial work mediation takes place, and what form of mediation actually takes place there. Judges in China do much more than mere adjudication; they also deal, for instance, with petitions, manage protests on the street, file cases, and execute judgments. Moreover, there may be differences in actual substantive mediation or judicial processes that are mediation in name but not in actual practice.
Second, and most crucially here, we need to understand how the mediation rate numbers are produced in Chinese courts. Accompanying the surge in mediation rate, the SPC implemented a judicial cadre performance assessment system to strengthen its control over the lower courts. The impact of this system on mediation has been neglected in the existing literature. We know that in other parts of the Chinese bureaucracy—for instance, in the field of environmental protection—performance reporting for cadre evaluation systems has resulted in shirking, game playing, and false reporting (Wang Reference Wang2013; Shi and Van Rooij Reference Shi and Van Rooij2014; Kostka Reference Kostka2015). Moreover, the simple numbers might hide large variation in mediation practices that we should not understand as one category. We thus need to understand what the court mediation rates actually mean and how the rate has been produced in the local court setting.
This article seeks to open up the black box of the mediation rates and understand how reported mediation rates in the courts are produced and what they actually signify. It will do so through an in-depth study of one local court.
Accordingly, the article finds that the surge in the court mediation rate does not represent a singular judicial mediation phenomenon. In the court studied, mediation rate increases at first were largely the result of low-level court involvement in mediation practices outside of the courtroom. This strongly affects our understanding of the meaning of court mediation rate increases, especially those before 2011, as they may well have little to do with the cases filed in the court. The article further finds that in the court studied the mediation rate was produced strategically by seeking the highest increase possible under the discretion of the definition of mediation in the incentive structure with the lowest cost and risk to the court. This undermined the objectives of the SPC incentives toward mediation, while also drawing away the courts' scarce resources toward unnecessary mediation practices. The article concludes by drawing out broader theoretical conclusions about how information asymmetries, discretion, and China's authoritarian system affect vertical coordination within the Chinese judiciary.
Theoretical Perspectives
To understand the meaning of the surge in the reported mediation rates, we must confront the hierarchical control structure of the Chinese courts. Most scholars have understood the mediation rates as simply meaning that lower courts have fully responded to superior court policies incentivizing mediation over adjudication (Liebman Reference Liebman, Woo and Gallagher2011a; Minzner Reference Minzner2011). Of course, they have good reason to do so, given that “core compliance” by lower courts with regime interests are noted in dozens of authoritarian states (Ginsburg and Moustafa Reference Ginsburg, Moustafa, Ginsburg and Moustafa2008). In China's authoritarian system courts have been set up in a strong hierarchical structure that demands loyalty to higher level policies and directives and allows for vertical control and full Party oversight (L. Li, Reference Li2015).
However, social scientists have developed different yet interrelated ideas that force us to look more critically at how courts as organizational units respond to hierarchical control. Principal–agent theory, as developed by economists and political scientists, theorizes hierarchical control of organizations (Moe Reference Moe1984; Sappington Reference Sappington1991; Waterman and Meier Reference Waterman and Meier1998). The superior principal seeks to alter behavior in the subordinate agent. The principal does so by incentivizing the desired behavior. Principal–agent scholars have shown that there are severe challenges in the implementation of these incentives in bureaucratic control because of information asymmetries and goal conflict between the principals and agents (Eisenhardt Reference Eisenhardt1989; Waterman and Meier Reference Waterman and Meier1998; Shapiro Reference Shapiro2005). This creates an opportunity for the agent to shirk in its duties but still get a positive incentive. This has been called a “principal–agent problem.”
Songer and colleagues have applied principal–agent theory to study interactions between the US Supreme Court and directly subordinate federal circuit courts. They view the Supreme Court as the principal who seeks to direct the work of circuit courts through incentives and monitoring. In their principal–agent analysis they found that, whereas on the one hand circuit courts were highly responsive to the Supreme Court, lower judges did find opportunities to shirk “to satisfy their own political interests” (Songer, Segal, and Cameron Reference Songer, Segal and Cameron1994; Posner Reference Posner2004).
This article was inspired by the work of Songer and colleagues, who theorized that the surge in mediation has occurred within the frame of a principal–agent relation in which the SPC, as a principal adopting incentives and indicators, stimulates mediation behavior in its agents, the local courts. Following the principal–agent theory, we must look further at the opportunities local courts have over control of information and to what extent they use these opportunities to shirk in mediation practice and cheat in generating mediation rates.
Discretion is key within the principal–agent relationship. The rules a principal uses to incentivize agents can have different levels of discretion in them, while in actual operation, even if the rules allow for limits, discretion agents can enjoy a wide de facto discretion because there is limited external oversight on their work (Davis Reference Davis1969; Lloyd-Bostock Reference Lloyd‐Bostock1992). Local agents may use their discretion to resist or reshape any of the principal's policies that may be unfavorable or impractical for them (Talesh Reference Talesh2009). In the study of regulatory enforcement, Hutter, for instance, showed that such local-level enforcement officials have used discretion in the law to redefine and operationalize unfavorable and unfeasible legal rules in a way that best fits their work conditions (Hutter Reference Hutter1997). Accordingly, we need to examine the rule-based and de facto discretion local courts have in responding to the SPC mediation-oriented incentive systems (Davis Reference Davis1969; Adler and Asquith Reference Adler, Asquith, Adler and Asquith1981; Van Rooij Reference Van Rooij2006). We need to understand also how local courts use the discretion they have in reshaping and redefining the rules of the incentive system and how this shapes their actual mediation work.
