Determinants of Patent Appeal outcomes - A study of Swedish firms

Determinants of Patent Appeal outcomes A study of Swedish firms

Master’s Thesis 30 credits

Department of Business Studies Uppsala University

Spring Semester of 2020

Date of Submission: 2020-06-03

Submitted by Bhavna Dadhich

Supervisor: David Andersson

Abstract

The number of appeal cases is increasing every year which becomes a problem in Sweden as the time duration of an appeal is longer than in other jurisdictions. Therefore, the aim of this study is to assess whether in the future, a firm that has certain characteristics will win or lose an appeal case. These characteristics will frame a strategy to protect their intangible assets from being opposed and appealed, which ​hopefully will save both time and costs for the firm. The study of appeal data shows that 41.54 % of the appeals are made by inter partes firms in Sweden and most of the firms who appeal are from the industrial goods sector. It also shows that out of total applications filed at the court of appeal 65.87% of applications are being filed and won by the opposition. To predict the outcome, a multiple logistic regression model is used on the variables such as total assets, sales, profits, intangible assets firm size, age. The findings show that net sales, firm size (number of employees), net profit and the differences in the age of the firm are significant for the outcome. This implies that firms can evaluate the possible outcome beforehand based on firm characteristics.

Keywords: patents, intellectual property rights, plaintiff, counterparty, oppositions, appeals.

Table of content

1. Introduction 3

2. Literature Review 7

2.1 Patents in the Swedish market 7

2.1.1 The Patent and Market court 9

2.2 Patent appeals and opposition in the Swedish market 10

2.2.1 The problem with duration of time 14

2.3 Protecting patents in the Swedish market 14

3. A statistical model 19

Multiple logistic Regression 19

4. Data 22

4.1 Data sources 22

4.2 Sample Selection 23

4.3 Data Description 25

4.4 Variables 27

The following section describes the variables from the data sources. 27

4.4.1. Dependent Variable 27

4.4.2. Independent Variables 27

5. Empirical specification and estimated results 31

5.1 Descriptive statistics 31

5.2. Multiple Logistic Regression Analysis 35

6. Discussion 39

7. Conclusion 42

7.1. Implications and contribution of the study 42

7.2. Limitations and further research 43

Appendix 49

1. Introduction

Technological development and innovation is the key driver for the economic growth of a country. Innovation brings benefits to society and it has greater importance when the society holds enormous natural resources and human knowledge, but disruptive effects ​could threaten to reap these benefits (Broughel and Thierer, 2019). ​The protection of Intellectual property (IP) is crucial in a globalized economy as it helps businesses to grow and innovate (Gould and Gruben, 1997). IP falls in the category of intangible assets such as patents, trademarks, designs, copyright, geographical indications and trade secrets which are all creations of human intellect.

Webster (2017) explains that, for many companies, intellectual property is a sizeable asset and holds approximately 70% of the company’s value. Intellectual property rights assist the inventor or firms to receive proper value for their patents, trademarks, designs, copyrights, and utility model rights. A patent is considered as a stimulating incentive for the inventor and is associated with the financial reward, as a result of the exclusive rights given to the patent owner. For the exclusive rights given to the owner, the owner contributes with the new technical knowledge which may stimulate further technical development in the society (Oredsson, 1988). The size of the firm, the competitive environment, and the nature of the technology are found to critically impact the patent rights of firms (Galasso and Schankerman, 2018).

Every year, there are several appeals filed in court. Appeals cause the patent holder to put substantial effort into defending the case for its validity, or infringement which involves higher cost and time, and in turn adversely impacts innovation, research, and development activities (Harhoff and Reitzig., 2004). This study examines appeals that are the result of opposition and litigation regarding protecting patents and it aims to devise a strategy for the firms to protect their own patents from being opposed or having to appeal. This is especially important in Sweden as there is an existing problem with the time duration and higher cost in trying a case.

Therefore, it is crucial for those who are opposed or are applying for a patent in Sweden to know

what they are up against in order to protect and maintain their patents because patents provide

vital advantages to being competitive in a global economy.

Intellectual property rights (IPR) plays an important role in fostering innovation by protecting the rights of the invento ​r. A patent is a form of intellectual property and the exclusive right of a patent is given to the inventor in the statutory law. It is a privilege giving the inventor, the sole right to make use of its invention for a limited number of years (Oredsson, 1988). The right for an invention arises if a patent is granted and registered (Borgenhäll and Waerme, 2018) prohibiting others from using the protected invention, and giving the right to the owner, also called patentee, to take legal action against infringements. The patents endow the patent holders with a passive right and the holder of the patent initiates enough efforts to defend his patent against either a validity challenge or against infringement of the patent right (Harhoff and Reitzig., 2004). On the other hand, anyone, except the patentee, who wishes to remove the exclusive right of a patent may file an application to the patent office within a specified period after the patent is granted or published with clearly stating the grounds for the opposition. An applicant whose patent application has been rejected or an unsuccessful party in an opposition case can appeal against the decision to the Patent and Market Court of Appeal. (PRV, 2020).

In Sweden, the Patent and Market Court of Appeal is the highest appellate court. There are special provisions for the intellectual property proceedings for the patent regarding the appellate procedure in patent proceedings in Sweden (PRV, 2020). According to the Swedish research council, Sweden was also a part of a knowledge-based community to help in increasing European competitiveness. Despite having laws and strong efforts towards regulations for the protection of intellectual property, several infringement cases turned into oppositions and then appeals. These cases have been registered in almost all the sectors in Sweden and across Sweden.

There are numerous factors that can affect the appeals. Multiple factors such as firm size, type,

experience, industry, the patent system, internationalization moderates the patent litigation

strategy and affects the firm (Yang., 2019). Many studies have been done to study opposition and

appeals in European countries however, there is a gap in the same research being conducted

amongst Swedish firms. Therefore, this study aims to acknowledge the gap. This paper will

focus on the patent opposition and appeals cases of Swedish firms and explore the firm

characteristics and the impact of those characteristics on the appeal outcome. The specific research question this study addresses is;

How do firms’ characteristics affect the outcome of patent appeals in Swedish firms?

