[author]Chen Gang

[content]

The Nostalgia in Judicature: Does Judges Favor ‘Lao Xiang’ in China

*Author: Chen Gang

Professor of the School of Economics, Southwest University of Political Science and Law

*Author: Sun Yanni

PhD Candidate of the School of Economics, Southwest University of Political Science and Law

Abstract: The fellow-townsman (lao xiang) is the continuation of the clan culture rooted in the traditional local society, is the social network that the individual is connected by the hometown after breaking away from the original acquaintancesociety and entering the colonial society. This paper, for the first time in theory and experience, reveals that the judge may have the behavior of favoring his fellow-townsman because of his identity in the judicial system. Using the micro data matching the judge and the lawsuit case, we found that the judge and the plaintiff were fellow-townsman, and had no significant effect on the probability of winning the lawsuit and the ratio of litigation costs borne by the plaintiff. But the judge andthe defendant are fellow-townsman, significantly reduced the probability of the plaintiff’s victory, and increased the plaintiff’s burden of litigation costs ratio. This result is robust in the case of using different methods and excluding the geographical bias of judges, which shows that there is significant judicial favoritism towards judges’ fellow－townsman in the judicial process，but the judge was only favoritism to the defendant fellows, not to the plaintiff fellow. Meanwhile, the favoritism ofthe judge towards the defendant fellows does not differ significantly depending on the judge’s age, education and grade, which means that the judge’s favoritism to the fellow-townsman may not diminish as the quality of the judge improves.

Key Words: judge; identity; judicial favoritism; fellow－townsman

1Introduction

Optimizing the business environment to further stimulate the vitality of various market entities and cultivate new growth momentum is an important strategic measure for deepening China's economic reform in the new era. ① Among them, equal protection of property rights and contractual rights of various market entities in accordance with the law is an important content and goal of optimizing the business environment. The growth experience of countries around the world shows that the absence of an effective property rights protection and contract enforcement system is the most important source of the economic stagnation of third world countries in history and their underdevelopment in the contemporary era (North, 1990). As we all know, both formal and informal institutions are helpful in protecting property rights and enforcing contracts, but with the growth of the market economy, the role of formal legal systems in protecting property rights and enforcing contracts is irreplaceable, especially the effective judicial system centered on the court is considered to be the foundation of a country's long-term economic growth (North, 1990). Therefore, it can be said that the business environment is essentially a legal business environment, and "the rule of law is the best business environment" (Xie Hongxing, 2019).

After the reform and opening up, China's economy has continued to grow rapidly for many years despite the imperfect legal system. Informal institutions such as relationships and reputation have played an important role in property rights protection and contract enforcement (Allen et al., 2005). With the deepening of market economic system reform, informal institutions such as relationships and reputation can no longer meet the needs of the expansion of market-oriented contractual transaction relationships, and various market entities have an increasing demand for high-quality legal systems. Data show that since 1979, the National People's Congress and its Standing Committee have reviewed and passed 263 formal laws. Other departments with legislative power have also promulgated more than 700 administrative regulations, tens of thousands of local regulations, and a larger number of departmental regulations and a considerable number of autonomous regulations. At the same time, the number of civil and commercial cases heard by people's courts at all levels in China increased from only more than 300,000 in 1978 to more than 10 million in 2016, an increase of more than 30 times in less than 40 years. However, it is no secret that in the context of continuous improvement of legislation, the problem of judicial favoritism that is prevalent in China's judicial system has been widely criticized and is an important obstacle to optimizing the business environment.

The existing literature on judicial preference has extensively discussed the issue of judicial preference caused by judicial localization and administrativeization (Liu Zuoxiang, 2003; Ma Huaide, 2003; Zhang Weiying and Ke Rongzhu, 2002; Long Xiaoning and Wang Jun, 2014; Lu et al., 2015; He & Su, 2015; Yu Yongcheng and Wei Jian, 2017). However, compared with the judicial preference caused by judicial localization and administrativeization, another form of judicial preference that is more hidden but may be equally widespread, that is, the behavior of judges favoring their fellow villagers not due to external pressure but due to identity recognition, has not received attention and attention from existing research literature.

Chinese society is rural at the grassroots level, and the differential pattern of social relations is the basic characteristic of rural society (Fei Xiaotong, 1985). In rural society, everyone's social relations are centered on "self", and spread outward in circles like ripples on water. Among them, the closest to the center is the family with which one has the closest relationship. The family spreads outward to form the clan, and the clan spreads outward to form the neighborhood. Therefore, the clan culture formed in rural society with blood and geographical relations as the bond is one of the most important characteristics of Chinese culture, and it still affects all aspects of Chinese society. Fellow villagers are the continuation of traditional Chinese clan culture. They are the social relationship networks linked by hometowns after individuals leave the original kinship and acquaintance society and enter the stranger society (Yang Yiyin and Zhang Shuguang, 2012). People's identity recognition of their fellow villagers is a positive emotional involvement and convergence tendency towards the language, culture, concepts, customs, etc. that people have in their hometowns, which will eventually be externalized into mutual help among fellow villagers (Jiang Yongzhi et al., 2012).

Judicial justice is the true and eternal life foundation of the judicial system (Xie Youping and Wan Yi, 2002), and judicial neutrality is the basic prerequisite for achieving judicial justice③. However, even in the ideal scenario of independent judges, judges, as socialized individuals, will inevitably be influenced by social culture and customs in the judicial process (Nugent, 1994). Therefore, when traditional clan culture infiltrates judicial activities, the serious consequence may be that judges have a preference for fellow villagers in the judicial process due to identity recognition. Using a set of manually collected and sorted micro data matching presiding judges with commercial litigation cases, this paper first theoretically and empirically reveals the possible behavior of judges in the judicial process due to identity recognition. The study found that although the judge and the plaintiff are from the same hometown, there is no significant impact on the probability of the plaintiff winning and the proportion of litigation fees borne by the plaintiff, but the judge and the defendant are from the same hometown not only significantly reduce the probability of the plaintiff winning, but also significantly increase the proportion of litigation fees borne by the plaintiff. This result is robust in the estimation using different methods and in the case of excluding the judge's regional preference, indicating that judges do have significant preference for fellow villagers in the judiciary. At the same time, this article finds that judges' preference for fellow villagers will not change significantly due to specific factors such as education, grade and age, which means that judges' preference for fellow villagers due to identity recognition may not weaken with the improvement of judges' quality.