Within the principal–agent incentive-based relationship, moreover, the primary goals of the principal can become displaced (Merton Reference Merton1940). This can especially happen when the principal introduces an incentive structure that rewards the achievement of a certain target. Agents may then make reaching the target their goal, even though doing so may not fulfill the original primary goal the incentive structure was set up for (Wilson Reference Wilson1989). Traffic police, for instance, serve the primary goal of enhancing safety and compliance with traffic law. The introduction of a target for number of citations may easily lead a high volume of easily produced cases, without necessarily addressing those traffic violations that have the biggest impact on safety. Another form of goal displacement happens when, in response to the principal's incentive structures, agents shift their focus only toward the measured aspects and away from other parts of their work outside of these measurements (Brodkin Reference Brodkin1986; Brodkin Reference Brodkin, Calista and Palumbo1990; Brodkin Reference Brodkin1997; Lipsky Reference Lipsky2010).
A related challenge for the principal is to ensure that the agent does not engage in so-called symbolic or creative compliance, where the agent may comply with the letter of the principal's demands but not its spirit (McBarnet Reference McBarnet1988; Gray Reference Gray2006; Gray and Silbey Reference Gray and Silbey2014). This is especially hard when principals do not have sufficient information about what agents actually do and have to rely on agents reporting summaries of their work through statistics or examples.
This body of work forces us to look at how the mediation performance in the local court actually meets the primary SPC goals of enhancing mediation, or whether some form of goal displacement has occurred where either court has sought to formally produce target mediation numbers without actually doing more or better mediation in their judicial work. Moreover, we need to understand whether the incentive system has had the unintended consequence of drawing judges toward creating higher mediation numbers to the detriment of other aspects of their job.
Methodology
To study how mediation takes place in local courts and how mediation rates are produced, this article utilized data collected during field researchFootnote 3 in China during 2011–2013. The data were collected for two years at a single court (herein called Court M). This study focused on how this court handled labor disputes. Labor disputes were selected because they allow for a broad understanding of diverse mediation practices. Labor cases make up the largest volume of civil cases in Court M, and also the highest number of mediation cases. In labor work, a wide variety of judicial work can also be studied—not only traditionally adjudicative work in the court room but also outside the court on the streets, where the judges are supposed to deal with labor-related conflicts (Su and He Reference Su and He2010), practicing “on site” mediation. Labor cases are also interesting because of their direct potential to cause social unrest (Fu and Choy Reference Fu and Choy2004; Shen Reference Shen2008; Fu and Peerenboom Reference Fu, Peerenboom and Peerenboom2009), and because of the high workload and pressures they create for courts (Michelson Reference Michelson2008).
Court M is located in the Yangzi River Delta, in a district that has been urbanized and industrialized over the past fifteen years. Because many factories are located in the court's jurisdiction, labor disputes make up a significant percentage of cases in that court. For instance, 13,000 new civil disputes were filed in the year 2012, among which labor disputes comprised about 7 percent. With about 170 persons on permanent staff, including ninety judges, the caseload is very high. For court mediation, a high mediation rate has been reported: 74 percent in 2011 and 60 percent in 2012. Although 85 percent of judges hold a bachelor's degree or higher, as a court within an industrial area Court M is neither cutting-edge nor daring.
Court M was thus selected because it has a high reported mediation rate. The high mediation rate makes it suitable for finding diversified practices of trial judges doing mediation. Practically, Court M also allowed for unique access to conduct this fieldwork.
The actual fieldwork was conducted through an intensive participant observation. The researcher (first author) spent six months gathering data in the courthouse (two months in 2011 and four months in 2013). Access was given to some internal reports and court statistics. More importantly, she was allowed to sit in during proceedings to directly observe the work of judges and officials on a daily basis. The researcher could thus directly observe how cases were dealt with in the courthouse, from the moment the court was involved in the dispute resolution process to formal litigation. To understand the actual mediation processes, the researcher participated, with the disputants' permission, in mediation sessions. After each observed session, the researcher conducted follow-up interviews discussing the process with the disputants, lawyers, judges, or mediators separately. The participant observation was not limited to the inside of the courthouse—the researcher also followed the court officials to the labor bureau or factories if they were leaving for “on site” mediation work.
In total, fifty-seven mediation sessions were observed, twenty-two interviews (some in 2011, most in 2013) were conducted with judges or mediators, and countless other observations were made. The interviews were not limited to the judges or mediators in Court M, but extended to judges from other basic people's courts. Those interviews lasted on average three hours or longer, sometimes taking place in more than one session.
Case Quality Assessment System and Mediation Rate Collective Method
In order to understand the practice of mediation and how mediation rates have been constructed, we shall first explain how the SPC as a principal sought to incentivize judicial work with its subordinate courts acting as agents. This allows us to later understand how the mediation practices in Court M, as well as the reported mediation rate, relate to the external incentive system. It also allows us to examine the level and type of discretion the incentive system leaves to local courts.
Chinese lower courts exist within an intricate system of external bureaucratic controls that the SPC uses to regulate their work. After local experimentation during the 1990s, the SPC officially in 2002 began to plan for a systematic evaluation of case quality and efficiency (Yan Reference Yan2011). After multiple rounds of research, in January 2008 the SPC published Guiding Opinions of the Supreme People's Court on Conducting Case Quality Assessment Work (for Trial Implementation).Footnote 4 After three years of practice, the SPC published a revised version in 2011,Footnote 5 and officially launched the case evaluation system known as the Case Quality Assessment System.