For this study, patent appeals data is collected from the Court of Patent Appeal between the years 1998 and 2016. The data includes 126 applicant firms and 126 counterparty firms , altogether 252 firms’ data were analyzed. The other two data sources used in the study are collected from PATlink and Serrano data to extract organization and financial information respectively.

Multiple logistic regression models were used to predict the outcome of appeals based on certain characteristics of the firm. It is computed on the differences of the corresponding values there are on a logit model. The binary (0/1) variable which is to be predicted is the dependent variable, in this case, the outcome. The various firms’ characteristics total assets, intangible assets, cash and bank, age of the firm, operating profit, net profit, sales, firm’s size and sector are independent variables for the study. ​The date of the application for the appeal (PRV) and the date on which the decision is given (PBR) is considered for calculating the time to the settlement of the case.

Sector data was analyzed to study which sector had the most appeals. The paper studies the various determinants of opposition, which is the root cause for appeals, and the patent system in Sweden for the opposition and appeals.

The study found that ​most of the firms that appeal are from the industrial goods sector followed by the corporate sector. The average time in taking the decision by the court comes to 32 to 33 months. It shows that out of total applications filed at the court of appeal 57 (34.13%) appeals were made by the patentee and 65.87% of applications are being filed and won by the opposition.

The regression model is computed and the results show that winning firms have higher net sales

and bigger firm size. This suggests that these two variables are related to the prediction of the

outcome of winning or losing patent appeal cases among Swedish firms.

The remainder of this paper is structured into seven sections as follows. ​Section II discusses the

literature review and patent system in Sweden, ​Section III presents a statistical- multiple logistic

regression model and ​Section IV ​discusses data and the computation of key variables. ​Section V

discusses the descriptive statistics and multiple logistic regression equation for the firms in the

sample and its results. Finally, ​Section VI and VII ​present the discussion and conclusions of the

study respectively.

2. Literature Review

2.1 Patents in the Swedish market

Sweden ranks fifth in the world behind the USA, China, South Korea, and Japan in terms of

patent applications (Rosenfeld, 2019). In Industry 4.0, which is called the Fourth industrial

revolution, Sweden filed the most patent applications out of all the European countries according

to the study by the Swedish Patent and Registration Office (PRV). Industry 4.0 relates to nuclear

technology and enabling technology and cloud computing amongst others. Between 2010 and

2016, a total of 2600 applications were filed in this field, amongst which one-fourth of all

European applications were filed by Sweden (Rosenfeld, 2019). This suggests patents are an

important intellectual property right in Sweden and motivates research in this area. A patent is a

form of intellectual property and the exclusive right of a patent is given to the inventor in the

statutory law. The inventor who is granted the patent is entitled to the rights for a limited number

of years that restrict others to use the patent. To maintain the patent for a limited number of

years, the inventor is obliged to pay the yearly fees for the same. The exclusive right includes the

right to assign, license, or pledge the right to a patent. In order for a patent to be granted, an

applicant must fulfill certain requirements. The patent holder must assure and disclose that the

invention is new, has an industrial application, and patent registration under the civil law protects

the property from infringement (PRV, 2020). The exclusive right for an invention arises if a

patent is granted and registered (Borgenhäll and Waerme, 2018) prohibiting others from using

the protected invention and giving the right to the owner also called patentee to take legal action

against infringements. The patents bestow the patent holders with a passive right and the holder

of the patent initiates enough efforts to defend his patent against either a validity challenge or

against infringement of the patent right (Harhoff and Reitzig., 2004). ​Every year thousands of

applications are being filed at the Swedish patent office that shows more and more firms are

filing for the patents and safeguarding their invention. Figure 1 shows a total of 370,658 direct

applications (total count by applicant origin) were filed in Sweden and granted 189,417 during the period between 1980 to 2016.

Figure 1: ​Swedish applications and patents grants 1998-2016. Source: WIPO, 2018.

A jurisdictional guide to Intellectual property (IP) litigation (2018) that discusses IP rights infringement argues that effective enforcement of IP rights against infringements and validation promotes the corporations to conduct the business in multiple jurisdictions. Many patents are registered in other jurisdictions and obtaining patent protection in other than their own national jurisdiction is typically expensive (Harhoff et. al., 2003). According to Putnam (1991), normally the applicants will seek international patent protection only for their most valuable inventions.

An application can be made in Europe or internationally. Swedish applicants can choose the

place as Sweden as the protected country then the application is moved to the national regulatory

authority for the required steps to proceed further with the application process. These patents can

also be granted at European Patent Office (EPO) as it provides a single patent grant procedure,

but not a single patent from the point of view of enforcement (Gowers, 2006). ​When a European

patent is granted and it's validity is not challenged by the opposition, it becomes a bundle of national patents or rights in those states (Jerak and Wagner., 2006; Harhoff and Reitzig, 2004).

Between 1978 and 1999, EPO has become one of the most important patent offices in the world after granting 479,133 patents out of the total 1,267,681 patent applications. Although, the granting period at that time was 4.2 years from the date of filing the application (Harhoff and Reitzig, 2004). After the new court was established in Sweden in 2016, Sweden is expected to be the most efficient jurisdiction in Europe (Ficks, 2017). On the contrary, if more patents are granted with negligence then the patents are challenged for its validity. So it is important for the policy makers to closely monitor the patenting activities.

2.1.1 The Patent and Market court

One who files an opposition must present the proper evidence that the granted patent fails to fulfill the requirement for the patentability; i.e., that the invention lacks novelty or the disclosure was poor or inadequate (Hall et al., 2004). Any third party can submit an objection in writing to PRV during the application process if it thinks that it can affect him/her if a patent is granted.

When PRV assesses the patent application, it considers all the issues raised during the application procedure. The party who files the opposition will get the notification if the patent is granted or if the objection is found significant, the notification will be sent to the patent applicant with an opportunity to comment. Thereafter, the opposition party has nine months to file a written opposition against the granted patent. After exchanging a few writs between the patentee and the opponent, PRO comes to the conclusion with three outcomes; either the patent will be revoked or amended or rejected. The losing party, whether the applicant or the objector, can appeal to the Patent and Market Court for the final decision within two months from the date of the decision. (Borgenhäll and Waerme, 2018).

