The Curious Case of China’s Guiding Cases System

The cases selected by the Supreme People’s Court (SPC) to serve as non-binding, but highly persuasive, “guiding cases” (77 issued from 2010-2017) usually stand for single bright-line legal rules that could be more easily and directly expressed through a binding judicial interpretation (30 issued in 2016), and then illustrated and elaborated on through “Model Cases” (149 such cases (典型案例) issued by SPC in 2016 ), as necessary.  If you aren’t clear on the difference between guiding cases and Model cases, you aren’t alone, but guiding cases can be quoted in judicial opinions, while Model Cases, like the millions of other ordinary Chinese cases made available online, cannot.

I have written before about the reasons why I think the system has not yet had a meaningful impact. Still, when I received a notice about a recent uptick in references to guiding cases, I thought I might need to reconsider. After taking another look, what I have found not only confirmed my impression of the guiding case system, but was also a painful reminder of the challenges and limitations of doing legal research in China. What should have been a quick inquiry became a wild ride, chasing the research trail deep down the rabbit hole.

Measuring Impact:

One way in which researchers have attempted to measure the success of the guiding cases (GC) system is by looking at how often other cases refer to the GCs. The idea is that the number of times a case is later referenced gives some reflection of how influential it is. This isn’t necessarily true, as sometimes a very influential ruling might actually result in fewer subsequent references, such as where a judgment holds that a particular type of claim will no longer be recognized, and parties simply stop raising such actions entirely.  Still, given the lack of reliable alternative indicators, the number of subsequent references to a GC is a seemingly straightforward, if imperfect, quantifiable measure, which provides some insight into the systems’ use.

Two major studies published almost back-to-back, one by Stanford’s China Guiding Cases Project[i] and one by Peking University (PKU), have considered the number of cases referencing GCs through the end of 2015. Both reports contain a wealth of authoritative analysis, and note the same major trends:

• Guiding Cases are almost never referenced: Over a five-year period, Stanford found a total of 181 subsequent cases, and PKU found 241.[ii] To provide a frame of reference, Chinese courts complete trial of well upwards of 10,000,000 cases per year.[iii]
• 50% of the guiding cases were never referred to at all. *Note though that cases 45-52 were only released in late April 2015, and numbers 53-56 were only released in November of that year, leaving little time for their application before the studies were released. These later cases, however, only amounted to about 20% of the total number of guiding cases in 2015.
• Almost half of the references found were to a single case; GC #24. [42.67% PKU; 41.98% Stanford]. That case concerns traffic accidents, and stands for a form of the ‘eggshell skull rule’ holding that the frailty of a faultless injured party does not reduce the other party’s liability for harm caused to them. It may be that lawyers working for insurance companies are sufficiently focused and vested in such a holding that they bring it up more frequently as compared to other cases, but the use of even this case is still exceedingly rare.

Although the GCs are handpicked to address common legal questions, it seems that either the courts don’t find them all that useful, or that simply counting the number of subsequent referring cases isn’t adequately capturing their impact. A new legal document released in 2015 requires judges to quote any guiding cases they consider in their written opinions, and this may bring more references to guiding cases to the surface.

Considering the Studies

It doesn’t immediately seem all that strange or significant that Stanford and PKU would find different numbers of subsequent cases. The researchers were working independently and may have used slightly different standards or research tools that led to more cases being excluded from Stanford’s list. It only means that those differences need to be teased out in order to understand what each team was saying.

In the following chart, data prepared by each study is placed side-by-side showing the number of cases referring to each GC. The green columns indicate guiding cases where more referring cases were found by PKU, red for where the Stanford Guiding Cases Project found more. Grey columns are those for which neither study found any referring cases at all, and the sole unshaded column shows where they both found the same case. You can see that for GCs that had any hits at all, the studies were almost never in exact agreement.

I had originally assumed that Stanford’s 181 cases quoting GCs were a subset of PKU’s 241, but looking at the differences between the two studies’ totals reveals a more complicated situation. Stanford found more, not less, subsequent cases relating to nine of the GCs [16%]. Merging the case sets from both studies and eliminating all duplicates leaves a pool of 286 unique cases- only 134 of which [about 47%] were found by both studies. This shows that each study includes cases that the other did not- meaning that it was possible that they weren’t talking about the same thing at all.

What do you mean by  ‘subsequent case?’

I had thought after a casual read-through of Stanford’s Guiding Cases Surveys^TM, Issue No. 2 that a subsequent case was one in which the court quoted a guiding case in its written opinion. After all, the most unique and exciting feature of the guiding case system, is that judges are permitted to quote them in explanations of their legal reasoning. On a more careful reading, it is less clear.

The 181 ‘subsequent cases’ Stanford identified are labeled as ‘explicitly’ referring to guiding cases. Their report indicates, however, that in 96 of these, the deciding court does not mention the GC at all, and in 38 others, the court is said to reference the cases either implicitly or explicitly.

A distinction is made as to whether the issue of a guiding case was first raised by the court itself, or by a party, with parties raising the GC in 134 cases, with the court responding in only 38 cases. As the chart to the left attempts to present, 85 cases (highlighted in yellow) are said to contain a court’s reference to a GC, either in response to a party’s argument or sua sponte.

For Stanford, the criterion for identifying subsequent cases seems to be that any case party or the court expressly refers to a guiding case. Party submissions and court transcripts are not easily available, so if no court ever referred to a GC, it would be difficult to identify as having referenced a guiding case- but sometimes second-instance cases mention that a party raised the issue of a relevant guiding case at the level below. A quick random spot check of some identified cases shows that the phrase ‘guiding case’ generally does appear in all cases, except a few, where a higher court mentions that it was raised at the court below.