This article may have policy and theoretical contributions in the following aspects. First, this article provides important clues and policy inspiration for China to optimize the business environment and promote the rule of law. As China's economy enters a new normal and the international economic situation and pattern are undergoing major changes, the Party Central Committee and the State Council have made major decisions and deployments to "accelerate the creation of a stable, fair, transparent and predictable business environment", and the rule of law is the foundation and guarantee for optimizing the business environment, and is the best business environment. However, the behavior of judges favoring their fellow townsmen in judicial practice violates the basic principle of judicial fairness and is an issue that needs to be addressed and resolved in the process of optimizing the business environment and promoting the rule of law. This study shows that the governance of judges' favoritism towards their fellow townsmen in judicial practice due to identity recognition may not simply rely on the formalization, specialization and professionalization of judges. A more effective possible measure is to improve the existing litigation recusation system.

Secondly, this article expands the research perspective on judicial favoritism. In theory, as a socialized individual, it is almost impossible for judges to be unaffected by their own subjective biases and external environment when making judgments (Nugent, 1994). Therefore, judicial favoritism is not only a phenomenon unique to the Chinese judicial system, but a common problem in countries around the world, including high-income countries with developed rule of law (Shleifer, 2012). Studies on corporate bankruptcy cases in France and Russia provide us with evidence that courts are influenced by external economic and political environments and prefer local companies (Esquerré, 2015; Lambert-Mogiliansky et al., 2007). Studies on China also found evidence that the administrative and localized nature of the judiciary caused courts to prefer local parties and politically connected companies (Zhang Weiying and Ke Rongzhu, 2002; Long Xiaoning and Wang Jun, 2014; Lu et al., 2015; He & Su, 2013; Yu Yongcheng and Wei Jian, 2017). Different from these literatures, this paper is the first to reveal in theory and experience the behavior of judges favoring fellow townspeople due to identity recognition.

Thirdly, this paper expands the research literature on the possible preference problems caused by identity. Identity economics, which has emerged in recent years, incorporates identity into the neoclassical utility function and successfully explains socioeconomic issues such as education reform, corporate performance, ethnic conflict, and family and marital relations (Akerlof & Kranton, 2000; Bertrand et al., 2015). In particular, the existing literature predicts and reveals the regional preference of incumbent officials for their hometowns (Do et al., 2017; Li Shujuan and Xu Xianxiang, 2016) and the racial preference for the same ethnic group (Dickens, 2018) that are common in countries around the world based on identity theory. This paper further expands identity to the study of judicial preference in China.

Finally, this paper enriches the research literature on the socioeconomic impact of clan culture. The clan culture rooted in rural society was suppressed for a long period of time after the founding of the People's Republic of China, but it has revived and revitalized after the reform and opening up, and has played a positive role in the development of rural enterprises, the transfer of rural labor, the supply of public goods, the alleviation of income inequality, the relaxation of financing constraints, and the promotion of employment and entrepreneurship (Peng, 2004; Tsai, 2007; Guo Yunnan and Yao Yang, 2013; Guo Yunnan et al., 2014; Xu & Yao, 2014). However, the "limited morality" of clan culture, while strengthening the short-radius trust within the clan network, also reduces the general trust of the clan network to the outside world, which not only hinders the development of cities (Greif & Tabellinni, 2010), but also is not conducive to the development of rural enterprises after the marketization is relatively mature (Ruan Rongping and Zheng Fengtian, 2013). The evidence provided in this article shows that when clan culture penetrates into judicial activities, it may cause judicial favoritism that is contrary to judicial justice, which helps to more comprehensively observe and evaluate the possible socioeconomic impact of clan culture.

2Literature Review and Theoretical Hypothesis

The judicial system centered on the court is the most important mechanism for interpreting and applying national laws, and fair justice is the eternal life foundation of the judicial system (Xie Youping and Wan Yi, 2002). A legal proverb says: "Law comes to the world with the help of judges." Therefore, judicial neutrality is an important prerequisite for achieving judicial justice, that is, judges must apply the law impartially in the judicial process and can only make judgments based on the facts of the case and the law, without being influenced by their own subjective biases and external environment. However, in judicial practice, it is almost impossible for judges, as socialized individuals, to be in a purely neutral position in the judicial process and make judgments without any influence (Nugent, 1994). Some influences may be internal or personal, such as the judge's religious beliefs, values, personality, and upbringing; some influences may come from the outside, such as social culture and norms, legal and political environment, etc. Therefore, the phenomenon that judges favor one party in the judicial process due to subjective bias or the influence of the external environment is by no means unique to a certain country or region, but is a common problem in countries around the world, including high-income countries with developed rule of law (Shleifer, 2012). A study on Italian labor litigation cases showed that court decisions were significantly affected by labor market conditions. In areas with higher unemployment rates, court decisions favored workers more. If the unemployment rate rose by 1 percentage point, the company's winning rate would drop by about 2.5 percentage points (Ichino et al., 2003). Another study on French corporate bankruptcy cases found that when the unemployment rate was high, the court was more inclined to rule on corporate restructuring rather than bankruptcy liquidation, sale and closure (Esquerré, 2015). Lambert-Mogiliansky et al. (2007) studied Russian corporate bankruptcy case data and found that local government intervention led to a clear preference for local companies in local courts, which protected local companies from bankruptcy liquidation and helped local companies evade tax obligations and debt responsibilities to the federal government.

In China, local courts at all levels are subject to local administrative power in terms of institutional setup, funding sources, and the appointment, dismissal, and promotion of judges. At the same time, the courts' own management system also has obvious administrative characteristics (Liu Zuoxiang, 2003; Ma Huaide, 2003), which often causes local governments to indirectly or directly interfere with the trial of cases in local courts through administrative means in order to protect the interests of local enterprises or enterprises with political connections to the local government. Therefore, the localization and administrativeization of the judiciary have weakened the independent and fair trial status of local courts and judges at all levels in China, resulting in judges being inevitably interfered with by local governments and higher courts in the judicial process, and there is a general preference for local parties and parties with political connections. A study of 620 economic case judgments in a grassroots court in Beijing showed that the winning rate of local enterprises was significantly higher than that of non-local enterprises. Among them, the winning rate of local plaintiffs against non-local defendants was the highest (38.4%), and the winning rate of non-local plaintiffs against local defendants was the lowest (25.9%) (Zhang Weiying and Ke Rongzhu, 2002). Another study using 2,724 judgments from Shanghai courts found that government agencies and enterprises with government connections had significantly higher winning rates (He & Su, 2013). Using more rigorous quantitative analysis methods, Long Xiaoning and Wang Jun (2014) found that enterprises located in the same place as the court had a significantly higher winning rate in first-instance litigation based on a study of intellectual property cases collected in the Gazette of the Supreme People's Court, but this was no longer significant in second-instance litigation. Lu et al. (2015) studied 3,323 litigation cases involving listed companies and found that the winning rate of politically connected enterprises was significantly higher than that of non-politically connected enterprises by 8.6%, especially in courts with simpler cases, worse legal environment, and where politically connected enterprises were located. Yu Yongcheng and Wei Jian (2017) also found that local plaintiffs had a significantly higher winning rate in first-instance civil cases in grassroots courts of four municipalities directly under the Central Government, and that the winning rate of parties within the system was significantly higher than that of parties outside the system.