An unsuccessful applicant or opponent should appeal to the national Patent and Market Court

(Patent- och marknadsdomstolen) within two months of the decision and a copy is sent to PRV

explaining the incorrect decision taken by PRV. The appealed application procedure is reviewed

by the PRV and can be altered if they agree that it was not correct. If there is no change in the previous decision, PRV refers the case from their side to the National Patent and Market Court (Patent- och marknadsdomstolen). If the appeal was made within the time limit, opposition cases are handled directly by the Patent and Market Court without prior review. The procedure is the same for infringement cases, it can lead to a legal proceeding against the infringer and disputes are settled in the national Patent and Market Court (PRV, 2020).

An appeal for annulment is the same as opposition. A patent can be declared invalid even after the opposition period has expired and if the opponent initiated the proceeding before the Patent and Market Court (Patent- och marknadsdomstolen), a division of the District Court of Stockholm. The Opposition party should also notify the PRV and others who have the license or undertaking for the patent (PRV, 2020).

The National Patent and Market Court (Patent- och marknadsdomstolen, PMD) which is a part of the District Court of Stockholm, tries to resolve the decisions taken by the PRV. The case, if not resolved by the PMD's, is appealed to the national Patent and Market Court of Appeal, a division of Svea Court of Appeal. Before 2016, appeals and oppositions were handled by Patentbesvärsrätten 1978-2016. After the case is evaluated and a need for further investigation is found, the court may give permission to appeal to the highest judicial body, the Supreme Court after contributing to the reinforcement of legal security. There are special provisions for the intellectual property proceedings for the patent regarding the appellate procedure in patent proceedings in Sweden. Besides the procedural provisions, it should be noted that the right to appeal is limited in patent proceedings (as regards appeal to the Supreme Court). In practice, for most of the patent cases, the Patent and Market Court of Appeal is the highest appellate court (PRV, 2020).

2.2 Patent appeals and opposition in the Swedish market

Sweden’s patent and registration office (PRV) in Sweden, is the authority for intellectual

property and it constantly works with cutting-edge ideas to strengthen the value of business and

competitiveness. According to the Swedish research council, Sweden was also a part of a knowledge-based community to help in increasing European competitiveness. Despite having laws and strong efforts of the regulatory for the protection of intellectual property, a number of infringement cases turned into oppositions and then appeals, and have been registered in almost all the sectors in Sweden and across Sweden.

Studies in Europe for the period 2000-2008 on patents and trademarks opposition found that Europeans use opposition more frequently than other countries and nearly 88 % of European patents are opposed by European opponents itself (Caviggioli et al., 2013). This is true in the case of Sweden also, where a higher proportion of opposition is seen from domestic defendants (Bjuggren et al., 2015). On the other hand, it was claimed by Bjuggren (2015) that Swedish patent opposition shrunk during the period in comparison to other European countries. A study during the period 2000-2008 argued that Germany had the largest number of cases in absolute terms among five other European countries whereas Sweden has registered the least, only 2.5%

of total cases registered in the five countries (Bjuggren et al., 2015). Studies during the years 1995 to 1999 showed that 235,396 patents were granted at the EPO and 14,087 of these patents (5.98%) were opposed (Harhoff & Reitzig, 2004) which was slightly lower compared to 5.3%

during the years 2000-2008 (Caviglioli et. al., 2013). Lanjouw and Schankerman (2002), Somaya (2002) and Allison et al (2003) found that the absolute amount of litigation had increased tremendously during the period 1978-1995 (Hall et al.,2004). According to WIPO (2018), the largest number of oppositions are in medical technology followed by pharmaceuticals. Earlier literature pointed out that the opposition rate in biotechnology is significantly higher than other emerging sectors and parts of large portfolios are more likely to be faced with more legal attacks and opposition (Schneider, 2010; Jerak and Wagner, 2006). This is supported by other authors ​in their article that the number of opposition in semiconductors, computing and software sectors is lesser than in the biotechnology or pharmaceutical sector (Graham et al., 2002).

Several authors have studied the determinants of opposition, they are important for the present

study as they are key factors in appeals and they influence the outcome of the appeals. A

comparative analysis on the post grant challenges, faced in biotechnology and pharmaceuticals,

semiconductors and computer software were analyzed between 1980 and 1997 and found that valuable patents are more likely to be challenged and the rate of opposition was 30 times higher at European patent offices (Graham et al., 2002). Econometric studies on biotechnology and pharmaceuticals found a positive relationship between the patent value and the probability of opposition (Harhoff et al.,2003; Harhoff and Reitzig, 2004). Lanjouw and Schankerman (1999), compared the number of claims with a measure of a patent’s breadth and found a high correlation between a number of claims with the value of a patent (Jerak and Wagner, 2006). Patent value is considered as a powerful tool for a successful defense against the opposition (Harhoff, 2003).

The patent of a higher value is more likely to be attacked and a successful defense against opposition and annulment indicates the strong value of a patent (Harhoff and Reitzig, 2004;

Harhoff et al., 2003). Other than the patent value, the size of the firms can be the reason for the opposition. It has been seen that leading companies’ patents are opposed more often than on average. For example, leading companies like Procter and Gamble, Unilever, ICI, Union Carbide and BASF face 10% higher opposition rates (Harhoff et al., 2003). On the contrary, according to Cooterand Rubinfeld, (1989) and Lerner, (1995), larger and settled firms are more likely to attack small firms because larger firms have more resources, knowledge and experience to defend their intellectual property.

Many authors have criticized the concept of opposition with the patent value. It was found that

the higher oppositions are not only related to the value of the patent but large-volume applicants

are more likely to be opposed. The large volume of applications was due to the strengthening of

the patent system which was lacking in the sufficient review of patent applications and then

granted the patent with insufficient quality (Hall et al.,2004; Caviglioli et. al., 2013). Due to the

shortage of available resources to the patent office, there was a fall in the number of granted

patents and the patents that were granted faced post validity challenges (Hall et. al., 2004). This

shows that if the patents are granted inefficiently, they will end up in opposition and then to

appeals. Another aspect of studying an opposition is that a patent that has faced several

oppositions and with more expensive procedures is considered more valuable than a patent that

has not faced any opposition previously (Harhoff et. al., 2003). In other studies, it was

demonstrated that an increase in patent litigation was strongly supported by increases in firm

values for the number of patents per dollar of R&D spending, capital expenditures, total R&D spending, market value, scale, liquidity level, and patent portfolio quality. These features show that the firm and the patent characteristics are the driving force for the opposition (Eskandarikhoee, 2015).

