[author]BIAN Renjun
[content]
*Author Bian Renjun
Abstract：China has been issuing IPR policies intensively since 2008. One of the primary goals of these policies is to deter infringements by raising their costs. The legislature follows the policies closely by raising the lower and upper limits of statutory damages and introducing punitive damages. In response, courts at all levels have started to grant higher-than-ever patent damages and vigorously publicized these cases in official reports. This shows that, in both public policies and legislative and judicial activities, “increasing damages can deter patent infringement” has been taken as a self-evident rule. However, pursuing high damages to archive strong deterrence has caused a large number of theoretical and practical dilemmas. For instance, punitive damages of up to five times have become available in all three major fields of intellectual property law since 2020, making Chinese Mainland a jurisdiction with the highest multiple and widest coverage of punitive damages around the globe. But a crucial theoretical question ⁃ how to justify the high multiple and wide coverage ⁃ has been left unanswered. In addition, the Chinese Patent Law has a lower limit for statutory damages, which has increased from RMB 5,000 yuan in 2001 to RMB 30,000 yuan in 2020. As a result, courts are faced with the dilemma of either violating the blackletter law or overcompensating the patent holder. In practice, it is common for Chinese courts to break the lower limit due to the low value of the patent-at-issue and the minor loss of the patent holder. This article first examines the “self-evident rule” of “increasing damages can deter patent infringement” and points out its limitations. Then, taking the more advanced deterrence theory in criminology as a reference, it constructs a theoretical model of the deterrent mechanism for patent infringement damages with multiple dimensions and intermediary conditions. In the end, this article conducts empirical analyses to test the above theories. It finds that, surprisingly, high damages have a very limited deterrent effect, while other two factors that are often overlooked ⁃ the certainty and timeliness of damages ⁃ present a significant and stable deterrent effect. Moreover, as far as the amount of damages is concerned, its subjective attributes have more deterrent effects than its objective attributes. Therefore, it is necessary to rethink the issue of whether to abolish the legal rules that simply pursue high damages and build a multidimensional deterrent mechanism with high certainty and timeliness in order to achieve an optimal deterrent effect.
1. formulation of the problem
Damages are often used as the main institutional tool to deter intellectual property infringement, both in the formulation of public policy and in legislative and judicial activities. At the same time, the use of damages for deterrence is often simplified to the use of the amount of compensation for deterrence, “high compensation” and “strong deterrence” is closely linked. However, in recent years, the disadvantages of simply pursuing high compensation have become more and more prominent, creating a large number of theoretical dilemmas and practical conflicts for legislation and justice. For example, in 2020, China has realized punitive damages of up to five times covering the entire field of intellectual property rights, which is rare in the world in terms of coverage and punitive multiplier. The theoretical question of the basis of its legitimacy has been left unanswered. Another example is that China's Patent Law has always set a lower limit for statutory damages, from five thousand dollars in 2001 to thirty thousand dollars in 2020. However, in judicial practice, a large number of infringement cases involving patents of low value, the right holder of the loss of minor, break through the lower limit of the legal compensation of the case is common, the court is in violation of the rules of the law and let the compensation of the dilemma of excessive.
With this kind of “high compensation” tendency is in strong contrast, China's intellectual property law for damages to play a deterrent effect of the specific mechanism of the research is relatively lack of. Improve the effect of the amount of compensation, can effectively deter infringement, the use of damages to play a deterrent effect of whether there are other alternative or even better dimensions, etc., are urgently need to study. This paper takes the patent infringement damages and its deterrence mechanism as the research object, and will construct both three-dimensional dimensions and intermediary conditions of the patent infringement damages deterrence theoretical model on the basis of the newly constructed theory of deterrence of patent infringement damages through the distribution lag model to test the theory of deterrence of patent infringement damages, to explore whether the relevant theory is in line with China's judicial practice, and then to discuss how to adjust China's macro policy, legislation and judicial activities related to patent infringement damages, and how to adjust China's patent infringement damages and deterrence. We will then discuss how to adjust China's macro policy, legislation and judicial activities related to patent infringement damages in order to obtain the optimal deterrent effect.