Compared with judicial favoritism caused by judicial localization and administrativeization, the problem of judges’ possible favoritism for their fellow townsmen due to identity recognition has not been observed in existing research literature. However, cross-national data studies including samples from developed countries show that in the economic field, the phenomenon of incumbent officials favoring their hometowns or ethnic groups due to identity recognition is widespread around the world (Do et al., 2017; Dickens, 2018). Studies in China have also found evidence that the identity recognition of local officials has significantly promoted the economic growth of their hometowns (Li Shujuan and Xu Xianxiang, 2016). Obviously, these research documents provide important references for observing whether Chinese judges have judicial favoritism due to identity recognition in the judicial process.

In China, fellow townsmen are the continuation of traditional clan culture. They are social networks linked by hometowns after people leave their original kinship and acquaintance society and enter stranger society. Social identity theory points out that groups with the same culture will form an "in-group" circle, while groups with different cultures will be classified as "out-group" circles. The behavior of members of the "in-group" circle has obvious "in-group preference" and "out-group discrimination" characteristics, that is, members of the in-group circle tend to allocate favorable resources to members of the in-group circle and give more positive evaluations to members of the in-group circle, while allocating fewer resources to members of the out-group circle and giving more negative evaluations to members of the out-group circle (Tajfel et al., 1982). In China, fellow villagers share a common regional culture such as language, customs, concepts and values. They will have a strong sense of identity with each other because of cultural similarities, and thus form a "fellow villager" circle with strong cohesion and important resource and information exchange functions, which is the so-called "in-group" circle in social identity theory. For example, folk organizations such as hometown associations or fellow villagers associations that are prevalent in the economic and even political fields④ are a direct reflection of the strong concept of fellow villagers in local culture.

It can be said that in a society of strangers, fellow villagers are the closest link to the center "self" in the social relationship network of people in a differential pattern (Yang Yiyin and Zhang Shuguang, 2012), and the strong identity recognition between fellow villagers due to cultural similarities will eventually be externalized into their mutual help (Jiang Yongzhi et al., 2012). In recent corporate governance research literature, Dai Yiyi et al. (2016) used data from A-share listed companies from 2008 to 2014 and found that if the chairman and general manager are fellow villagers with the same dialect identity, the company's agency costs can be significantly reduced. Lu Yao and Hu Jiangyan (2014) used data from listed companies in Shanghai and Shenzhen from 2000 to 2009 and found that the hometown relationship between the CEO and the board of directors has a significant positive impact on the company's risk level. Another study using data from listed companies in Shanghai and Shenzhen Stock Exchanges in China from 2000 to 2013 found that when the CEO and the directors of a company are from the same hometown, the company will have a higher tendency to violate regulations, and the probability of the company being audited after violating regulations is correspondingly lower (Lu Yao and Hu Jiangyan, 2016), which means that the hometown relationship between the CEO and the directors of the company may increase the possibility of collusion between them to violate the law.

Therefore, it is not difficult to understand that when the clan culture rooted in the traditional rural society permeates into judicial activities, the behavior of judges preferring fellow villagers out of identity recognition may not be uncommon. Moreover, the behavior of judges preferring fellow villagers within the scope of their discretion may not be considered illegal. Therefore, this paper proposes the following theoretical hypothesis:

Theoretical hypothesis: Under certain other conditions, judges have significant preference for fellow villagers in judicial activities.

3Data and Equations

3.1Data

In order to quantitatively identify whether judges have a preference for their fellow townspeople in the judicial system, we tracked and collected all the judgment documents of first-instance (civil) commercial cases presided over by judges in office in March 2019 from 2014 to 2018 in four intermediate people's courts (including two intermediate people's courts in the east, and one intermediate people's court in the central and western regions) and can be found on the China Judgment Documents Network. There are two points that need to be explained and emphasized for this unique manually collected data: First, the main reason why we chose judges of the intermediate people's courts as the observation sample is that according to the provisions of China's Civil Procedure Law, the grassroots people's courts have jurisdiction over first-instance civil cases, but "cases with major impact in the jurisdiction" are under the jurisdiction of the intermediate people's courts. Therefore, civil cases heard by the intermediate people's courts are more representative because they have a "major impact" in the jurisdiction. Second, for the research of this article, if we want to identify whether judges have a preference for their fellow townspeople in the judicial system, the inconsistency of the location of the plaintiff and the defendant in the litigation case is very critical information. Because if the plaintiff and the defendant are from the same place, the case does not have the necessary information basis to identify whether the judge has a preference for his fellow townspeople. Therefore, we screened all the first-instance commercial litigation cases heard by all judges one by one, and on this basis deleted the cases where the location information of the defendant and the plaintiff could not be identified. After sorting, we finally obtained 1,628 samples of judgment documents of first-instance commercial cases presided over by 128 judges.

According to the causes of litigation, these 1,628 judgment documents of commercial litigation cases can be roughly classified into four categories: contract disputes, trademark disputes, loan disputes and other disputes (see Figure 1). Among them, there were 536 contract disputes, including contract disputes such as sales contracts, construction contracts, and loan contracts; there were 960 trademark disputes, including commodity rights and unfair competition disputes, copyright disputes and work distribution rights disputes; there were 60 loan disputes, including disputes of private lending, financial lending and corporate lending; ⑥ There were 72 other disputes, including management disputes, execution objections, equity transfers and other types of disputes.