The party that loses the opposition may appeal their cases to the higher court. The determinants

of patent appeals and the outcome of the appeals are influenced by various components. It can

be said that factors that have influenced the appeals, consequently affected the firms, depending

on what kind of stakeholders they are: as a disputant (plaintiff, defendant); as to their size

(individuals, SMEs, and large enterprises) and in what kind of industrial setting they are in

( ​Yang​, 2019). In an article, court outcomes were studied and found significant disparities in the

outcome for the different industries. The analysis reveals the significant relationship between the

various variables and litigation outcomes (Allison et. al., 2016). The patentees that have a higher

quality patent portfolio are more likely to win (Eskandarikhoee, 2015). A similar study on the

outcomes on patent challenges found out that branded firms that had peak sales, won a majority

of court decisions on AI patents (Grabowski et.al.,2017). Studies on the determinants of patent

oppositions and opposition outcomes in the European market found that the probability of a

patent to be opposed by industry rivals is positively associated with an increase in company sales

and the number of citations received (Sterlacchini, 2016). Studies also show that the outcome of

the appeal is a result of the strategies pursued by the firms (Somaya, 2001). Appeal outcome is

not just influenced by the firm characteristics or the strategies framed by them but is varied by

jurisdictions too. Mejer and van Pottelsberghe (2012), found substantial differences in litigation

outcomes on the same patent across European jurisdictions. The outcome of these ligitations

were studied and found a measurable effect on the firm which depended on both characteristics

of the firm and the outcome of the trial. Defendants or the losing firm are negatively affected by

a loss or a settlement, while a win leaves the value unchanged (Cremers et al., 2013).

2.2.1 The problem with duration of time

The judgment on the appeals is a matter of concern, as the length of time for the judgement may vary for different jurisdictions. From a Swedish perspective, the major concern seems to be the time duration for the cases, which is comparatively longer in Sweden (Bjuggren, Domeij and Horn, 2015). Sweden has suffered relatively more time consuming judicial proceedings in comparison to other major patent jurisdictions in Europe such as Germany, the Netherlands, the UK and France. Although a speedier alternative known as an interim or interlocutory decision is available in Sweden which takes on average 3 months for the judgement (Ficks, 2017). No such speedy judgement is available in other European jurisdictions. According to Graham et al.

(2003), the average length of time taken for the appeals and opposition comes to three years at EPO although variation was found in the length for the individual cases. The normal time frame from filing a lawsuit to reaching the final judgment is 35.5 months and 37.5 months for revocation, which is longer in Sweden than other European countries (Bjuggren et al., 2015).

Efforts have been made to improve the quality of the patent and to reduce the length of the patent litigation (Hall et al., 2004; Bjuggren et al., 2015). The patents that have more opposition with more citations, more European coverage or multiple oppositions, take a longer time in judicial proceedings (Graham et al., 2002). Apart from time, the cost of filing a case or litigation suit is also considered as an important factor in appeals.. Each party must carry its own costs and these costs cannot be charged to the loser (Graevenitz, 2008). Whereas in Sweden, the losing party reimburses the cost to the prevailing party and in a few circumstances, litigation cost is shared by both parties (Westerberg et. al., 2018).

2.3 Protecting patents in the Swedish market

Intellectual property is considered to be of financial value and the aim is to protect and

strengthen the value of an asset. Along with that, IP protection helps the creator or inventor to

earn recognition or get financial benefits out of it (WIPO). It is observed that IPR not only

enables the inventors, creators, and businesses to prevent unauthorized exploitation of their

creations but also helps them to receive compensation on their investment (Taplin and Nowak,

2013). In Sweden, inventor creation is protected and gives them the right to exploit the invention through patent rights. The right to use the invention arises through applying for the patent registration and being granted. Sweden’s creativity, innovation and digitization are at a level of global competence and to continue with the same knowledge economy, the government has initiated and worked together with the Swedish Patent and Registration Office (PRV) to strengthen intellectual property rights (IPR). PRV experts in intellectual property collect not only the technical, but also legal and business information from all around to help in taking strategic decisions. Along with that, Sweden possesses a Patent Cooperation Treaty (PCT), an international agreement that allows one to file a single case in a country's language and get an international filing date for the case.

Patents are protected when they are granted by the authority. There are three ways to make the application, it can be made (i) nationally, (ii) throughout Europe pursuant to the European Patent Convention (EPC), or (iii) internationally pursuant to the Patent Cooperation Treaty (PCT).

When an application is made to Europe or internationally, an applicant can choose the place as Sweden as the protected country. Then the application is moved to the national regulatory authority for the required steps to proceed further with the application process. The application authority in Sweden is the Swedish Patent and Registration Office (PRO) works with both national and foreign applications. In all the three ways of the application, after formal and technical examination, the patent is granted in a national Swedish patent registration. In the case of rejection of an application, the applicant can appeal to the Swedish Patent and Market Court.

The time limits for granting a patent vary from case to case but generally, the final decision is made within two years from the date of filing an application. The patent remains in force for 20 years for the application date but for a few sectors, it can be extended to a maximum of 5 years.

The patentee has to pay yearly fees to retain the patents (Borgenhäll and Waerme, 2018). The

cost for the national applications and patents is 3000 SEK which can be paid through different

payment methods (PRV, 2020). It includes fees for government services, copies and printouts of

public documents. The total cost of the patent at the EPO comes to roughly three times more than

applying for a national application but it is less expensive than if it is applied for in more than

three designated states. This makes it appealing for the applicant to promote business in multiple markets of the European countries (Hall et al.,2004).

Hence, as previous literature has established the importance of firm characteristics in patent opposition and appeal cases, this paper examines the relationship of said characteristics with outcomes of appeal cases concerning Swedish firms. Specifically, firms’ size, intangible assets, total assets, firms’ age, sales, operating profit, net profit/loss, sector and cash and bank are analysed in predicting the outcome of appeals on both the applicant and counterparty firms. The value of an intangible asset, total assets, sales, profits and cash and bank that appear in the balance sheet of a firm played an important role and therefore, have been included in this study.