2.Deterrence theory construction: three-dimensional dimension and intermediary conditions
Similar to the traditional civil law research, the discussion of deterrence theory in intellectual property law is mostly placed under the macro topic of the function and effect of damages, and there is little specialized research on deterrence theory and its operation mechanism. This section intends to draw on the more in-depth study of the deterrence theory of criminology results, combined with the characteristics of patent infringement, trying to build both three-dimensional dimensions and intermediary conditions of the patent damages deterrence mechanism theoretical model.
2.1 patent infringement damages should pursue deterrent effect
Intellectual property law has been concerned about the issue of damages, about its function and effect of the study is not a few, mainly focusing on compensation, deterrence and punishment of three points. It is generally believed that the rule of intellectual property damages should pursue the dual effect of compensation and deterrence. The compensation function aims to fill the damage and restore the interests of the right holder to the state in which the infringement did not occur. Domestic scholars mostly trace back to the principle of comprehensive compensation of civil law and apply it to intellectual property law through the logical reasoning from general to special. The civil law system countries to comprehensive compensation for the basic principle of tort damages, China's doctrine has been the principle of comprehensive compensation as the basic principle of the whole civil compensation.
Deterrence function, also known as the preventive function, usually from the rational person assumption, hope that by depriving the infringement benefit, guide the potential infringer to give up infringing behavior, the existing research often deterrence function as the compensation function can not be realized as a suboptimal choice. On the one hand, from the intangible nature of intellectual property rights, the causal relationship between the infringement and the consequences of the damage becomes complicated, and there may even be a situation where the economic interests of the right holder are instead gained after the infringement; on the other hand, with the development of network technology, infringement is frequent, and in many cases, it cannot be detected or prosecuted. In other words, in the context of intellectual property infringement, it is very difficult to fill in individual cases, and even if individual cases can be realized, it does not certainly make up for the overall loss of social welfare. Therefore, most intellectual property scholars believe that damages should play a deterrent function to correct the systemic failure of the compensation function.
It should be noted that, although some scholars also advocate that the function of punitive damages also includes punishment, but the general view is that, even if punitive damages, its main function is still deterrence, not punishment. Therefore, in view of the purpose of this paper is to discuss whether to increase the damages can deter patent infringement of this general problem, there is no need to exclude the circumstances of punitive damages or separate distinction. In addition, although the existing literature does not distinguish between patents, copyrights and trademarks, however, as the lower concept of intellectual property rights, patent infringement damages should pursue the dual effect of compensation and deterrence is not controversial.
2.2 Theoretical Construction of Patent Infringement Deterrence
Taking the rational man hypothesis as the logical starting point, it may be intuitively concluded that raising compensation can strengthen the inference of deterrence, which is also the theoretical basis for frequent raising of damages in recent years in public policy, legislation and judicial activities. But does high compensation necessarily lead to strong deterrence? Or from the bottom of the logic, raising the cost of engaging in a particular behavior, so that the loss is greater than the gain, can necessarily play a reverse incentive effect? Not exactly. The theory of penal deterrence, which also starts from the rational man assumption, has given a negative answer.
Deterrence theory originated from the classical school of criminology, Bianqin and Beccaria from the “rational man hypothesis”, put forward the potential offender can be based on cost-benefit analysis and the instinct to avoid harm to make a rational decision to commit a crime or not. However, modern criminological theory and empirical research results have questioned the “rational man hypothesis” and the deterrent effect of punishment. Some scholars have argued that criminals are not rational in the economic sense, and that their reasons for committing crimes are complex and varied, involving physical, psychological, social and cultural aspects. Some empirical studies have also pointed out that the restriction of the death penalty policy has not led to an increase in the macro-crime rate, thus negating the deterrent effect of the penalty from the practical level.