The Case Quality Assessment System is a number-oriented system. In total the Case Quality Assessment System contains thirty-one indicators that evaluate and incentivize the quality of lower courts. There are eleven measures for the fairness of the trial,Footnote 6 ten measures for the efficiency of the trial procedure,Footnote 7 and ten measures for the outcome of the trial.Footnote 8
These indicators are designed to understand, improve, guide, and supervise the courts' work.Footnote 9 The system did so by higher level courts setting targets and by ranking the lower level courts according to the performance indicators (Xiang and Guo Reference Xiang and Guo2012; Zu Reference Zu2013) until the SPC stopped ranking the High People's Courts in December 2014 (Hu Reference Hu2014; Lou Reference Lou2015). For example, before December 2014, if the higher level court believed that the faster the courts processed the cases the better it was, the “number of cases a court solves” was set as one of the key indicators, so a higher level court could rank the lower courts and the lower level courts could rank their judges (Yan Reference Yan2014). As shown by recent studies, the control of the ranking system determines the promotion of judges, the disparity among judges' payment scales, and personal honors and other social benefits (Zhang Reference Zhang2012; Yang Reference Yang2014; Zhao Reference Zhao2014; L. Li Reference Li2015). Such incentives are similar to the cadre evaluation system through which all Chinese officials are evaluated, but tailored to fit the Chinese judiciary's work (Manion Reference Manion1985; Huai Reference Huai, Hamrin and Zhao1995; Huang Reference Huang1996; Edin Reference Edin2003; Whiting Reference Whiting, Naughton and Yang2004; Minzner Reference Minzner2009bb; Birney Reference Birney2012).
The SPC's reemphasis on mediation began in 2002. The SPC increasingly started to emphasize and incentivize mediation in courts, as it responded to an upsurge in petitions from citizens seeking justice directly with central level authorities in Beijing. Courts, and ultimately the SPC, were held responsible for the rapid rise in petitions (Zhang Reference Zhang2004; Liebman Reference Liebman2009; Fu and Cullen Reference Fu, Cullen, Margaret and Gallagher2011). To implement the top-down mediation policy, the Chinese Case Quality Assessment System made the mediation ratio a key index in court evaluation work in 2011, in an attempt to increase mediation in the courts. Within the courts this was translated into individual personnel policy, and mediation rates became an important evaluation standard for the courts and their presidents, and also for individuals of lower level rank and file judges (Cai Reference Cai2013). Mediation performance in these systems is only evaluated through one index: the mediation rate, namely the number of cases “that are concluded through mediation in a certain period of time” (Chen Reference Chen2013). The SPC, as a principal, thus clearly sought to incentivize the behavior of its agents, the lower courts.
Originally, before 2011, the SPC used a rather crude and ambiguous standard to measure mediation performance. This allowed the lower courts wide discretion in how to define mediation and also what aspects of judicial work should be seen as part of the overall work against which the mediation rate was to be calculated. Higher courts implementing the Case Quality Assessment System did have some information about the work done in lower courts. All courts use a computer system, literally called “the court's collective data management system (fayuan zonghe xinxi guanli xitong)” (Xiang and Xiang Reference Xiang and Xiang2012; Wang and Liang Reference Wang and Liang2013). Under this system all courts digitally record many aspects of their court work, and this gives higher courts direct access to crucial information for the Case Quality Assessment System, including the cases that are filed, the time of filing, and the way cases are resolved. This system very likely reduced the information asymmetry the higher courts as principal faced when implementing the mediation incentive system at the lower courts acting as their agents.
To deal with the discretion problem, in 2011 the revised version of the Case Quality Assessment System imposed a more detailed formula which distinguished the summary trial cases from the common trial cases (Lin Reference Lin2012). However, even after 2011 no explicit national standard has been provided for what can and cannot contribute to mediation cases (Wang Reference Wang2012). Thus lower courts retain considerable discretion on what to report, and a so-called principal–agent problem of bureaucratic control arises with the SPC not being able to specifically dictate what the courts should actually do.
The reliance on a singular number to represent mediation performance strongly simplifies what is a complex practice. The definition of “mediation” used to calculate the mediation rate always adopts a “result-oriented” approach instead of a “means-oriented” approach. This implies that the mediation rate is based on the conclusion of the cases instead of on the substantive quality of the dispute resolution process. One example is disputant settlement. Even if disputants negotiate a case by themselves (instead of having a third party mediate, i.e., a court involved) and the negotiation leads to an agreement, as long as the case is concluded through mediation-related court documents, the case is still regarded as “being mediated” and contributes to the court mediation rate (Cai Reference Cai2013). This “result-oriented” approach is essentially different from the academic depiction of court mediation. Consequently, not all cases included in the mediation rate have gone through processes that are substantively qualified to be called mediation. For the SPC as principal, if it seriously wanted to change the actual judicial process to become more mediation-based this situation would of course be detrimental. Lower courts could claim higher mediation rates without substantively doing much more mediation in the process, simply by redefining the meaning of mediation (Hutter Reference Hutter1997; Talesh Reference Talesh2009). In so doing they could engage in creative and symbolic compliance (McBarnet Reference McBarnet1988; Gray Reference Gray2006; Gray and Silbey Reference Gray and Silbey2014). This can easily lead to goal displacement, where the original goal of a substantive change in the judicial process simply results in formal changes in the type of final documents courts produce (Merton Reference Merton1940). For our understanding of mediation in Chinese courts this also has important implications. The Chinese court mediation rate ought not to be understood as a strict depiction of a change in the substantive judicial process, but should be seen as a depiction of a formalistic result—that is, the cases that are concluded with mediation-related court documents.
The Implementation of the SPC's Top-Down Mediation Policy in Court M
Even prior to the SPC's Case Quality Assessment System, the high court in M Province had already established, in 2004, a similar evaluation system to measure and rank lower courts' work. All intermediate courts in M Province sought to compete for achieving mediation targets in a ranking system set up by the provincial high court, and the intermediate Court M was no exception. After the SPC unified and standardized the national Case Quality Assessment System in 2008, the High Court M followed and adapted to the new SPC standard. Among the thirty-one indicators, mediation rate was recognized as one of the key indicators, among others, together with jie'anlü (the percentage of cases that are closed within the time limit) and fagailü (the percentage of cases whose original judgments were rescinded and which were remanded to the original court for retrial due to falsified judgments).