Intangible assets that include research and development, patents, goodwill and other intangible assets are of greater importance. They are considered for the study because it is expected that all future economic benefits that are associated with these intangible assets will flow to the firm based on the indefinite and finite life of an asset (IFRS, 2017). On the other hand, cash and bank balances show how much money is available to the firm immediately for spending. Investment in intangible assets and cash availability may vary depending on the type, size and age of the firm.

Sales reflect the revenue and profit that show the financial health of the firm. In addition, the

sector is included to study the relationship with predicting the outcome and also to investigate

which sectors in Sweden have the most appeals cases. The age of the firm, size of the firm and

time for the settlement are also considered. To investigate the research question, these variables

are studied with the help of the multiple logistic regression model. The results from this study

will help to answer the research question of how these variables are helpful in predicting the

outcome of appeal cases of Swedish firms.

Table 1: Summary of Literature Review

Authors Main Findings

Patents in the Swedish market

Rosenfeld (2019) Swedish ranking in Europe and in the world.

Borgenhäll and Waerme (2018);

PRV (2020); Harhoff and Reitzig (2004)

Exclusive and passive rights for the patent holders.

Harhoff et. al. (2003) To conduct business in multiple jurisdictions.

Jerak and Wagner (2006); Harhoff and Reitzig (2004)

Bundle of national patents or rights in the states.

Patent appeals and opposition in the Swedish market

Caviglioli et. al. (2013) European countries use opposition more frequently than any other country.

Graham et al (2002) Valuable patents are more likely to be

challenged.

Harhoff et. al. (2003); Harhoff and Reitzig (2004);

A positive relationship between patent value and opposition.

Hall et al. (2004); Caviglioli et. al. (2013) Lack in reviewing and insufficiently granting the patents.

Allison et. al. (2016) Significant relationship between various variables and outcomes.

Grabowski et.al. (2017); Sterlacchini (2016) Opposition and appeal outcomes are

associated with the company’s sales.

Somaya (2001) The outcome of the appeal is a result of the strategies pursued by the firms.

Bjuggren, Domeij and Horn (2015);

Ficks (2017)

Relatively more time-consuming judicial proceedings in comparison to other major patent jurisdictions.

Protecting patents in the Swedish market

PRV (2020) Working towards strengthening intellectual

property rights.

Taplin and Nowak (2013) Prevent unauthorized exploitation of creations of the inventor.

3. A statistical model

Multiple logistic Regression

In order to predict ​the outcome of the appeals and answer the research question in a systematic manner, a statistical- multiple logistic regression model is used. ​Logistic regression models have been used across a range of studies involving patent characteristics in the study of patent litigation and at least a few studies have used it in predicting win/loss ratios. This provides information about the hierarchical nature of patent case outcomes and the interaction of variables that influence the outcome (Cowart et.al., 2014). In appeal cases, the two firms, the applicant and the counterparty ​may have different outcomes that depend on different factors. Here, to predict the outcome, various firms’ characteristics are used for the study. First, descriptive statistics of the data is constructed and then the multiple logistic regression model is used on the dataset.

Multiple logistic regression is an extension of logistic regression and is used when there is one nominal variable and two or more measurement variables. It is used to find out how the measurement variables affect the nominal variable, also known as the dependent variable. The outcome of the dependent variable is dependent on the independent variables. The binary variable which is to be predicted is called the dependent variable, in this case it is the outcome.

The outcome in logistic regression analysis is coded as 1 or 0, where 1 indicates that the outcome of an appeal is winning and 0 indicates that the outcome of an appeal is losing. Firstly, the data has been modeled for future predictions and it describes whether in the future a firm that has certain characteristics will win or lose the appeal.

Two separate logistic regression models are run on the data. The first ​multiple logistic regression

is computed on the differences between the independent variables to check that certain properties

of the model like whether the independent variables are significant or not. The model related to

this is specified in equation 1 below.

equation 1 Where,

D ​

= denotes the differences of applicant firm -intangible assets and counterparty firm - intangible assets

D ​

= ​…

Similarly D ​

, D ​

… , D ​

are defined as above. This gives us a set of dependent and independent variables where the 1st variable is the outcome that has values of either 0 or 1 and other variables are the observed differences of the corresponding variables.

The second regression model is based on the independent variables relating to the characteristics of the firms as well as the differences that have been explained above. It is done to determine which characteristics influence or have a relationship with the probability of winning or losing a patent appeal case. The specification for this is shown in equation 2 below.

equation 2

X ​

, X ​

X ​

X ​

denotes the independent variables of the firm.

P(y=1) and P(y=0) is the expected probability that the outcome is winning, X ​

through X ​

are

distinct independent variables, and through are the regression coefficients. (MLRA,2013).

For the second regression, ​ stepwise regression is computed on the independent variables and on the differences between the independent ​ variables to determine the impact of firm characteristics on winning and losing. Stepwise selection is a combination of both forward and backward selection where it starts with no predictors and then continuously adds the most contributive predictors (like in forward selection) and then after adding each new variable or predictor, removes the variables that do not contribute to the improvement in the model or is no longer statistically significant (like in backward selection). So if the older variable's coefficient is not significant then that variable is removed and a next variable is placed into the model and so forth.

It is expected that this model will be helpful in predicting the outcome i.e. winning and losing.

4. Data

4.1 Data sources

To study the opposition cases in patents, three different data sources within Sweden were required for the study. To gather the information on all the appeal cases, data was collected from the court of Patent Appeal (PBR), the second source was PAtLink and the third source was Serrano data to extract the financial information of the firms that filed the opposition.

Patent Appeals Data

The Court of Patent Appeal is an administrative court that files all the appeals, reviews decisions taken by the patent and registration office; makes decisions in relation to trademarks, patents, design; and makes authorizations to publish periodicals (PBR, 2020). The Swedish court of patent appeals (PBR) (värsrätten) database contains appeals made on Swedish patents, trademark and designs filed between 1978 and 2016. The complete data set contains 21,641 opposition cases including 9,849 for patents, 9,828 for trademarks and 1,963 for designs. The data includes information on the type of property rights, the application number, year, plaintiff, counterparty, addresses, agents, process dates, duration and decisions of the opposition.