The clash of views and the frequent exchange of theories and empirical studies have promoted the development of the theory of penal deterrence, and the design of the relevant system has undergone a metamorphosis from feloniousness to a combination of three-dimensionality and intermediary conditions. The three-dimensionality means that the deterrent effect of punishment involves at least three dimensions: certainty, timeliness and severity. Certainty refers to the probability that an offender will be punished for a particular offense, timeliness refers to the time interval between the offender's commission of the offense and the imposition of the penalty, and severity refers to the severity with which the offender is punished for a particular offense. Theoretically, other things being equal, an increase in either dimension should lead to an increase in deterrent effectiveness and a decrease in crime rates. However, empirical data find that the deterrent effects of the three dimensions mentioned above are not identical. In particular, severity, whose deterrent effect is controversial, has been steadily diminishing in importance.
Transplanting the three dimensions of deterrent effectiveness of penalties to the patent field is equally applicable. On the one hand, a rational potential infringer will evaluate the costs and benefits before making an infringement decision, and the main component of the costs is the product of the probability of compensation and the amount of compensation, which is the certainty and severity of compensation. On the other hand, the characteristics of human nature determines that most people are only concerned about the immediate costs and benefits, and relatively insensitive to the costs and benefits of the more distant time nodes, which is the timeliness of compensation. In other words, similar to the deterrent effect of criminal punishment, the damages in the field of patents to play a deterrent effect, also involves the three dimensions of certainty, timeliness and severity. Moreover, theoretically, the enhancement of any of these dimensions will lead to the strengthening of the deterrent effect of damages, and thus curb infringement behavior.
The complexity of the mechanism by which penalties exert deterrent effects involves mediating conditions in addition to their three-dimensionality. It is generally believed that the objective attributes of the penalty cannot act directly on the offender's behavioral decisions, and need to be converted into the offender's subjective perception of the penalty before it can exert an influence on his or her subsequent decisions. The Norwegian criminalist Johannes Andenæs has pointed out that “the decisive factor in achieving the desired deterrent effect is not the objective existence of the captured danger itself, but the subjective judgment of the potential offender of that danger.” Experts in criminal law and criminology in China also recognize the above viewpoint, and refer to the subjective perception of the penalty by the offender as the “intermediary condition” or “intermediate link” for the deterrent effect of the penalty to be realized.
Similar to the deterrent effect of the penalty mechanism, patent infringement damages to play a deterrent effect also involves intermediary conditions. This is because both need to convert the objective change of a legal attribute into the subjective perception of the perpetrator, and in the process of information transfer, it is inevitable that there will be bias or even error. Statistics show that most patent infringers in China are individuals, small enterprises and other subjects lacking legal knowledge, whose channels for obtaining information are very limited and easily influenced by the media and individual cases. Even for large enterprises with well-developed legal departments, it is by no means easy to accurately identify class cases and understand the level of remedies in such cases when the channels are open and supported by a number of official and third-party databases. Therefore, what may happen is that the objective level of damages has changed, but because the relevant information cannot be accurately and timely transmitted to potential infringers, the individual infringement behavior and the overall infringement rate do not follow the changes in the objective level of damages.
In summary, in the field of patents, the mechanism by which damages exert a deterrent effect is very complex. On the one hand, the deterrent effect is not only a single dimension of severity, but also involves certainty and timeliness, which is the three-dimensional characteristics of the deterrent mechanism of patent infringement damages; on the other hand, the above three dimensions in order to play a deterrent effect, need to be perceived by the potential infringer, which is the intermediary condition characteristics of the deterrent mechanism of patent infringement damages. The above mechanism is represented by a flow chart, as shown in Figure 1:
Figure 1 Theoretical Model of Patent Infringement Damage Deterrent Mechanism
It should be noted that the deterrent effect of patent infringement damages is realized by depriving economic benefits, and the deterrent effect is inevitably different from the deterrent effect of criminal punishment through the imposition of coercive measures on the human body in terms of effect and path. Therefore, this part of the deterrence mechanism model of patent damages constructed with reference to the deterrence theory of punishment, only for the theoretical model that has not been tested in practice, and needs to be interpreted and further improved in combination with the empirical results of the third and fourth parts below.