As a result of the local adaptation of these national-level incentive systems, judges in Court M in 2011 indicated a strong pressure to meet the mediation targets as one of the chief targets. On the 21st of each month, the evaluation results punctually landed on each judge's table, including the results of their mediation rate. The judges' names were listed in descending order, from the best to the worst. The results contained not only the statistics from the previous month, but also the average results of the season and of the last half-year. Fluctuations in the sequence of the names aroused some discussion in the offices. Judges and judicial clerks whispered stories to each other about the number of fluctuations—for example, who had always been a front-runner, whose performance had improved, and who never cared about the ranking. In this system, comparison was unavoidable: comparison with the judges' own previous personal record and the comparison with other judges. Clearly, court leaders, rank and file judges, and even the trial judicial clerks felt the pressure of competition to get a high mediation rate.
So what do these reported mediation rates actually mean? In most existing studies it is implicitly assumed that the reported mediation rate reflects mediation work done in the courthouse, and particularly during adjudication processes. In Court M, according to its data in 2011, in many instances the court's judges were involved in mediation work even before disputants had filed a case. Fifteen percent of Court M's labor cases are mediated in on-site trials (On-Site Mediation). In these cases court officials participate in dispute resolution processes organized by the labor bureau organized outside the court and before litigants have actually filed a case with the court. The second category, with 19 percent of all labor cases, is composed of so-called judicial confirmation cases (Confirmed Mediation). In the latter cases judges will confirm mediation agreements that parties themselves have negotiated, either themselves or aided by a people's mediation committee (a grassroots government mediation organization). Also in these cases litigants have not filed a case with the court. Eight percent of labor cases are mediated under a third category in which mediation occurs in the court during the case-filing stage (Case-Filing Mediation). In a fourth category, cases are mediated in the court during the adjudicative process (Judicial Mediation, 32 percent).
Thus what emerges is a very different picture of mediation than the face-value mediation rates suggest. Rather than largely affected adjudicative work, the SPC mediation incentives have spurred a growth of judicial work largely outside of the court. What we see is a creative compliance with the higher court demands that has led to a goal displacement. Let us look in more detail at each type of mediation below to understand better how this could happen and what it means for our understanding of mediation practices and how the mediation rate is produced. Also we shall look at how this has changed over time as the amount of discretion of what constituted mediation in these rates decreased and the Court shifted its mediation working routines.
On-Site Trials as Low-Cost Mediation
In 2011, following directions from the provincial high court and the SPC, Court M created a collaboration agreement for labor disputes with the local labor bureau. According to the agreement with the labor bureau, the original working scenario was that trial judges should participate in the dispute resolution process of the labor arbitration committee, hear cases through trials, and nip the disputes in the bud. This goal was clearly stated in the intermediate court's instructions: when disputants submitted a case to trial that met certain criteria, an on-site trial could accept it immediately. If the case could be heard immediately, the on-site trial should proceed without delay.
However, the actual operation of an on-site trial by Court M deviated from the original plan, as pro forma low level judicial involvement in on-site trials became a favored way for Court M to boost its mediation statistics without taking on more actual cases and doing more actual judicial work.
Court M accomplished this in several ways. First, the court restricted its actual involvement in the on-site trial mediation process. Instead of actual judges, most of the work in on-site trails was relegated to judicial clerks, as judges were simply too busy with real trials inside the courthouse.
Second, the court judicial clerks induced parties to file cases that the labor arbitrators had successfully resolved with the court. If the clerk succeeded, he/she took the already “resolved” cases to the court and had them go through the case-filing process, and they were concluded in a Civil Mediations Agreement. In other words, the clerk simply collected cases that had already been successfully mediated by the labor arbitration committee. These cases went through the case-filing process so that they could be counted as court mediation cases in the statistics.
Third, the judicial clerks played an insignificant role in the actual dispute resolution processes in the on-site trial. Clerks simply were there to ensure the quality of the settlement agreements, but not to actually help reach such settlements, even though formally the clerk was part of the mediation process and arbitrators would delegate to them legal authority in the procedure. In one case, for example, a worker claimed compensation for the illegal termination of a contract and asked for an overtime work payment of 10,000 yuan. During mediation, both parties focused on compensation for the illegal termination of the contract. The factory offered an overall compensation of 6,000 yuan for all claims. The judicial clerk had to make sure that other claims were also referred to in the mediation agreement besides the compensation for illegal termination, or they could lead to the claimant reopening the dispute.Footnote 10
Fourth, judicial clerks filtered cases that could boost mediation numbers without causing much trouble to the court. In most cases, the judicial clerk accepted every case that was mediated by an arbitration committee for a judicial mediation agreement. Cases were not accepted if (1) there was a flaw in the mediation procedure (in an overtime payment case, legal documents could not be provided to identify the agent of the factory, so the judicial clerk refused to take the case back for a judicial mediation agreement);Footnote 11 (2) there was an agreement that might cause trouble for the court. For example, in one case of an illegal termination of a labor contract, the worker was very irritated and complained a lot about the factory and the mediation process. The judicial clerk did not bring the agreement to go through the case-filing procedure; he told the arbitrator that they could still have it as a labor arbitral mediation agreement but not as a court document. As the judicial clerk explained, having disputes end in a judicial mediation agreement was risky because the court could become inextricably entangled in the dispute, especially if the parties showed signs of restlessness.Footnote 12
In sum, Court M here responded to the external mediation incentive system by trying to get higher mediation numbers. They did so in a way that brought about the lowest cost to the court. As such, Court M as an agent responded to the incentives of the principal, yet did so in a way that did not meet the substantive goals of increasing actual court involvement in mediation or enhancing the actual judicial mediation processes. Clearly here we see a form of creative and symbolic rather than substantive compliance with the spirit of the SPC policies. The court used the discretion that the incentive system allows to claim more mediation work that it was actually never much involved in, and that would also have occurred without the court's involvement. The discretion in the original incentive system with its vague operationalization of what is to be counted as mediation, and the emerging practice of on-site mediation created the opportunity for Court M to do so. The result was a goal displacement (Merton Reference Merton1940), where the original goal of the mediation target incentive system—to enhance mediation in judicial work—got lost and instead Court M found a way to attach its name to already ongoing practices. As one senior judge in Court M stated, “Sending court staff to the labor bureau is done to meet the statistical requirement. If there were no assessment statistics and targets, we would not do that. To be honest, we have too many cases on hand, how can we take the time to step into the shoes of the labor bureau?”Footnote 13
Boosting Mediation with Judicial Confirmation
Court M has also sought to boost its mediation numbers by directing disputants into the judicial confirmation procedure. Court M has done so under a legal rule in the People's Mediation Law that allows parties who have reached a mediation agreement themselves or through the help of people's mediators to go to court to have it judicially approved.Footnote 14 According to this rule, the initiative should be with the disputing parties to turn to the court. In practice, however, the judicial confirmation in Court M has worked out quite differently.