PAtLink

The second data source was PAtLink which is referred to as the Swedish market for patents and

it contains all the information on organization numbers of Swedish firms that have registered

patents. This source is mainly used to match the organization number with the appeal cases

number given in the PBR. This helps to identify the unique number and names of the firms in

Sweden. The application number given in the PBR data was matched with the PAtLink to find

out the organization number. PAtLink databank contains all the information such as application

id, application number, unique id, family id, application year, kind of application, IPR type,

organization number (Swedish house of finances, 2020).

Serrano database

The third data source was the Serrano database, which is considered as the first and the most prominent database with financial history (Swedish house of finances. 2020). It contains all the financial data of Swedish firms based on the financial statement of all the Swedish companies that are registered at the Swedish Companies Registration Office (Bolagsverket). The Serrano database contains financial information from 1990 and covers most legal firms in the Swedish business community. It contains 10,800,286 observations. The organization numbers identified in the PAtLink were used in the Serrano data to extract the financial information of the companies.

4.2 Sample Selection

Swedish court of patent appeals (PBR) (värsrätten)

The patent appeal data is collected from the (PBR) (värsrätten) database that contains all the appeals made on Swedish patents. The complete data set contains 21,641 appeal cases including 9,849 for patents and others for trademarks and designs. There were 1,278 patent appeals for the study period which included all the appeals filed to Swedish patent and registration office (PRV), by the foreign firms and by the domestic firms. ​The appeals made to (PRV) are known as ex-parte, an appeal where there is only one party. There were 876 appeals made to PRV during the period of study. When two parties are involved in the appeal, it is called as inter partes where who filed the appeal is known as the plaintiff, and the opposition party is known as the counterparty. In inter partes appeals, the firms can be either foreign or Swedish, or both of them can be from Sweden. The focus of the study is to investigate inter partes appeals where both plaintiff and counterparty were from Sweden. The purpose of choosing inter partes is the availability of financial data of Swedish firms. It was not feasible to extract the financial information of the firms outside Sweden within a short time.

The appeal cases were sorted between 1998-2016 and 167 firms were found that include the

information about the case number, application date, decision dates, information of plaintiff,

counterparty, their agents, and organization names. Out of the total inter partes appeals (foreign and domestic) made at the court, 167 (41.54%) appeal was related to Swedish firms. Earlier studies have shown that in Sweden, domestic patents account for 41% of the appeal patents (Bjuggren et al., 2015). After analyzing 167 firms, it was found that 57 (34.13%) appeals were made by the patentee (inventor/ owner) and 110 (65.87%) applications were filed by the other than the patentee, which reflects that patents are mostly opposed and appealed by the counterparty. The possible reason for this could be that the invention lacks novelty, or the disclosure was poor or inadequate for the granted patents.

Out of 167 inter partes appeals, 41 appeal cases were dropped due to the rejection of the applications and pending decisions. The application is rejected if it does not fulfill the criteria set by the technical administrator in consultation or by the experienced patent expe ​rt. As ​far as pending decision cases are concerned, a few cases were dropped from the study. There were 126 inter partes appeal cases ​for the study.

Table 2: Data collection process for the appeals

Data selection Number of Appeals

Total appeals 1,278

PRV 876

Inter partes appeals (foreign & Sweden) 402

Inter partes appeal (only Sweden) 167

Rejected and pending decisions 41

Inter partes appeals (Sweden) 126

PAtLink

The next information is collected from PAtLink. The application number given in the PBR data

for the 126 appeals were matched with the PAtLink to find out the organization number for both

applicant and counterparty firms and in total 252 organization numbers were identified. To

collect the information for a few applications, Swedish patent office databases are used. The data was searched using the case number and application number given in the PBR appeals.

Financial data

After collecting the organization numbers of all 252 firms from PAtLink, these numbers are matched to those in the Serrano database to extract the financial information for all the firms.

The financial information includes all the financial data of the year in which the application was made for the appeal. It includes all the data related to financial statements and other relevant firm data such as firm type, firm size, date of registration and so on. The financial data related to the firms was kept for further analysis.

Table 3: Financial data collection required

Firms Number of firms

Applicant 126

Counterparty 126

Total financial data 252

4.3 Data Description

In this study, there are applicants who file the case and the corresponding party is the counterparty firm. Cases are paired on a case to case basis such that if the applicant firm wins or loses the case, then the counterparty is the opposing firm, that means that if the applicant firm wins then the counterparty loses and if applicant loses then the counterparty wins the case.

126 pairs were put together from this information and the structure is such that the first group

(group 1) is the applicant firm and the second (group 2) is the corresponding firm in the case,

which is automatically the opposing firm. The dependent variable is the outcome variable which

is binary in nature, meaning that if a firm within the sample wins the appeal case, this variable

has the value of 1. Similarly, if the firm loses, this outcome variable has the value of 0.

The independent variables are denoted by X ​

, X ​

X ​

X ​

where

X ​

= intangible assets X ​

= Operating profit/loss X ​

= Net profit/loss X ​

= Total assets X ​

= Net sales X ​

= Cash & bank X ​

= Sector X ​

= firm size X ​

= Age of firm

Next, the following differences are calculated on the independent variables to check whether the outcome is dependent on the difference or not, and that is denoted by “D”

D ​

= difference of ​ applicant firm - intangible assets and counterparty firm - intangible as​sets D ​

= difference of ​applicant firm - operating profit/loss and ​counterparty firm - operating profit/loss

D ​

= difference of ​applicant firm - ​Net profit/loss and ​counterparty firm -​ Net profit/loss D ​

= difference of ​applicant firm - ​total assets and ​counterparty firm - ​total assets D ​

= difference of ​applicant firm - ​ net sales and ​counterparty firm - ​net sales

D ​

= difference of ​applicant firm - ​ cash & bank and ​counterparty firm - ​cash & bank D ​

= difference of ​applicant firm -​ sector and ​counterparty firm​ - sector

D ​

= difference of ​applicant firm -​ firm size and ​counterparty firm -​ firm size.