3. Test of Deterrence Theory: Research Design
3.1 Research Questions and Research Assumptions
In the process of constructing the theoretical model of the deterrence mechanism of patent infringement damages above, there is a process from shallow to deep, that is, first examining the simple model of “increasing damages can deter patent infringement”, pointing out its limitations, and then proposing a complex model with three-dimensionality and intermediary conditions on the basis of this model. The empirical tests in this Part and Part IV will also repeat the above process, first exploring whether the simple model is consistent with judicial practice, and then empirically testing the three-dimensionality and mediating conditions. In other words, the following section intends to answer, through empirical data: first, whether increasing the amount of damages (i.e., severity) can curb infringement; second, whether increasing the certainty and timeliness of damages can curb infringement; and, third, whether an objective change in one of the dimensions of damages needs to be converted into the infringer's subjective perceptions first in order to have the effect of curbing infringement.
It is worth noting that testing the mediating condition from a theoretical model logically requires two hypotheses on each of the three dimensions: objective level and subjective perception. Unfortunately, with the exception of severity, the mediating conditions of certainty and timeliness require questionnaires or interviews that are beyond the reach of the case database constructed in this paper. Therefore, this paper only uses severity as a proxy to set hypotheses for its objective level and subjective perception, while, in the subsequent conclusion section, it only comments on whether severity exerts a deterrent effect with the qualities of mediating conditions. The hypotheses to be tested in this paper include:
Hypothesis 1-1: When the objective level of the amount of damages ordered by the court to be paid by the infringer increases, the corresponding patent infringement will decrease;
Hypothesis 1-2: When the subjective level of the amount of damages awarded to potential infringers increases, the corresponding patent infringement will decrease;
Hypothesis 2: when the certainty of the court ordering the infringer to pay damages increases, the corresponding patent infringement will decrease;
Hypothesis 3: when the timeliness of the court ordering the infringer to pay damages is strengthened, the corresponding patent infringement will decrease.
3.2 research method
Both the existing simple model and the complex model constructed in this paper are essentially causal models, which believe that there is a causal relationship between the specific dimension of damages and the specific behavior of potential infringers. Randomized controlled trials are ideal for testing causation, but are not common in the HSS. As a result, the search for causation in the humanities and social sciences has often had to retreat to quasi-experiments that draw on seemingly random differences produced by external circumstances.
Take, for example, the causal relationship between the severity of damages and the infringement of rights, which is the subject of this paper. Due to the existence of various subjective and objective factors, the amount of compensation awarded by the court is not static; some periods are higher and some are lower. Then, the period with higher compensation can be regarded as the experimental group, and other periods with lower compensation can be regarded as the control group. If, controlling for other possible influences, the rate of infringement in the experimental group is significantly lower compared to the control group, it can be concluded that higher damages lead to a decrease in the macro-rate of infringement, i.e., that there is a causal effect between the severity of damages and infringement.
It is worth noting that, from the perspective of the research question in this paper, considering different periods as subjects receiving different treatments is not the only feasible experimental design, but also considering different courts or regions as control and experimental groups, or from the perspective of changes in the law (e.g., the Patent Law of 2020 raises the upper and lower limits of the statutory damages and introduces punitive damages), and considering before and after the adoption of the law as control and experimental groups, respectively. experimental group. However, this paper argues that regression analysis on time-series data is most desirable for two main reasons. One, the relationship between damages and torts is not only causal, but also dynamically causal. That is, changes in the damages dimension in one period may not only affect the macro-infringement rate in the same period, but may also have an impact on the state of infringement in subsequent periods. Second, this paper seeks to explore the underlying logic of how damages affect tort behavior, rather than focusing on specific institutional tools. In other words, this paper hopes to provide factual data not only for the existing debate over statutory and punitive damages, but also for any future macro-policy, legislative, and judicial activity that seeks to adjust the severity, certainty, and timeliness of damages to deter infringement.