In Court M's jurisdiction, few disputants proactively come to the courthouse to initiate a judicial confirmation; usually, the courts take the initiative. Courts often receive information on the mediation of collective action disputes from the local government. Based on that information, Court M judges have gone on-site and conducted the judicial confirmation of the mediation agreements. They did so to make sure that the already settled cases could be judicially confirmed.
Judicial confirmation was especially attractive for Court M's mediation rate because nearly all judicial confirmation cases in Court M were collective disputes in 2011, and each collective dispute involved more than thirty parties. In China, courts can separate collective into individual cases, and tend to do so to boost their judicial performance statistics. For example, a collective case regarding one hundred plaintiffs is considered one hundred individual cases in the official statistics. This results from the fact that statistics regarding the cases handled by a court are one of the crucial indexes to measure the courts' work, so the more cases the courts handle the better (Chen and Xu Reference Chen and Xu2012). In 2011 in Court M, the largest case brought in more than 120 disputants.Footnote 15 As a judge stated, “It is cost-effective for the court to collect cases with the least effort.”Footnote 16
Interestingly, in contrast to on-trial mediation, judicial confirmation, which seems like a low-cost and easy way to boost mediation numbers, proved to be rather cumbersome for the court. Court M judges and clerks often ended up getting involved in the actual dispute resolution process, instead of merely examining and confirming the already prepared mediation contracts. They did so in part because the local government wanted the court to be more involved as a legal authority in the mediation process itself, especially as these were often collective disputes with workers protesting on the street. In these sensitive cases governmental officials hoped that the court's presence would increase the authority of the dispute resolution. In one case, a government official kept referring to the judicial clerk as an “expert in law” and encouraged the parties to consult with him about legal questions.Footnote 17
But even if the government wanted Court M to be more involved earlier in the mediation process, the court could have done so at the lowest possible short-term cost simply by going through the motions and confirming whatever agreement came out. Yet, they did not do so; Court M judges and clerks actually actively participated in the mediation and the agreement drafting at various stages. Court M got more involved earlier on in order to make sure that the agreement reached would stick and not come back to the court. Similar to other courts, Court M held that only a rigorous mediation agreement that finds a solution of the key issues at stake can have a sustainable effect to end the dispute (Research Group of the Zhejiang High Peoples Court 2010). Governmental mediators care for the more immediate concern of ending an ongoing protest and getting workers off the streets.Footnote 18 In Court M there also clearly were tensions between the court and governmental mediators. It was in the court's interest to participate in the earlier parts of the dispute resolution process and convince governmental mediators to follow a long-term perspective of reaching mediation deals with rigorous and fair agreements. The court did not always succeed. Sometimes, this even resulted in the court rejecting a judicial confirmation initiated by local governments.Footnote 19 The following example shows under what conditions such a rejection may happen.
In 2011, the labor contracts of three hundred workers were terminated because a factory was relocated. To appease the workers' grievances, the local government (jiedao) conducted mediation and invited the court to confirm the mediation agreement. In the draft agreement, the local government only listed the compensation amount without mentioning major issues such as overtime work payment and labor insurance. The court believed that by leaving those issues unaddressed, this agreement would not put an end to the dispute. So the court suggested inserting clauses addressing those issues into the contract. The government officials at first agreed, but later retracted their promise. The government's main concern was that, if the clauses were inserted, workers might claim more in the future. The government had already paid the compensation, and it could not (or was unwilling to) afford more. Officials also had concerns about the civil mediation agreement itself, and worried that asking workers to sign more documents for judicial confirmation would delay the dispute solving process, thus frustrating the goal of making the workers stop assembling as soon as possible.Footnote 20 With the mediation agreement unchanged, the court refused to provide a judicial confirmation, leaving the dispute to conclude with a non–legally confirmed mediation agreement.