3.2Equations and Variables

In order to identify whether judges have favoritism towards their fellow townsmen in judicial practice, this paper sets the following benchmark regression equation:

Among them, the subscripts i and j represent the i-th case and judge j, respectively. The explained variable winnerp on the left side of the equation is a variable that reflects and measures the plaintiff's victory; the explanatory variable tomap on the right side of the equation is a dummy variable that measures whether the judge and the plaintiff are from the same hometown, the explanatory variable tomad is a dummy variable that measures whether the judge and the defendant are from the same hometown, and the explanatory variables localp and locald are dummy variables that measure whether the plaintiff and the defendant are local parties (i.e. parties in the same place as the court); W is a vector composed of individual characteristic variables of judges, Z is a vector composed of other case characteristic variables that may affect the defendant's victory, Π and Ψ are the regression coefficient matrices of the judge characteristic vector and the case characteristic vector, respectively; ν and μ are the court fixed effect and the year fixed effect, respectively, and ε is a random disturbance term.

This paper uses two indicators to measure and reflect the plaintiff's victory. The first indicator is the dummy variable win_r that measures the plaintiff's victory. If all the claims raised by the plaintiff are supported by the judge in the judgment document, then the plaintiff's victory dummy variable is assigned a value of 1; otherwise, it is assigned a value of 0. Given that in many litigation cases, the plaintiff's claims have not been fully supported by the court's judgment, but have not been completely rejected by the court. The court only supports part of the plaintiff's claims. In this case, it may be unreasonable to define the plaintiff as the winner or loser absolutely. Therefore, the second indicator used in this article to measure the plaintiff's victory is the proportion of litigation costs borne by the plaintiff (referred to as "litigation cost ratio"), which is based on the practice of He & Su (2013). According to the provisions of the "Regulations on the Payment of Litigation Costs of the People's Courts", "litigation costs shall be borne by the losing party", and "partial victory and partial defeat" shall be determined by the court according to the specific circumstances of the case. In practice, if the plaintiff's claims are more supported in the court's judgment, then the court will tend to let the plaintiff bear less litigation costs and the defendant bear more litigation costs when dividing the litigation costs. Therefore, the proportion of the plaintiff's claims supported by the court's judgment is negatively correlated with the proportion of litigation costs borne by the plaintiff, so the plaintiff's litigation cost ratio is an indicator that reversely measures the plaintiff's victory.

The explanatory variable tomap, which measures whether the judge and the plaintiff are from the same hometown, is determined as follows: if the plaintiff's place of residence and the judge's place of origin are in the same prefecture-level city, the plaintiff and the judge are considered to be from the same hometown, and the variable tomap is assigned a value of 1; otherwise, it is assigned a value of 0. The explanatory variable tomad, which measures whether the judge and the defendant are from the same hometown, is assigned the same value. It should also be noted that the reason why the hometown relationship between the two is determined based on whether the judge's place of origin and the plaintiff's (defendant's) place of residence are in the same prefecture-level city rather than the lower-level district or county level is mainly because the judges of the above-mentioned four intermediate people's courts that we tracked only provided us with information on their hometowns at the prefecture-level city level. However, the boundaries of hometown relationships in Chinese society are inherently very flexible (Yang Yiyin and Zhang Shuguang, 2012). For example, if a person is abroad, people from the same country are considered to be from the same hometown; if a person is in another province, people from the same province are considered to be from the same hometown; if a person is in another city, people from the same city are considered to be from the same hometown. Therefore, the definition of the hometown relationship between the judge and the plaintiff (defendant) has not been specific to the grassroots district and county level, and may not have a substantial impact on the research results of this article.

The control variables of the equation also include judges' individual characteristics and case characteristics variables that may affect the plaintiff's success. The judges' characteristics variables include four variables: gender (male), age (age), education (educ) and grade (grad). Among them, the judges' age is the age of the judges when they hear a specific case; the gender, education and grade of the judges are all dummy variables. When the judges are male, graduate students and senior judges, these three dummy variables are assigned 1 respectively; otherwise, they are assigned 0. ⑦ The case characteristics variables include two dummy variables to measure whether the defendant has hired a lawyer () and whether the defendant has attended the trial in court (). Among them, if the defendant has hired a lawyer and attended the trial in court, these two dummy variables are assigned 1 respectively, otherwise they are assigned 0. The case characteristics variables also include a variable (comp) that approximates the complexity of the case by the number of words in the judgment document, because the judgment issued by the court includes the cause of action, the facts and reasons of the dispute, the facts and reasons recognized by the judgment, etc. Therefore, the number of words in the judgment documents of more complicated cases is often more. When specifically estimated, the natural logarithm of the number of words in the judgment document is included in the equation.

Table 1 reports the descriptive statistics of each variable. Among the 1,628 judgment documents, a total of 600 judgment documents supported all the claims raised by the plaintiff, accounting for 36.86% of the total judgment document samples. At the same time, the average proportion of litigation costs borne by the plaintiff was 38.42%. In addition, among all the litigation case samples, the proportion of litigation cases in which the judge and the plaintiff were from the same hometown (the plaintiff's hometown) was 18.92%, and the proportion of litigation cases in which the judge and the defendant were from the same hometown (the defendant's hometown) was 51.11%, which was much higher than the proportion of litigation cases in which the judge and the plaintiff were from the same hometown. The main reason is that civil litigation cases in China are generally under the jurisdiction of the court where the defendant is located, and most judges in local courts are locals⑧. This also reminds us that if we want to more accurately identify the judge's preference effect for the defendant's hometown, we may need to control the judge's possible regional preference for local parties in the equation.

3.3Comparative Description

This article intuitively describes and compares the plaintiff's success in different situations. Table 2 reports the plaintiff's success in the cases of local plaintiffs (defendants) and non-local plaintiffs (defendants). Column (1) shows that there are 276 cases in which both the plaintiff and the defendant are local parties, of which 116 cases are won by the plaintiff, with a plaintiff's success rate of 42.03% and an average proportion of litigation fees borne by the plaintiff of 29.12%. A logical assumption is that when both the plaintiff and the defendant are local parties, there will be no regional preference in the judge's judgment. Therefore, by comparing the differences between the plaintiff's success in column (1) and other situations, it can be intuitively revealed whether the judge has obvious regional preference (local protectionism) in the judiciary.