What isn’t tallied is how often parties mention cases other than the guiding cases; be it in the Beijing IP court that specially allows such citations and handles about 4,000 cases a year, or in other courts, where lawyers have been known to introduce other cases as persuasive authority. In the latter situation, there is unlikely to be any record in the written opinions, but we might at least learn something of the former through an empirical study of the IP court’s published opinions.

PKU breaks down its data differently, but also addresses the explicit/Implicit distinction, and the active or passive application of the cases. They refer to the ‘application’ of GCs rather than ‘reference’ or ‘consideration’ of the cases as Stanford does, but looking at their analysis suggests they probably mean the same thing, and are not making a distinction based on the depth of analysis.

The chart below attempts to present some of PKUs data in a way to make it easily comparable to Stanford’s, but cannot guarantee I have done so correctly. PKU is on the right.

PKU identified 241 total cases, and noted that judges made explicit applications of guiding cases in 79 of them, and implicit application in 156 of them.[iv] The 79 explicit application cases are divided into active application- where the court brings up the issue of guiding cases themselves (59 cases), and those made in response to a party raising the issue (20). The team identifies 156 cases as including an implicit application, which is generally defined as the situation where a party has raised a guiding case and the court does not directly address them, but makes a ruling in accordance with the guiding case’s holding.[v]

I am not 100% sure if the area in the white ‘Party Raised’ section of my diagram for each study represents the same thing, but am fairly confident it does, and would welcome further insight.

But, if the studies are looking at the same issues, why did PKU and Stanford identify different cases as referencing/applying the GCs?

Timing is everything

One simple difference between the studies is the period in which they collected data. The PKU team makes clear that they stopped collecting data on 11/25/2015, while Stanford says they collected data through December 2015.

As Stanford points out, however, cases do not appear online quickly, and only 45% of the cases that they found were uploaded within 50 days, with 33% taking over 100 days to go online.[vi] This percentage is only of the 181 subsequent cases they found, however, and does not consider other cases that took so long to go up they weren’t online at the conclusion of the study, or that have never been uploaded at all.

Neither study would have been able to accurately capture to all cases tried during their research period, but the extra month of research would have allowed Stanford to discover additional cases as they came online, even some that were decided months before. The end of the year in particular can lead to a rush of new cases being uploaded as uploading cases is part of judges’ performance evaluations. Based on the data in Stanford’s appendix, 19 cases were uploaded in the last month, only 5 of which were adjudicated in that month. This cannot explain, however, why Stanford found fewer total cases.

The unpredictable delays in Chinese cases being uploaded, and the large number of cases that simply never appear online, make it almost impossible to ever feel confident that a research sample is complete. Sometimes the information available online seems to be in a constant state of flux with new cases appearing and even old cases being taken down. This may just be growing pains, as the courts adjust to new case publication requirements, but it undeniably undermines research today.

Some Databases are more equal than others?

The two studies also used different databases. PKU used its own collaborative project’s database pkulaw.cn, while Stanford used the courts’ public database. This might have a direct impact in terms of case availability, and could also have an indirect effect based on the quality of each website’s search engine.

Going to the court website now and doing a search for the term ‘指导案例’ , a common way of referring to guiding cases, and limiting the search to dates through 12/31/2015, shows 752 results—more than both of the two studies findings combined! And that’s without even trying alternate search terms- such as the specific case name, or ‘指导性案例’. The same search on a third database, shows 840 hits.

From past experience, however, I know that some of these hits will be duplicates that the site has not caught and removed. In another research project, I have seen five separate instances of a single case entered in the courts’ database, with publication dates changing, case numbers being updated, different names being redacted, and other minimal changes being made each time. The databases are also plagued by problems beyond their control, like courts failing to upload opinions, the use of identical case names for unrelated cases[vii], and so forth. Without looking carefully at each case, you can’t really know the sample size.

Conclusion

The defining characteristic of guiding cases is that they can be quoted in opinions. That something so seemingly straightforward as counting the number of times they were actually quoted has become so complicated is a testament to how hard it is to do research like this in China. This isn’t just a problem for scholars, but for judges and legislators as well, as they try to understand China’s legal landscape.

The difference between 181 cases or 241 referencing cases isn’t critical – either number is a long way from showing that the systems are working. The GC system is still young, and may become more prominent as time goes on, but already other more organic attempts to create a body of common law, like that of the Beijing IP court, seem to be moving forward more smoothly. What is more important is the need for continuing efforts towards greater transparency so that a reliable and clear picture of judicial practice can take shape.  This includes not only ensuring that online case databases are complete and easily accessible, but also that judges feel comfortable fully explaining their reasoning and the basis of their opinions.

[i]  Mei Gechlik, Minmin Zhang, and Liyi Ye, Cumulative Analysis of All Subsequent Cases Referring to Guiding Cases in China (2014 Q1-2015 Q4), STANFORD LAW SCHOOL CHINA GUIDING CASES PROJECT, Guiding Cases Surveys^TM, Issue No. 2, Jan. 2016, http://cgc.law.stanford.edu/guiding-%20cases-surveys/

[ii] Two of these turned out to be duplicated, so the actual total was 239.

[iii]  16,713,793 in 2015 alone. http://www.dffyw.com/sifashijian/ziliao/201603/40182.html

[iv] Two notes here, 1) they actually found 139 cases after two duplicates are removed, but I do not know where they fall on this chart so have not adjusted the numbers. 2) 6 cases, not featured on the chart, were cited in a unique way that Stanford did not consider, but attaching  or citing secondary materials.

[v] 1 of these cases is a ‘broad meaning’ implicit application, which I have not addressed here, it is only one case- but seems to be where the court does cite a GC, but doesn’t say they are doing so.

[vi] See Stanford at note 1, Section IV.

[vii] See e.g.:  (2015)高行初字第3号 and (2015)高行初字第3号