In summary, this paper intends to estimate the dynamic causal effect between damages and infringement through distributional lag modeling.
3.3 Data Source and Sample Scope
Starting from the research problem and research methodology, the data required in this section have not been counted by relevant organizations or previous scholars, therefore, it is necessary to use the adjudication documents as the original material to obtain relevant data and establish a data set.
Based on the comprehensiveness of the case collection and the convenience of access, this paper takes “Beida Law Treasure” as the data source, and at the same time, restricts the sample scope in terms of time and trial level. In terms of time, this paper only examines patent infringement cases concluded within the scope of January 1, 2014 to December 31, 2020, in order to take into account the timeliness and sample representativeness. In terms of trial level, this paper only examines patent infringement cases of first and second instance, and does not involve retrial cases. Therefore, the sample scope of this paper is the patent infringement cases concluded by national courts between January 1, 2014 and December 31, 2020, except for retrial.
According to the above conditions, the authors of this paper carried out in the “Beida Fabulous” with the conditions of “subject matter: infringement of invention/utility model/design patent disputes” “type of instrument: judgment” “Decision Date: 20140101-20201231” “Trial Procedure: First/Two-Instance” advanced search, a total of 23,389 judgment documents that meet the requirements were collected. After that, the judgments that did not meet the search conditions [e.g., “(2019) Yu 01 Zhi Min Chu No. 825” judgment was a judgment of tenure dispute] and the first-instance judgment of the appeal cases were eliminated, and finally 19,596 cases that met the requirements were obtained.
3.4 Variable Setting
Starting from the research problem and research method, and considering the accessibility and representativeness of the data, the variables in this section are selected as follows:
3.4.1 Dependent variable: macro infringement rate
Criminological tests of deterrence theory mostly use the crime rate - the number of criminal cases per 100,000 population - as the dependent variable. Corresponding to the patent field, the ideal variable to describe the macro-infringement situation should be the infringement rate, i.e., the number of infringement cases per specific number of active patents.
However, the statistical caliber used to calculate the number of infringement cases is a thorny issue. Theoretically available parameters include: the number of actual infringement cases, the number of infringement cases sued by right holders and the number of infringement cases recognized by the court. Among them, for the number of actual infringement cases, there is no available official or third-party data, nor can a researcher or a team collect and obtain them within limited time and resources; for the number of infringement cases sued by right holders, although there is available official data, the calculation of the infringement rate based on this cannot exclude the situation that the plaintiffs have made a mistake in judgment or even sued in bad faith. Therefore, taking into account the accessibility and representativeness of the above parameters, this section will calculate the infringement rate on the basis of the number of infringement cases recognized by the court, which is expressed by the formula as follows:
Macro Infringement Rate = Number of cases in which the court found infringement / Number of valid patents (10,000)
It is worth noting that calculating the infringement rate based on the number of cases in which the court found infringement is not perfect, and there are at least two problems. One, it is impossible to take the overall rise in the number of cases into account. According to the data released by the State Intellectual Property Office, the number of new and concluded civil patent cases in China's courts has increased year by year. In this case, if the infringement rate also shows a rising trend, it is not ipso facto possible to conclude that the infringement situation is getting more and more serious. Second, there may be a dark number of problems. Patent infringement behavior is scattered and hidden, in many cases can not be found, even if found, the patentee based on cost and benefit considerations, not all will go to court. Therefore, compared with the actual infringement rate, the macro-infringement rate calculated in this paper based on the number of cases in which infringement was found by the court may be on the low side.