Columns (2) and (3) report the plaintiff's success in the cases where the plaintiff is a local party and a non-local party, respectively. Among them, in the case where the plaintiff is a local party (column (2)), the plaintiff's winning rate is 53.06%, which is 11.03 percentage points higher than that in column (1). At the same time, the proportion of litigation costs borne by the plaintiff is 25.73%, which is 3.39 percentage points lower than that in column (1). In the case where the plaintiff is a non-local party (column (3)), the plaintiff's winning rate is 27.69%, which is 14.34 percentage points lower than that in column (1). At the same time, the proportion of litigation costs borne by the plaintiff is 45.61%, which is 16.49 percentage points higher than that in column (1). The above changes show that local courts in China may indeed have the problem of local protectionism that favors local parties.

Compared with the situation in column (1), the changes in the plaintiff's winning situation in the cases where the defendant is a local party and a non-local party (columns (4) and (5)) also intuitively reveal the possible judicial local protectionism problem of the courts. In the case where the defendant is a local party (column (4)), the plaintiff's winning rate is 30.06%, which is 11.97 percentage points lower than that in column (1), and the proportion of litigation costs borne by the plaintiff is 42.54%, which is 13.42 percentage points higher than that in column (1). In the case where the defendant is a non-local party (column (5)), the plaintiff's winning rate and the proportion of litigation costs borne by the plaintiff are not surprisingly increased and decreased respectively (compared with column (1)), among which the former even increased by a staggering 22.17 percentage points (0.6420-0.4203) compared with column (1).

Table 3 compares the plaintiff's success in different situations where the plaintiff (defendant) and the judge are from the same hometown. Column (1) reports the situation of 644 cases where the judge is not from the same hometown as the plaintiff or defendant. Among them, there are 356 cases where the plaintiff wins, and the plaintiff's success rate is 55.28%, and the proportion of litigation fees borne by the plaintiff is 27.93%. Obviously, in the cases where the judge is not from the same hometown as the plaintiff or defendant, the plaintiff's success reflects the judge's judgment result that is not affected by the identity of the same hometown. Therefore, by comparing the plaintiff's success in the cases where the judge and the defendant (plaintiff) are from the same hometown, it can be intuitively revealed whether the judge still has obvious preference for fellow villagers in the judicial system.

In column (2) where the judge is from the same hometown as the plaintiff and the defendant, the plaintiff's success becomes less optimistic than in column (1). Among them, the plaintiff's success rate is 30.77%, which is 24.51 percentage points lower than that in column (1). At the same time, the proportion of litigation fees borne by the plaintiff is 31.64%, which is 3.71 percentage points higher than that in column (1), which means that judges may prefer the defendant's fellow townsman in judicial practice, but the degree of preference for the plaintiff's fellow townsman is relatively lower, or even non-existent.

The situation of litigation cases where the judge and the plaintiff are from the same hometown as each other in column (3) also shows that there may be no preference for the plaintiff's fellow townsman in judicial practice, because in litigation cases where the judge and the plaintiff are from the same hometown as each other, the plaintiff's winning rate has decreased instead of increased compared with that in column (1) (from 55.28% to 44.16%), while the proportion of litigation fees borne by the plaintiff has increased instead of decreased (from 27.93% to 30.6%). At the same time, the phenomenon of judges preferring the defendant's fellow townsman is once again supported by the statistical results of litigation cases where the judge and the defendant are from the same hometown as each other in column (4). Among them, the plaintiff's winning rate is only 18.75%, which is 36.53 percentage points lower than that in column (1). At the same time, the proportion of litigation costs borne by the plaintiff increased to 48.18%, an increase of 20.25 percentage points (compared with column (1)).

The comparison results in Table 3 show that judges may indeed have obvious favoritism towards fellow villagers in judicial practice, but this is mainly manifested in the judges' preference for the defendant's fellow villagers. The judges' preference for the plaintiff's fellow villagers is not statistically significant. The judges' lack of preference for the plaintiff's fellow villagers may be related to the tradition of emphasizing "ritual order" rather than "rule of law" in rural society (Fei Xiaotong, 1985). We will give a more detailed explanation of this later.

4Judges’ Preference for Fellow Villagers: Empirical Evidence

4.1The plaintiff wins the case

This paper first uses the binomial logit model to estimate the impact of the hometown relationship between the judge and the plaintiff (defendant) on the plaintiff's chance of winning. All equations control for the court fixed effect and the case verdict year fixed effect. Table 4 reports the marginal effect of each explanatory variable on the plaintiff's chance of winning. In the equation of column (1), the explanatory variables only include the dummy variable measuring whether the judge is from the same hometown as the plaintiff and defendant. The marginal effect of the plaintiff's hometown () is positive, and the marginal effect of the defendant's hometown () is negative, and the latter passes the 1% statistical significance test, indicating that the judge's hometown relationship with the plaintiff does not significantly increase the plaintiff's chance of winning, but the judge's hometown relationship with the defendant significantly reduces the plaintiff's chance of winning, indicating that judges have a significant preference for the defendant's hometown in judicial practice.

In column (2), the dummy variable measuring whether the plaintiff and defendant are local parties is also included in the equation. Because civil litigation is generally under the jurisdiction of the court where the defendant is located, and the proportion of local judges in local courts is relatively high, the proportion of litigation cases in which the judge and the defendant are from the same hometown is also correspondingly higher. Therefore, the effect of judges’ hometown preference may be overestimated in column (1) because the judges’ regional preference is not controlled. The regression results show that the marginal effect signs of local plaintiffs () and local defendants () are consistent with the expectation of judicial local protectionism, which are positive and negative respectively, but they both fail to pass the 10% statistical significance test. Importantly, after controlling for judges’ regional preference, the marginal effect sign of plaintiffs’ hometown is still insignificantly positive, and the marginal effect of defendants’ hometown is still negative at the 1% statistical significance level, indicating that the judges’ hometowns significantly reduce the probability of the plaintiffs winning.

Column (3) incorporates the judges’ personal characteristic variables based on the equation in column (2). Among them, the marginal effects of judges’ gender (male), age (age) and grad (grad) all pass the 10% statistical significance test, which means that judges may indeed be affected by individual characteristics such as gender and grad when making decisions (Nugent, 1994). At the same time, after controlling for the judge's individual characteristic variables, the marginal effects of local plaintiffs and local defendants are not significant. At the same time, the marginal effects of the plaintiff's fellow villager and the defendant's fellow villager are positive and negative, respectively. The former fails to pass the statistical significance test, while the latter still passes the 1% statistical significance test.