D ​

= difference of ​applicant firm - ​age of the firm and ​counterparty firm-​age of firm

Hence, there are 10 variables, the first variable (dependent) is the outcome that has values either 0 or 1 and the other D ​

to D ​

are the observed differences of the corresponding variables. The ​‘R’

programming tool was used for the computation of variables.

4.4 Variables

The following section describes the variables from the data sources.

Applicant & counterparty- Plaintiff is a firm or a person that claims in court that a law was broken by another firm or person as a result of doing something wrong or illegal. The plaintiff in this study is referred to as the applicant and can either be an inventor or the other party who has been affected by the granted patent. The opposite party is the counterparty. In this study both the applicant and the counterparty are Swedish firms.

4.4.1. Dependent Variable

Court Outcome- ​The outcome of the appeal was either winning or losing, or the application was rejected, or any other reasons. This variable has been used for the cases where the decision for the firm was winning or losing. A binary variable was created to represent a win (1) and loss (0) outcome in the appeal cases.

4.4.2. Independent Variables

The following are the independent variables used in the study:

PRV and PBR decision dates- The date of application for the appeal is the PRV decision date and the date on which the decision is given is the PBR decision date. The difference between these has been considered to calculate the time to settlement.

Sector - This variable is used to identify in which sector firms are more likely to win appeals.

The branch sector grouped the branches into eleven overall sectors. They are Energy &

Environment, Material, Industrial goods, Construction industry, Shopping goods, Convenience

goods, Health & Education, Finance & Real estate, IT & Electronics, Telecom & Media,

Corporate Services and Others. This a discrete variable.

Firms age - The registration date of the firm has been used to calculate the age of the firm. This has been considered to identify if the outcome of an appeal is affected by the firm’s age. I hypothesise that as firms grow older and gain more experience, they are more likely to make an appeal if they believe that they have a higher chance of winning. Making such informed decisions comes from experience and observations in their industry. Hence, I hypothesise, that older firms who have more experience are more likely to win a case. Therefore, the study aims to study this interesting relationship between the age of a firm and the outcome of the appeal.

Size of the firm ​- ​It is a discrete variable that displays the legal entity's size category based on the number of employees to identify whether the firm falls in the category of large, medium or small.

More the number of employees in the firm, the bigger the firm size is. It has been assumed that larger firms have more resources than the smaller firms and hence, the hypothesis is that they can allocate more resources to fight and win the case.

Intangible Assets- Intangible assets are non-physical assets and they include, for instance, research and development, patents, and goodwill. Research and development show how much a company is investing in innovation and then registering and protecting their inventions through patents. It has been observed that companies that have higher value of patents are more likely to win an opposition. It was also found that patents with higher economic value are more likely to be litigated as compared to patents with a lower value (Harhoff and Reitzig, 2004). This variable has been considered to study the above mentioned relationship. Intangible Assets are continuous variables.

Operating Profit & loss - Operating profit or loss measures the profitability of the firms before

taking into account interest and taxes. It is a significant number for the managers that reflects the

revenue and expenses that can be controlled by them ​(Business literacy institute, 2020). It is a

continuous variable and ​it is considered to see how well the applicant and counterparty firms

managed with the company’s products and services. This variable is considered to study if it has

an impact on the outcome of the appeal.

Net profit or loss - ​Net profit or net loss shows a company’s financial health and for the firms it is important to se ​e from the competitors, investors and shareholders’ perspective. It is taken as a continuous variable. ​This variable is considered to study if it impacts the outcome of the appeal.

Sales - ​Sales is a part of the operation of a business and shows the total revenue of the company.

It is a continuous variable and is considered to see how sales of an applicant firm or counterparty firms are related to the outcome of winning or losing. Operating profit and loss, net profit or loss and sales all indicate to some extent, how well a firm is doing. With these three variables, it is expected that there will be a positive association with the dependent variable.

Cash and Bank - ​Cash and bank balances are current assets, and are extremely useful and powerful as they are the basis of any operation or activity of the firm (Bragg, 2020). This variable is considered important for the study as it highlights the position of available cash with the applicant firm who appealed and the counterparty who defends. It helps to identify whether higher liquidity influences winning or losing of appeals for a firm. The expectation is that firms with higher liquidity are more equipped with resources to tackle court cases and so more likely to win appeals in this case. It is a continuous variable.

Total assets - ​Total assets are all fixed assets and current assets of a firm which equal to the sum of the total equity and total liability of a firm. It is the economic value that is expended over time to yield a benefit for the firm (Bragg, 2020). This variable has been considered to see how assets of applicant firms and counterparty firms affect the outcome of appeals. In some studies, total asset is used as a proxy for firm size, with higher total assets indicating larger firms. While it is not used in such a way in this study, as was the case with firm size, a similar positive relationship with the likelihood of winning is expected for this variable. ​Total assets is a continuous variable.

Table 4 below presents a summary of the variables, highlighting the variable name, summary

description, source as well as the expected direction of the coefficients. The expected direction is

based on findings in earlier literature as well as the assumptions and hypotheses made above.

Table 4: Description of variables

Variable Description Expected Source

Firm size Based on the employees, large, medium and small firms

Coefficient(+) Serrano

Intangible fixed assets

R&D, patent, goodwill Coefficient(+) Serrano

Operating profit/loss Income statement, EBIT Coefficient(+) Serrano Net profit/loss Income statement, after

interest and taxes

Coefficient(+) Serrano

Total assets Current and non-current assets

Coefficient(+) Serrano

Net sales Revenue-(allowances, discount and returns)

Coefficient(+) Serrano

Sector Type of Industry Coefficient(+) Serrano

Cash & bank Liquidity of firms Coefficient(+) Serrano

Age Date of registration Coefficient(+) Serrano

5. Empirical specification and estimated results

5.1 Descriptive statistics

The empirical analysis is based on data that studies Swedish patents appeals with application years from 1998 to 2016. Table 5 below shows that out of the total 126 appeal applications, 44 were filed by the patentee, inventor, or owner, and 82 were filed by parties other than the patentee. This reflects that patents are more opposed and appealed by the counterparty. A possible reason for this could be that the invention lacks novelty or the disclosure was poor or inadequate for the granted patents. On the other hand, patentees win 10 (22.72% success rate) cases out of the 44, showing that the patent being opposed were valid. Empirical evidence in the UK, studied the outcome of validity and infringement cases and found that 38% appeal cases were won by the patentee, which means that the patent was considered to have been infringed on and/or valid (Cremers et al., 2013).