In order to minimize the impact of the above problems on the statistical results, this paper proposes to take two measures. First, in the process of causal inference, “case base” and “year-end effect” are introduced into the model as control variables, so as to exclude the impact of year-on-year increase in the number of cases and the concentration of court cases at the end of the year on the macro-infringement rate. Secondly, the model not only takes the infringement rate of all patent infringement cases as the dependent variable to make causal inference, but also makes additional causal inference for a specific subset of cases - those in which the plaintiffs and defendants are both companies. It is generally believed that corporate-type plaintiffs are superior in terms of financial resources and expertise, and are more capable of detecting infringement and bearing the subsequent costs of litigation than plaintiffs of other subject types. At the same time, corporate-type defendants are also easier to monitor and recognize than general defendants due to the wider scope of their production and business activities. Therefore, compared with all patent infringement cases, the subset of cases in which the plaintiffs and defendants are all corporations is more controllable in terms of the dark number problem, and the relevant statistical results are closer to the actual infringement situation. In addition, this paper intends to analyze the time series data rather than panel data, which can further weaken the influence of dark number by paying more attention to the trend of infringement rate over time than the absolute value.
3.4.2 Independent Variables
For hypothesis 1-1, hypothesis 1-2, hypothesis 2 and hypothesis 3, the independent variables and their definitions to be selected in this section are respectively:
Average amount of damages: the average value of the amount of damages in cases where the court ordered the infringer to pay damages in a specific time frame. Number of large damages cases: the number of cases in which the court ordered the tortfeasor to pay damages amounting to more than one million dollars in a given time frame. Probability of damages: the ratio of cases in which the court ordered the infringer to pay damages to all patent infringement cases within a specific timeframe. Damages Lag Time Mean: the time in years from the occurrence of infringement to the court ordering the infringer to pay damages within a specific timeframe.
3.4.3 Control Variables
Potential infringers, as people in society, will not only consider the legal consequences of their decision to infringe or not, but will also be affected by factors such as their socio-economic status and cognitive level. Therefore, in addition to the case base and year-end effects mentioned above, this paper will also select a number of other control variables to be included in the model. Specifically, this paper intends to select GDP and per capita disposable income of urban and rural residents as the variables to measure the socio-economic situation, and at the same time select the proportion of people with education above high school and the proportion of people with education above bachelor's degree to reflect the cognitive level of the parties.
4. Deterrence Theory Test: Presentation of Results
4.1 Statistical Description: Overview of Torts and Damages
4.1.1 Overview of Infringement Behavior
Figure 2, i.e., the trend graph describing the macro patent infringement rate in China over time. It can be seen that since 2014, China's patent infringement situation has the following characteristics in the time trend.
First, China's macro patent infringement rate is generally on an upward trend, especially between 2015 and 2019. The minimum value appeared in the first quarter of 2015, and every 10,000 valid patents involved only 0.60 infringement cases; the maximum value appeared in the fourth quarter of 2018, and every 10,000 valid patents involved 5.22 infringement cases, which was 8.7 times more than that of the first quarter of 2015.
Figure 2 Time trend graph of China's macro infringement rate (in quarters)
Secondly, China's macro patent infringement rate shows obvious seasonal changes, and the year-end effect is obvious. Taking 2020 as an example, the macro infringement rate in the first quarter was the lowest for the whole year, only 1.35; the infringement rates in the second and third quarters were the third and the second, 2.62 and 2.32, respectively; the infringement rate in the fourth quarter was the highest for the whole year, reaching 3.23, which was about 2.39 times of the first quarter.
Figure 3 Time Trend of Infringement Rate of Patent Infringement Litigation in China (in quarterly basis)
However, from the above trend, it does not automatically lead to the conclusion that the patent infringement situation in China is getting more and more serious in a particular period of time. From the definition of macro-infringement rate as the number of infringement cases per 10,000 active patents in the “Variable Setting” section above, there are at least three possible explanations for the increase in macro-infringement rate: overall increase in the number of cases, relaxation of infringement standards and increase in infringement behaviors. After reviewing the above possibilities one by one, it can be found that the overall increase in the number of cases is not the only reason for the increase in the macro-infringement rate, and the laxity of the infringement judgment standard is not in line with China's legislation and judicial practice. After excluding other possible explanations, it can be concluded that the increase in the macro infringement rate in Fig. 2 reflects, to a certain extent, the deterioration of the patent infringement situation in China.