The incentive system, as it had originally existed with broad discretion as to what constitutes mediation, allowed Court M to seek these collective confirmation cases. Again we see that a goal displacement occurred, with the Court moving resources outside of its court and going beyond its legal obligations in proactively seeking these cases and getting involved in the actual mediation process rather than simply confirming. With these cases, we learn that Court M responded to the SPC incentive system by maximizing its mediation rate, finding the greatest number of cases, and managing the process to reduce longer term risks and costs. In contrast with the on-site trials, Court M here did end up doing real mediation work, albeit not in the adjudicative process. It is important to note what Court M did not do, at least as far as we were able to learn from the fieldwork. Instead of boosting its numbers through on-site and confirmed mediation, Court M could have tried to outright falsify reporting and report higher in-court mediation statistics. There is no indication that this actually happened. The digital system for recording court work, which also recorded work on these on-site trials, provides higher courts with direct information on the day-to-day working operations of the lower courts. This has reduced information asymmetry, making it much harder to produce fake numbers because these cannot simply be reported at the end of a reporting period but must be recorded on a daily basis, thus requiring a much more deliberate and complex form of fraud involving a much larger number of court personnel. Instead, we see that the court strategically uses the discretion the rules allow to get the best numbers at the lowest cost.
Limiting Discretion Ends Out-of-Court Mediation Practices
By the middle of 2012, when our fieldwork was resumed, a sudden change had occurred in Court M. The court's mediation rate dropped from 74 percent to 60 percent. During 2012, there was virtually no more confirmed mediation, and by 2013 on-site mediation also had ended almost completely.Footnote 21 These changes in Court M's mediation practices occurred in direct response to changes the SPC made in its revised version of its Case Quality Assessment System in 2011. The 2011 version of the system, as we saw, provides a more detailed formula for what is to be included in the mediation rate. The new formula, which is not published, seems to restrict the definition of mediation to practices that occur in the courthouse. Therefore, on-site mediation and confirmed mediation would no longer count for a court's mediation performance.
In 2012, to echo the changes at the central level at the local level, Court M adopted a policy under which the mediation rate could only include what had gone through official case-filing, or where the disputants had at least come to the courthouse for dispute resolution.Footnote 22 All of this completely disincentivized both on-site and confirmed mediation practices. Not surprisingly, both out-of-court mediation practices that had been such a vital part of a mediation rate booster strategy decreased and eventually ended. From the middle of 2012 to the middle of 2013, there were no more confirmed mediation cases. And in the same period Court M conducted only thirty-nine on-site mediation cases. The on-site cases continued for a while because this procedure had been embedded in a working agreement with the labor bureau, and it took some time to terminate it. As for judicial confirmation, as it was not institutionalized notwithstanding its existence as a procedure in the law, Court M simply stopped joining the sessions organized by the government or local bureaus.
Clearly here the changed incentive structure that decreased the rule-based discretion in what constitutes mediation for inclusion in the mediation rate was at play. The new definition narrowed the discretion in the rules, and the court no longer include out-of-court mediation practices as part of court mediation. This directly impacted its mediation practices. Interviewees have clearly indicated that, now that the rules in the incentive structure were changed, their attitude toward out-of-court mediation was altered. For instance, one judicial clerk who was in charge of on-site trial stated in 2012: “How can I possibly like the on-site trial? I went to the on-site trial every day and took the case back. There was lots of paperwork. Of course I would participate in the mediation, but the goal of my job is very vague. The court has wasted too much human capital and resources on it.”Footnote 23
Even in 2011, judges clearly indicated that, were it not for the inclusion in mediation rates, they would not want to spend court resources on mediation outside of the court. As a senior judge stated,
The mediation rate requirement is around 65 percent, it cannot be achieved through the existing channels (the inside-courthouse mediation). What if disputants in a litigation process do not want to mediate? We cannot force them to sign the in-trial mediation agreement. In short, judicial confirmation and on-site trial happened in a special time with a special purpose.Footnote 24
What we thus see is that the SPC incentive system does shape mediation behavior in the court. When it still allowed broad discretion in what was mediation, Court M responded by focusing on high volume and relatively low-cost forms of mediation. By simply limiting the discretion in the mediation definition of the incentive system, the SPC was able to end out-of-court practices in Court M that neither helped reached the primary goal of enhancing mediation in judicial processes nor provided the most efficient use of scarce court resources to meet the court's core job challenges of managing its in-court workload. What is interesting here is that a simple change in the central level rules could have such a large effect at the local level. Here again we found no indication that the court or its judges sought to report fake numbers, and maybe also here the digital system for recording cases would have made this a complex and risky affair. What we thus see here is that the standard principal–agent problem of information asymmetry and challenges for principals in verifying agent performance is not clearly at play. Rather, it seems that at Court M the goal displacement that occurred was more a matter of the rule-based discretion the original policy allowed.
Mediating while Filing Cases
With the 2012 change in the mediation definition, Court M had to reach mediation rate numbers through processes that occur inside the court. The first place it tried to do so was at the court's doorstep, in the case-filing procedure. Case filing occurs in the court's case-filing division. This division, which is responsible for making decisions about which cases the court will accept, has functioned as a sort of minicourt. It has its own subdivisions: the case-filing committee, the committee for the preservation of evidence, the committee for serving notices, and the petition committee (Liebman Reference Liebman, Woo and Gallagher2011a). The case-filing division is very powerful, because it has large discretion of which cases to accept and thus guards access to litigation. With the development of the so-called “speedy trial” and “small claim trial,” the case-filing division is also responsible for actually hearing cases through trials (Ningbo Yinzhou Court 2010).
Mediation at the case-filing division of Court M serves a dual purpose. It can help resolve easy cases and thus decrease the workload for trial judges. And of course mediation work here will help raise the court's mediation rate. As such, at Court M, if the case-filing division gets cases with small claims and clear facts, it should try to organize mediation sessions and end the disputes. This has established Court M's case-filing division as a mediator in its own right.
Court M established a procedure where each dispute that qualified would be assigned to a case-filing mediator organizing a mediation session with both parties.Footnote 25 Successful mediation concluded in mediation agreements; without success the case would move to the trial judge.