Column (4) incorporates the case characteristic variables based on the equation in column (2). Among them, the marginal effect signs of the two variables, defendant hired a lawyer () and defendant attended the trial in court (), are negative as expected, but fail to pass the significance test. At the same time, the marginal effect of the complexity of the case (comp) is negative at the 1% statistical significance level, indicating that in more complex litigation cases, the probability of the plaintiff winning is lower, which is consistent with theoretical expectations. Because generally speaking, excluding the differences in the litigation capabilities of the plaintiff and the defendant, the more complex the case, the less likely the plaintiff is to win (Zhang Weiying and Ke Rongzhu, 2002). It is worth noting that after controlling for the case characteristic variables, the marginal effect size and significance level of the plaintiff’s fellow villagers and the defendant’s fellow villagers have not changed significantly compared with before. Among them, the marginal effect of the plaintiff’s fellow villagers is still insignificantly positive, and the marginal effect of the defendant’s fellow villagers is still negative at the statistical significance level of 1%.

The equation in column (5) includes all explanatory variables. The marginal effects of the plaintiff's fellow villager and the defendant's fellow villager are still positive and negative, respectively. The former is not statistically significant, while the latter passes the 1% statistical significance test, indicating that the judge's fellow villager status does not significantly affect the plaintiff's chance of winning the case, but the judge's fellow villager status significantly reduces the plaintiff's chance of winning the case. This means that judges do have a significant preference for fellow villagers in the judicial process, but this is mainly manifested in the judge's significant preference for the defendant's fellow villager status, and the preference for the plaintiff's fellow villager status is not significant, which is consistent with the findings of the descriptive statistics in the previous article. The fact that judges do not have a significant preference for the plaintiff's fellow villager status may be related to the tradition of attaching importance to "ritual order" rather than "rule of law order" in rural society. Because "ritual" is a social norm recognized by society, it is maintained by tradition, unlike law, which is implemented by external power. Therefore, in rural society, litigation has become a shameful thing, indicating insufficient education (Fei Xiaotong, 1985). At the same time, fellow villagers are the continuation of clan culture rooted in traditional rural society, and because cultural factors such as litigation concepts may be passed down from generation to generation in social groups through interpersonal intergenerational transmission, they have long-term historical continuity (Guiso et al., 2006). Therefore, although China is now in the process of transforming from a traditional rural society to a modern industrial society, the litigation concept of "no litigation" in rural society still remains firmly in the general public (Fei Xiaotong, 1985) and will have a lasting impact on people's behavior and decision-making. Therefore, it is not difficult to understand why judges do not have obvious preferences for plaintiffs' fellow villagers in the judicial process, because safeguarding rights and resolving disputes through legal proceedings does not conform to the "ritual governance" tradition in rural society, so that such behavior is difficult to get the preference of judges' fellow villagers.

In addition, it is noteworthy that the scale of the judges' fellow villagers' preference effect may be much larger than the scale of the judges' regional preference. Because after controlling for all explanatory variables (column (5)), on average, the judge and the defendant being from the same hometown reduces the probability of the plaintiff winning by about 27.36%. At the same time, even if the impact of local plaintiffs and local defendants on the plaintiff's winning probability is significant, local plaintiffs only increase the plaintiff's winning probability by about 9.8%, and local defendants only reduce the plaintiff's winning probability by about 3.31%, which is far smaller than the marginal effect of the judge and the defendant being from the same hometown on the plaintiff's winning probability.

4.2Plaintiff's litigation costs

Next, this paper uses the plaintiff's litigation cost ratio, that is, the proportion of litigation costs borne by the plaintiff, to measure the plaintiff's success, and estimates whether the plaintiff's and defendant's hometown relationship with the judge significantly affects the plaintiff's litigation cost ratio. Because the value range of the plaintiff's litigation cost ratio is between 0 and 1, we use the truncated tobit model to estimate the equation, and the regression results are reported in Table 5.

The equation in column (1) only includes two explanatory variables: the judge and the plaintiff are from the same hometown () and the judge and the defendant are from the same hometown (). Among them, the signs of the estimated regression coefficients of the variables and the variables are negative and positive, respectively, and both pass the 1% statistical significance test. Column (2) is based on the equation in column (1) and includes the two variables of local plaintiff and local defendant. The signs of their regression coefficient estimates are consistent with the expectation of judicial local protectionism, which are negative and positive, respectively, and both pass the 1% significance test, indicating that local plaintiffs significantly reduce the proportion of litigation costs borne by plaintiffs, while local defendants significantly increase the proportion of litigation costs borne by plaintiffs, which means that local courts in China may indeed have a preference for local parties. It is worth noting that after the inclusion of local plaintiffs and local defendants, although the signs of the estimated values of the regression coefficients of the variables are still negative and positive respectively, the former fails to pass the 10% significance test, while the latter still passes the 1% significance test. Columns (3) to (5) successively include the judge's personal characteristics and case characteristics variables on the basis of the equation in column (2). Among them, the estimated values and significance levels of the regression coefficients of the variables have not changed significantly compared with those in column (2). At the same time, the estimated values of the regression coefficients of the variables are always positive and can pass the 1% significance test.

The regression results in Table 5 show that the judge and the plaintiff being from the same hometown does not have a significant impact on the proportion of litigation fees borne by the plaintiff, but the judge and the defendant being from the same hometown significantly increases the proportion of litigation fees borne by the plaintiff. Under the condition of controlling all explanatory variables (column (5)), the judge and the defendant being from the same hometown increases the proportion of litigation fees borne by the plaintiff by about 16.62%. The above results are consistent with the findings in Tables 3 and 4, which also reveal that judges have a significant preference for their fellow villagers in judicial practice. Moreover, the judges' preference for their fellow villagers is only manifested in preference for the defendant's fellow villagers, but not for the plaintiff's fellow villagers.

4.3Sample cleaning

The regression in the previous article found that judges have significant preference for fellow townspeople in the judicial system, but it is mainly manifested in the judges' preference for the defendant's fellow townspeople, rather than the plaintiff's fellow townspeople. Since civil litigation in China is generally under the jurisdiction of the defendant's local court, the preference effect of judges for the defendant's fellow townspeople found in the previous regression may contain some information about the judges' preference for local defendants. Although we have included local plaintiff and local defendant variables in the equation, this may not be able to fully control the judges' preference for local plaintiffs. Therefore, this article further extracts the local defendant sample from the sample for regression. If the judges' preference for the defendant's fellow townspeople found in the previous regression contains some information about the judges' preference for local defendants, then when we use the local defendant sample for estimation, the effect of the judges' preference for the defendant's fellow townspeople will be greater than the effect of the judges' preference for the defendant's fellow townspeople estimated in Tables 4 and 5.