First of all, the overall rise in the number of patent infringement cases is undeniable, but it is not the only reason for the rise in the macro infringement rate. According to the data released by the State Intellectual Property Office, the number of new and concluded patent civil cases in China's courts has increased year by year, and the cases collected in this paper also show a rising trend. At the same time, testing the correlation between the number of patent infringement cases and the macro infringement rate, the correlation coefficient of the two is as high as 0.839, with a P-value of less than 0.001. However, this is not the only reason for the rise in the macro infringement rate. Figure 3 plots the time trend again for the infringement rate of patent infringement lawsuits (i.e., the proportion of cases in which the court ruled on infringement to all patent infringement cases) to exclude the effect of the overall rise in the number of cases on the macro-infringement rate. As can be seen from Figure 3, after excluding the overall rise in the number of cases, the infringement rate still shows an upward trend similar to that of Figure 2, especially in the part from 2015 to 2019, which is even more concentrated and obvious than the upward trend of the macro infringement rate in Figure 2. It can be seen that although the overall rise in the number of cases is an important reason for the rise in the macro infringement rate, it is not the only reason.
Secondly, the infringement judgment standard is gradually relaxed and not in line with China's legislation and judicial practice. As early as 2001, the supreme people's court through judicial interpretations to establish the basic principles of patent infringement judgment - literal infringement principle and the principle of equivalent infringement. Subsequently, the principle of literal infringement has gone through the process of comparing all the technical features from the necessary technical features, and the principle of equivalent infringement has gone through the process of comparing the technical features from the whole to each of them, and the infringement judgment standard tends to be tightened. In addition, in 2008, the Patent Law added the prior art defense, and in 2009, the judicial interpretation clearly stipulated the principle of donation and the principle of estoppel, which further tightened the standard of patent infringement judgment from different perspectives. It can be seen that the gradual relaxation of the infringement judgment standard is not in line with China's legislation and judicial practice, and naturally cannot be a reason to support the rise in the macro-infringement rate.
In summary, since 2014, especially during the period from 2015 to 2019, the macro infringement rate in China has gradually increased, and after reviewing and excluding other possible explanations one by one, we can draw the reasonable conclusion that the patent infringement situation in China has become more and more serious in a specific period of time.
4.1.2 Damages Overview
Figures 4, 5, 6 and 7 are the trend charts describing the three dimensions of China's patent infringement damages over time, of which Figures 4 and 5 correspond to the objective and subjective attributes of the severity of damages, and Figures 6 and 7 correspond to certainty and timeliness respectively. It can be seen that China's patent infringement damages have the following characteristics in the time trend.
First, in terms of severity, the average value of damages shows an obvious upward trend, from 71,371.4 yuan in 2014 to 110,420.4 yuan in 2020, an increase of 54.7%. At the same time, the number of large damages cases with awards of more than 1 million yuan has also increased simultaneously, from 6 cases in 2014 to 42 cases in 2020. Among them, the top three cases in terms of damages awarded were Huawei v. Samsung in Fujian High Court in 2017 (80.5 million yuan), Huitian v. Siliwei in Shenzhen Intermediate Court in 2020 (40.07 million yuan), and Gree v. Ao Sheng in Guangdong High Court in 2019 (40 million yuan).
Figure 4 Time Trend of Average Patent Infringement Damages in China (in quarters)
Figure 5 Temporal trend of the number of large damages cases of patent infringement in China (in quarters)
Secondly, in terms of certainty, the probability of damages has gradually increased over time, while its distribution has become centralized from discrete to maintain at around 80% by 2019 and 2020. Figure 6 portrays the trend graph of the probability of damages (i.e., the proportion of cases in which the court ordered the infringer to pay damages to all patent infringement cases) over time. Taking 2020 as an example, the probability of damages in its first, second, third and fourth quarters were 81.65%, 81.57%, 81.11% and 79.89%, respectively. In other words, four out of every five patent infringement cases that were concluded with a judgment involved damages.