Court M invested considerable resources in mediation work at the case-filing stage. Court M not only assigned nine judge assistants to do case-filing mediation work, it even hired two people's mediators with extensive local-level mediation experience to come and work in the courthouse. Although formally employed by the local bureau of justice, the two people's mediators in Court M are actually working for the court.
From September 2012 to May 2013, nearly twenty cases were successfully mediated by one of the people's mediators each month, approximately one-fifth to one-third being labor cases.Footnote 26 And since one judge assistant could mediate around ten cases per month,Footnote 27 the mediators of the case-filing division contributed significantly to the mediation rate rise.
Court M had a strong interest to make sure that all mediation in the case-filing division could count toward its mediation rate. The challenge was to get the work done by the externally hired people's mediators to count as court mediation. According to Chinese law, the mediation these people's mediators concluded was to result in a so-called “People's Mediation Agreement.” This agreement could itself not count toward the mediation statistics and also was not strongly enforceable by the parties. Court M could confirm such an agreement either through the judicial mediation confirmation procedures or through a so-called “Civil Mediation Agreement.”Footnote 28 As said, the former needs to go through the confirmation procedure, while the latter would require a full filing of the case with the court. In fact, the People's Supreme Court has never ruled that mediation cases concluded by in-courthouse people's mediators can be granted a Civil Mediation Agreement. This practice was mostly “legalized” by some high courts, including the High Court in M Province and some other high courts such as Heilongjiang high court in 2009 (Heilongjiang High Court 2009).
Originally, before 2012, Court M could have simply boosted its mediation rate by judicially confirming the mediation agreement reached by the people's mediators who had been contracted to work at the case-filing division. After 2012, however, as we saw, judicial confirmation mediation could no longer count toward mediation rates. Court M thus had to adapt its procedures, and it started to strongly favor confirming the people's mediators' cases by filing them within the court for a Civil Mediation Agreement. This of course involved more work for the court. As a result, the focus on mediation statistics began to come into conflict with the court's interests in managing its workload. It started to file cases that already were resolved and could end with a simple judicial confirmation outside of court, adding them as part of their workload.
Court M used its legal authority and power to guide these cases toward a full case filing for Civil Mediation Agreement and inclusion in its mediation statistics. Formally, in M Province, the disputants are to choose which confirmation procedure to use. Yet that hardly happens in practice. Court M's filing division mediators predefine how disputants select the path, and simply ask the disputants to sign prepared documents which often include a form for case filing. Hardly any explanation is provided.Footnote 29 The disputants are only informed that they would get their legal document later. Disputants, during interviews, indicated that they had no idea what sort of document they would be provided to confirm the mediation agreement, but they believed that whatever document they got carried authority and enforceability.Footnote 30
Here again we see that Court M strongly responded to the external incentives system but tried to use discretion to redefine it in a way that best enhanced its mediation performance and served its broader interests. The result was again a form of goal displacement, this time with the court undertaking unnecessary extra work by filing cases for a civil mediation agreement which could simply be confirmed, all in order to get better mediation numbers. In doing so, the court bent and reshaped legal rules, using its legal authority over litigants who simply did not know better, and benefiting from a lack of oversight on its choices about which mediation procedures to follow during case filing.
Trial Judges Mediating Collective Disputes
Following the 2011 revisions in the SPC Case Quality Assessment System, courts received more specific targets for their mediation work. According to what our informants told us in 2013, trial judges needed to reach a mediation rate of 40–45 percent of their own cases.Footnote 31
Some cases trial judges handle are naturally suited for mediation. For instance, if a labor case has received a dichotomous arbitral award, or if the facts and the law of the dispute are clear, judges find it relatively easy to help the disputants reach an agreement because disputants from both sides quite clearly know what they can expect.Footnote 32 In these cases, mediation is quite easy, and judges do not need to devote much effort. Most judges in Court M have stated that 30 percent would be a normal rate for cases where they do not need to render judgments for first instance civil cases.Footnote 33 Beyond 30 percent, however it is increasingly unlikely that cases are naturally suited for mediation. It is likely that cases have complex fact patterns, conflicting interpretations of law, and disputants who are simply not inclined to reconcile and find a middle ground and accept a settlement.
However, there are still cases beyond the 30 percent of naturally suitable mediation cases that judges themselves want to mediate. As mentioned in some existing studies, Chinese judges push or even coerce some litigants into settlement (Balme Reference Balme and Peerenboom2010; Fu and Cullen Reference Fu, Cullen, Margaret and Gallagher2011; Minzner Reference Minzner2011). This fieldwork found, however that in these cases coercing mediation on litigants does not arise from judges wanting to get higher mediation numbers. As one judge explained: “If you see a judge trying hard for mediation, he must be faced with a difficult case.”Footnote 34 Judges will thus try and force litigants into mediation if they see a certain risk of going through normal adjudication, and where adjudication can result in “wrongful cases” (as registered in the Case Quality Assessment System), difficult disputants who may later appeal or petition, or simply the risk of spending too much of their highly constrained time on cases (Minzner Reference Minzner2009aa; Y. Li Reference Li2015). Judges will only choose to force mediation through in these cases if the difficulty of such mediation, which is considerable, is less than the risks associated with going forward with adjudication. In Court M, coerced mediation by trial judges was not frequent and only contributed marginally to the overall mediation rate. Accordingly, for trial judges mediation was the preferred method in about 30 percent of easy cases and in a small number of high-risk cases. Yet the mediation rate target for trial judges is 45 percent. Judges have responded to this challenge, as much as possible, by trying to get collective disputes into mediation. As we saw, under Chinese law, such collective disputes can count as individual cases on the judge's and thus also the court's performance record. For example, in Court M there was a case involving 267 disputants. The boss had abandoned the factory and fled overseas in 2012 in City M. Subsequently, all 267 disputants filed claims before a labor arbitral tribunal for salaries, compensation, and social insurance. After the court mediated the salary claims, 212 of the disputants accepted the mediation agreement.Footnote 35 The judges in this case, even though they did the work for only one trial mediation, could add 212 trial mediation cases to their records. This strategy plays an important role in increasing both the trial judges' mediation rate and the court's mediation rate. The above case contributed significantly to the trial judges' mediation rate. If this single case was removed, the percentage of cases mediated by the trial judges would drop from roughly 48 percent to 37 percent, which made the judges fail to reach the mediation rate targets.