Table 6 reports the estimation using the local defendant sample. The results show that the judge and the defendant are from the same hometown, which significantly reduces the probability of the plaintiff's success (columns (1) and (2)), and significantly increases the proportion of litigation costs borne by the plaintiff (columns (3) and (4)), which is consistent with the findings in the previous article. More importantly, the estimation using the local defendant sample finds that if the judge and the defendant are from the same hometown, the probability of the plaintiff's success will be reduced by about 27.22%, and the proportion of litigation costs borne by the plaintiff will increase by about 14.91%. The impact effect is slightly lower than that in Tables 5 and 6, which means that the judge's preference for the defendant's hometown captured in Tables 5 and 6 does not contain information about the judge's preference for local defendants. In addition, it is worth noting that in the regression using the local defendant sample and measuring the plaintiff's success by the proportion of litigation costs borne by the plaintiff (columns (3) and (4)), the judge's preference for the plaintiff's hometown also becomes significant. If the judge and the plaintiff are from the same hometown, the proportion of litigation costs borne by the plaintiff will be reduced by about 15.13%.

4.4Robustness test

In order to test the robustness of the previous results, this paper conducted the following additional tests. First, given that in the binomial equation, a high or low plaintiff win rate in the sample litigation cases will cause the equation estimation results to be sensitive to the cumulative distribution function (CDF) of the latent variable selection, this paper uses different methods to estimate the impact of the hometown relationship between the judge and the plaintiff and the defendant on the plaintiff's chance of winning.

Columns (1)-(3) in Table 7 are the estimation results using the binomial probit model, where the marginal effect signs of the variables and are positive and negative, respectively, and the former fails to pass the 10% significance test. The latter still passes the 1% significance test, indicating that the judge and the defendant are from the same hometown. The probability of the plaintiff's victory is significantly reduced, but the judge and the plaintiff are from the same hometown. The probability of the plaintiff's victory has no significant impact.

Columns (4)-(6) in Table 7 use the difference in the proportion of litigation costs borne by the plaintiff and the defendant as an indicator of the reverse measure of the plaintiff's victory. Since the difference between the proportion of litigation costs borne by the plaintiff and the defendant is between [-1, 1], the truncated tobit model method is used to estimate the equation here. The results show that the signs of the variable and the regression coefficient of the variable are negative and positive as expected, respectively, but after the local plaintiff and local defendant variables are included in the equation, the regression coefficient of the variable does not pass the significance test, and the regression coefficient of the variable still passes the 1% significance test very robustly, indicating that the judge and the defendant are from the same hometown, which significantly increases the proportion of litigation costs borne by the plaintiff, but the judge and the plaintiff are from the same hometown. The impact of reducing the proportion of litigation costs borne by the plaintiff is not significant.

In summary, the regression results in Table 7 are highly consistent with the regression results in Tables 4 and 5 above, indicating that the judge and the plaintiff are from the same hometown and do not significantly affect the plaintiff's chance of winning, but the judge and the defendant are from the same hometown and significantly reduce the plaintiff's chance of winning, which means that the judge does have a significant preference for the defendant's fellow villagers in the judicial process, which supports the robustness found in the previous article.

Secondly, this paper deleted the litigation cases where the defendant was a non-local party and the plaintiff was a local party, and only retained the estimated equations for the litigation cases where the plaintiff was a non-local party and the defendant was a local party, which more cleanly eliminated the possible impact of the judge's regional preference on the regression results. Because, in the litigation cases where the plaintiff was a non-local party and the defendant was a local party, the judge's regional preference for the local party was incorporated into the constant term or random disturbance term in the equation, and would not interfere with the regression results of the defendant's fellow townsman.

Columns (1)-(3) in Table 8 use the Logit model to estimate the impact of the judge and the defendant being fellow townsmen on the plaintiff's chance of winning. The marginal effect signs of the variables are all negative, and they all pass the 1% statistical significance test, indicating that the judge and the defendant being fellow townsmen significantly reduces the plaintiff's chance of winning. Columns (4)-(6) use the tobit model to estimate the impact of the judge and the defendant being fellow townsmen on the plaintiff's litigation fee ratio. The regression coefficient signs of the variables are all positive in each column, and they all pass the 1% significance test, indicating that the judge and the defendant being fellow townsmen significantly increases the proportion of litigation fees borne by the plaintiff. In general, Table 8 is the estimated result after excluding the judge’s regional preference, which is consistent with the previous research findings, indicating that judges have a significant preference for the defendant’s fellow townsman in the judicial system. At the same time, after excluding the judge’s regional preference, the impact of the judge and the defendant being from the same hometown on the plaintiff’s chance of winning and the proportion of litigation costs borne are reduced (compared with Tables 4 and 5). After controlling for all explanatory variables, the judge and the defendant being from the same hometown reduces the plaintiff’s chance of winning by about 25.64% (column (3)) and increases the proportion of litigation costs borne by the plaintiff by about 11.85% (column (6)).

5Judges' Characteristics and Hometown Preference

Judges’ behavior of favoring fellow townspeople due to identity recognition in judicial practice violates the basic principle of judicial fairness, is a great hazard to judicial credibility, and is also an important obstacle to optimizing the business environment. So, whether the behavior of judges favoring fellow townspeople due to identity recognition will be weakened or even eliminated with the improvement of judges’ quality is a question worthy of further discussion and has important policy implications. Because if the improvement of judges’ quality helps to weaken or even eliminate judges’ favoritism towards fellow townspeople, then the phenomenon of judges favoring fellow townspeople will be effectively governed with the development of China’s vigorously promoted work on the formalization, specialization and professionalization of judges. On the contrary, if the improvement of judges’ quality does not help to weaken judges’ favoritism towards fellow townspeople due to identity recognition, then it may be necessary to further improve the existing litigation recusation system to avoid the phenomenon of judges favoring fellow townspeople.

Next, this paper estimates whether the judge's preference for the defendant's fellow townsman will weaken as the judge's quality improves by incorporating variables such as the judge's education level (educ), grade (grad) and age (age) into the regression equation, which reflect and measure the judge's quality (see Table 9).