Table 5: PBR decision on the appeals filed during 1998-2016

Applicant Win Loss Total

Patentee 10 34 44

Other than patentee 44 38 82

Total 54 72 126

The data in Table 6 (see appendix) displays the legal entity's size category based on the number

of employees, taken from the ​Serrano ​Database. The size category with a higher number of

employees represents a larger firm size. The maximum frequency of appealed firms falls in the

category of large firms which shows that larger firms appealed more than smaller firms. A

possible reason for larger firms to appeal more is that they may have larger resources in terms of

employees who can invest their time and energy, higher number of patents, more experience in

the industry and good reputation in the market. Those firms that may have already faced opposition before and been successful may be encouraged to oppose and appeal more.

Table 7 (see appendix) shows the sector wise classification which was provided by the ​Serrano database that grouped the branches into eleven overall sectors. This sector wise data shows that the maximum number of firms belong to the industrial goods category followed by the corporate services. Industrial goods include goods which are bought and used for industrial and business purposes. The latter, includes activities or services provided to provide support based on specialized knowledge and technology to serve internal and external customers. The industrial sector in Sweden is well developed and has risen to 70% in the last two decades (Swedish market, 2020). The data describing the study of firms such as Volvo, Saab, Ericsson, ABB, Electrolux, etc., shows this trend. In this case, a larger number of industries in the industrial sector introduces more competition and could be one of the reasons for a higher number of appeals in this sector.

Tables 8 and 9 show the ​summary statistics for the appeal outcomes and the firm characteristics

of the sample for both applicants and counterparties.. The outcome is denoted by 1 for winning

and 0 for losing and shows that overall, 42.85% out of the total appeal cases filed at the court

were won. The minimum time taken by the court in making a decision on appeal cases is 3

months whereas the maximum time is 67 months. The average time for the settlement comes to

32 to 33 months. The data also shows that the average firm size and age fo ​r applicant fir​ms are

higher than the ​counterparty ​firms. It is also observed from the two tables (Tables 8 and 9), that

the range of some of the variables is very wide on both the datasets of applicant and counterparty

firms ​. In​tangible assets, operating profit and loss, net profit and loss, cash and bank have highly

deviated from the mean so we can conclude that there is a possibility that there are outliers

present in the data.

Table 8: The descriptive statistics for ​ Applicant firm

Dataset Min 1st Qu. Median Mean 3rd Qu.

Max.

Outcome (0/1) 0.42

Intangible F.A 0 0 0.00 1.28 0.11 112.97

Operating profit/loss -27.38 -.02 0.01 3.36 1.02 59.26 Net profit/loss -4.21 0 0.05 12.87 2.73 240.44 Total assets 0.00 0.26 3.69 150.46 64.63 2036.90

Net sales 0 0.23 1.45 46.84 26.31 531.15

Cash & bank

---

0 0.00 0.05 4.49 0.64 144.44

*Sector 10.00 20.00 20.00 32.41 45.00 99.00

*Firm.size 0.000 2.000 5.00 4.484 7.00 7.00

*Age of the firm 1.00 21.00 32.00 38.603 54.75 91.00

*Duration 3 22 33 32.59 42 67

Number of applicant firms (N=126) Amounts in SEK (in 00,000)

*Integers, (in 00,000) is not applicable.

A few values are 0 because of small very small amounts and calculations using two decimals.

Table 9: The descriptive statistics for ​ Counterparty

Dataset Min. 1st Qu.

Median

Mean

3rd

Qu. Max.

Intangible F.A 0 0 65 1.33 0.06 127.22 Operating profit/loss -31.69 -0.00 0.03 2.87 0.45 68.11

Net profit/loss -0.89 0.02 0.07 6.27 0.96 172.27

Total assets 0.01 0.24 2.54 108.83 26.88 2158.38

Net sales 0 0.18 1.66 36.10 15.86 617.44

Cash & bank

---

0 0.05 5667 4.45 0.53 144.44

*Sector 10.00 20.00 20.00 31.07 40.00 99.00

*Firm size 0.00 3.000 5.000 4.603 7.000 7.000

*Age of firm 1.00 16.00 27.50 31.40 44.25 91.00

*Duration 3 22 33 32.59 42 67

Number of applicant firms (N=126) Amounts in SEK (in 00,000)

*Integers, (in 00,000) is not applicable.

A few values are 0 because of small very small amounts and calculations using two decimals.

5.2. Multiple Logistic Regression Analysis

To predict the outcome of winning or losing, multiple logistic regression is computed on the differences between the two groups using equation (1), and the results are presented in Table 10.

The ​significance level is expressed as a p-value between 0 and 1. In the table, D​

is related to total assets that are significant at 5% and 10% level, as the p-value is less than 0.05 and 0.1 respectively. Similarly, cash and bank(D ​

) is significant at the 10% level of significance as the p-value is less than 0.10. However, other variables such as firm size, intangible assets, operating profit and loss, net profit and loss, net sales and sector are not significant at any of the 1%, 5% or 10% levels of significance. It can also be observed that the standard error (SE) is lower compared to others in case of D ​

and D ​

which is also the reason for them being significant. D ​

(total assets) has a positive coefficient which shows that having more total assets has a positive impact i.e., it is more inclined to produce an outcome of 1 than 0. D ​

has a negative coefficient which shows that having more cash and bank has a negative impact i.e., it is more inclined to produce an outcome of 0 than 1.

The two deviances given are null and residual. The value of null deviance is 172.09 on 125

degrees of freedom and the residual deviance is 159.25 on 116 degrees of freedom. Deviance is

used to measure the goodness of fit for a model. A higher number indicates a bad fit. The null

deviance indicates how efficiently the response variable is predicted by a model. From this, it

can be calculated that 9 (125-116) independent variables decreased the deviance to 159.25 from

172.09, a significant reduction in deviance. The Residual Deviance has reduced with a loss of

nine degrees of freedom.