Figure 6 Time Trend of Probability of Patent Infringement Damages in China (in quarters)
Third, in terms of timeliness, the average value of the lag time for damages was relatively stable in 2017 and before, basically remaining in the range of 1.2 to 1.5 years, and began to rise in 2018, peaking in 2020. Figure 7 depicts a graph of the trend in the mean value of damages lag time (i.e., the time elapsed from the occurrence of the infringement to the time when the court ordered the infringer to pay damages) over time. Starting from Figure 7, it can be found that the mean values of damages lag time for the four quarters of 2020, which is the peak, are 1.60 years, 1.64 years, 1.70 years and 1.65 years, respectively.
Figure 7 Temporal trend of our damages lag time averages (in quarters)
4.2 Statistical Inference: Can Raising the Amount of Damages Deter Patent Infringement?
This section intends to estimate the dynamic causal effect between damages and infringement by means of a distributional lag model. Among them, the dependent variable is the macro-infringement rate in a specific period, and the independent variables include not only the average value of damages in the current period, the number of cases with large damages, the probability of damages, and the average value of the lag time of damages, but also the above values with a one-year lag and two-year lag, in order to capture the time lag required for information transfer. At the same time, the case base will be introduced into the model as a control variable to exclude the effect of an overall rise in the number of cases on the macro-infringement rate. Expressing the above model with the formula is:
Macro-infringement rate t = β0 + β1 × mean value of damages t + β2 × mean value of damages (t-1) + β3 × mean value of damages (t-2) + β4 × number of cases with large damages t + β5 × number of cases with large damages (t-1) + β6 × number of cases with large damages (t-2) + β7 × probability of damage t + β8 × probability of damage (t-1) + β9 × probability of damages (t-2) + β10×mean lag time of damages t + β11×mean lag time of damages (t-1) + β12×mean lag time of damages (t-2) + β13×case base + β14×year-end effect + β15×gross domestic product (GDP) + β16×per capita urban and rural residents' disposable income + β17×the share of cases with education level above high school + β 18×Percentage of education above bachelor's degree + εit
Tables 1 and 2 show the results of running the above distributional lag model for all cases and cases where both the original defendant and the defendant are corporations, respectively, and the following observations can be made.
First, high compensation does not ipso facto lead to strong deterrence. In Table 1, the test results of the three variables representing objective severity are not significant (i.e., p-value is less than 0.05); Table 2 restricts the sample scope to cases in which the original defendant is a company, and the above conclusions still hold. It can be seen that changes in objective severity do not have any deterrent effect on patent infringement.
Second, the deterrent effect of certainty and timeliness of damages is obvious. In Table 1, the partial regression coefficients of the variables reflecting certainty and timeliness - probability of damages (current period), probability of damages (two years ago), and the mean of lag time of damages (current period) - are -2.273, -3.095, and -1.646 respectively, which are significant at the level of 0.05 respectively. Table 2 restricts the sample to cases where both the original defendant and the defendant are corporations, and the partial regression coefficients of certainty remain significant. Therefore, although the severity of damages in judicial practice is less than expected, its certainty and timeliness but presents a clear deterrent effect, to a certain extent, verifies the patent infringement damages play a deterrent effect with three-dimensional dimensions of the characteristics.
Finally, the objective change of a certain dimension of damages, need to be converted to the subjective perception of the infringer in order to play a deterrent effect, in this information transfer process, there may be errors and delays. In the case of severity, for example, changes in the average value of damages, a variable representing its objective attributes, do not have any deterrent effect on infringements, while the number of large compensation cases (two years ago), a variable representing subjective severity, shows some deterrent effect. At the same time, the deterrent effect of the number of large compensation cases is not timely, and there is a delay of about two years. The above data also confirms the expectation of the theory construction part of this paper: at least in terms of severity, the deterrence of infringement by damages has the characteristics of mediating conditions - potential infringers are limited by their own knowledge, ability and access to information, and their subjective perception of damages may be one-sided and lagging.