First, we included the interaction term (×educ) between the judge's education level and the defendant's fellow townsman into the equation (columns (1) and (4)). In theory, judges with higher education may have rich legal theory literacy and stronger rule of law concepts due to their higher education, which may prompt them to be more faithful to the law and case facts in judicial practice, and thus judge cases more objectively and fairly. The results show that the marginal effect of the interaction term ×educ on the plaintiff's probability of winning is positive (column (1)), and the regression coefficient on the plaintiff's litigation cost ratio is negative (column (4)), and the latter passes the 1% significance test, indicating that judges with higher education may have weaker preference for the defendant's fellow townsman.

Second, we include the interaction term (×grad) between the judge's rank and the fact that the judge and the defendant are from the same hometown as each other in the equation (columns (2) and (5)). According to the "Interim Provisions on Judges' Ranks" promulgated by the Organization Department of the Central Committee of the Communist Party of China, the career advancement of judges is based on "the judge's position, moral and professional performance, professional level, trial work performance and years of service". Judges can be promoted "only after the promotion period expires and they have passed the assessment; those who fail the assessment shall have their promotion postponed; those who have outstanding moral and professional performance, professional level and trial work performance may be promoted in advance". Therefore, judges with higher ranks not only have more judicial experience, but also may have better professional level and trial performance, which may make judges with higher ranks more objective and fair in judging cases. The estimation results show that the marginal effect of the interaction term ×grad on the plaintiff's chance of winning is positive (column (2)), and the regression coefficient on the plaintiff's litigation fee ratio is also positive (column (5)). The latter passes the 1% significance test, indicating that the scale of the plaintiff's litigation fee ratio increased by the judge's hometown being the same as the defendant increases with the judge's rank, which means that the judge's preference for the defendant's hometown does not weaken with the judge's rank.

Finally, we also include the interaction term (×age) between the judge's age and the judge's hometown being the same as the defendant into the equation (columns (3) and (6)), because older judges tend to have longer work experience and richer judicial experience. The regression results show that the marginal effect of the interaction term ×age on the plaintiff's chance of winning is negative [column (3)], and the regression coefficient on the plaintiff's litigation fee ratio is positive [column (6)]. The latter passed the 1% significance test, indicating that the scale of the increase in the proportion of litigation fees borne by the plaintiff when the judge and the defendant are from the same hometown will increase with the judge's age, which means that the judge's preference for the defendant's fellow townsman will not weaken with the judge's age.

In general, the regression results in Table 9 do not show too many positive signals. Although the regression results in columns (1) and (3) show that the judge's preference for the defendant's fellow townsman may weaken with the improvement of the judge's education level, the regression results in other columns show that the judge's preference for the defendant's fellow townsman will not weaken with the improvement of the judge's rank and age. Therefore, we have not yet obtained sufficient empirical evidence to support the fact that the improvement of judges' quality will weaken the judge's preference for fellow townsman behavior. This means that simply relying on the formalization, specialization and professionalization of judges may not effectively govern the judge's preference for fellow townsman behavior due to identity recognition in the judicial system. A more effective correction measure may be to improve the existing litigation avoidance system.

6Concluding remarks

As China's economy enters a new normal and the international economic situation and pattern are undergoing major changes, the CPC Central Committee and the State Council have made major decisions and arrangements to optimize the business environment, which has made the Chinese economy's need to establish a high-quality judicial system increasingly urgent. However, it is no secret that, against the background of continuous improvement of legislation, the problem of judicial preference in China's judicial system has been widely criticized, and judicial preference is also an important obstacle to optimizing the business environment. There are existing literature on judicial preference, which has extensively discussed the problem of judicial preference caused by judicial localization and administrativeization, but another more hidden form of judicial preference, that is, judges may prefer fellow villagers in the judiciary due to identity recognition, has not yet received attention and attention from the academic community.

Fellow villagers are the continuation of clan culture rooted in traditional rural society. They are the relationship network linked by hometown after individuals leave the original relatives and acquaintances society and enter the stranger society. Individuals' identity recognition of fellow villagers is a positive emotional involvement and convergence tendency towards the language, culture, concepts, customs, etc. of their hometown people, which will eventually be externalized as mutual help among fellow villagers. Therefore, when clan culture rooted in traditional rural society permeates into judicial activities, whether it will cause judges to favor fellow villagers due to identity recognition is a research topic with important theoretical and practical significance.

This paper reveals for the first time in theory and experience the possible behavior of judges favoring fellow villagers due to identity recognition in judicial practice. Based on the micro data matching presiding judges with commercial litigation cases, the study found that after controlling for variables such as judges' regional preference, judges' personal characteristics and case characteristics, although the judges' fellow villagers with the plaintiff had no significant impact on the plaintiff's chance of winning and the proportion of litigation fees borne, the judges' fellow villagers with the defendant significantly reduced the plaintiff's chance of winning and increased the proportion of litigation fees borne by the plaintiff. This result is robust when using different estimation methods and excluding the judges' regional preference, indicating that judges do have significant preference for the defendant's fellow villagers in judicial practice. At the same time, further research has not found empirical evidence that judges' preference for fellow villagers will change significantly with the judges' education, rank and age, which means that judges' preference for fellow villagers due to identity recognition may not weaken with the improvement of judges' quality.

This article expands the research perspective on judicial favoritism, enriches the research literature on the problem of favoritism that may be caused by identity recognition, and helps to more comprehensively understand and observe the impact of clan culture in contemporary social economy. At the same time, at the policy level, the behavior of judges favoring fellow villagers due to identity recognition violates the basic principle of judicial fairness and is a difficult problem that China needs to pay attention to and solve in order to optimize the business environment and promote the rule of law. However, the empirical evidence provided by this article shows that the behavior of judges favoring fellow villagers due to identity recognition may not weaken with the improvement of the quality of judges. Therefore, to effectively govern the behavior of judges favoring fellow villagers due to identity recognition, it may not be possible to simply rely on the formalization, professionalization and specialization of judges. A more effective measure may be to improve the existing litigation recusation system.

The original article was published in the 6th issue of "China Economic Issues" in 2022 (pages 82-98), and was reprinted from the WeChat public account "Judicial Big Data".

Assistant Editor: Yang Ziyue

Editor-in-Chief: Tan Jun

Review and Reading: Ji Weidong