In the case of the trial judges, we again see a strong responsiveness to the external incentive structures. With the decreasing discretion in the mediation rate definition, Court M has started to put pressure on its trial judges to meet these targets. Again here they have used discretion in the system, this time in how to define what constitutes one mediation case, to boost their numbers. Again we see limited outright resistance to the incentive structure but rather attempts within the confines of the system to make it workable. And again we see creative and symbolic rather than substantive compliance with the policy demands.
Conclusions
This analysis of Court M's recent practices sheds new light on China's court mediation surge. The study revealed that the mediation rate does not simply represent more mediation in adjudicative work, as the original SPC policy intended. Rather, the mediation rate surge, at least in Court M, represents mediation work done outside the court, by lower court judges and judicial clerks and in collaboration with labor bureaus and people's mediators. Moreover, it shows that a considerable portion of the actual increases in trial mediation are attributable to case-accounting mechanisms—for example, counting each litigant as a singular case in large collective cases. This means that any normative discussion of what the surge of the mediation rate means (Su and He Reference Su and He2010; Fu and Cullen Reference Fu, Cullen, Margaret and Gallagher2011; Liebman Reference Liebman, Heilmann and Perry2011b; Minzner Reference Minzner2011), as so many scholars have done, should first of all specify actually which aspect of the actual mediation work the rate represents. Here, even in one court, we found great variation not only in the mediation practices that generate the mediation rate, but also in changes over the short time span of two years. We need more in-depth studies from other locations to capture the meaning of the mediation rate surge as it happened across the country over a longer period of time. As far as Court M is concerned, we can conclude that the net results of the SPC mediation incentives were creative compliance (McBarnet Reference McBarnet1988; Gray Reference Gray2006; Gray and Silbey Reference Gray and Silbey2014) and goal displacement on two levels (Merton Reference Merton1940; Wilson Reference Wilson1989). First, the mediation rate surge resulted in higher numbers but not in more and better mediation in actual judicial processes, as mediation originally was situated outside of the court. And second, as the court sought to raise its numbers through investing time and resources outside of its judiciary processes, it took away valuable support for its core mission of adjudication. Thus, the case study underlines that measurement performance tools can have serious adverse effects. They not only can lead to an adaption of policy goals limited to the dimension that is actually measured—in the Chinese case new practices solely focused on “number making”—but they can also make it even less clear for the public authority and scholars to discern what is actually happening at the local level.
This study highlights the functioning of hierarchal control in the Chinese court structure. Similar to work in other policy arenas in China studying vertical control (Huang Reference Huang1996; Edin Reference Edin2003; Van Rooij Reference Van Rooij2006), we see that SPC as principal can have a strong impact on the work carried out by its lower court agents. Court M responded strongly to the SPC incentive structures as they were replicated from Beijing through the province and municipality into the court. We did not find any indication that Court M tried to make use of information asymmetries to shirk its duties outright and produce inflated or fake numbers. We found that, most likely, lower court judges had trouble doing so because key elements of their work are recorded in a digital system that is directly accessible by higher courts. And as such we do not seem to have a classic form of a principal–agent problem, where information asymmetry obstructs oversight and control (Eisenhardt Reference Eisenhardt1989; Songer, Segal, and Cameron Reference Songer, Segal and Cameron1994; Waterman and Meier Reference Waterman and Meier1998; Shapiro Reference Shapiro2005).
Of course, this may be expected in China's authoritarian setting (Nathan Reference Nathan2003), where hierarchical control seems natural. Within the study of Chinese governance it is an interesting finding nevertheless. Many scholars have highlighted how fragmented vertical hierarchies are (Lieberthal Reference Lieberthal, Lieberthal and Lampton1992; Mertha Reference Mertha2009), and how often Beijing is not able to successfully implement its policies at the local level. Many studies have highlighted the problem of local protectionism, which has for instance undermined the local implementation of environmental regulation (Ma and Ortolano Reference Ma and Ortolano2000; Van Rooij and Lo Reference Van Rooij and Lo2010) or arable land protection rules (Sargeson Reference Sargeson2002). In these arenas local governments have at times simply gone against national law and policies simply because they could because the central government lacked oversight and control. By contrast, in this case this did not happen. We see here, as also evidenced by other studies (Huang Reference Huang1996; Edin Reference Edin2003), that the cadre evaluation system incentivizes local level actors' behavior.
Court M sought to comply with the incentive structures and create more mediation. In its compliance, however, the court reshaped the meaning of court mediation as far as it possibly could, given the text of the incentive system. It did so to ensure that it could gain from the incentive system, yet at the lowest possible cost. Scholars of government regulation point to similar phenomena when enforcement agents or regulated firms shape the meaning of legal rules when operationalizing the law into their working environment, highly similar to what happened here (Hutter Reference Hutter1997; Lange Reference Lange1999; Talesh Reference Talesh2009). The result is a form of “creative compliance” (McBarnet Reference McBarnet1988; McBarnet and Whelan Reference McBarnet and Whelan1991), where the court here achieves the formal targets, yet not the substantive goals of the incentive system.
