Construction Toy Industry beyond the Lego brick.
Determining the value indicators for patents
Sabina Makhmudova
10235132
Amsterdam, June 29, 2014
BSc Business Studies Thesis Supervisor: Dr. Ranjita Singh Academic year: 2013-2014 Semester 2, Block 3
2
Table of Contents
Abstract ... 4
1. Introduction ... 5
2. Literature review ... 7
2.1 Industry life cycle ... 7
2.2 Construction toy industry and The Lego group ... 9
2.3 Patents ... 10
3. Conceptual Framework ... 12
3.1 Patent value indicators ... 12
3.1.1Forward Citation ... 12 3.1.2. Backward Citation. ... 13 3.1.3. Number of claims ... 14 3.1.4. Inventors ... 15 4. Methodology ... 16 4.1. Research design ... 16 4.2. Sample... 17
4.2.1. Construction toy industry ... 17
4.2.2. The Lego Group and patent indicators ... 18
4.3. Data collection ... 18
4.4. Variables: Dependent variable ... 19
4.4.1. Forward Other Assignee Citation ... 19
4.5. Variables: Independent variable ... 19
4.5.1. Backward Citation ... 19
4.5.2. Number of claims ... 20
4.5.3. Number of Inventors... 20
4.6. Variables: Control variable ... 20
4.6.1. Year Patented ... 20
4.6.2. Time to Grant ... 21
4.6.3. Number of Patents per year ... 21
4.6.4. Number of foreign backward citations ... 21
4.6.5. Number of National Classes ... 22
4.6.6. Number of Patent Classes ... 22
4.6.7. Number of Attorneys ... 22
3
5. Results ... 23
5.1. Industry and Lego evolution ... 23
5.2. Patent indicators results ... 25
5.2.1. Forward Citation adjusted for time ... 26
5.2.2. Forward citations as a proportion ... 26
5.3. Results summary ... 30
6. Discussion ... 31
6.1. Construction Toy Industry ... 31
6.2. The Lego Group ... 33
6.3. Patent value indicators ... 34
7. Theoretical implications and further research ... 36
8. Managerial and Business implications ... 37
9. Limitations ... 37
10. Conclusion ... 39
References ... 40
4
Abstract
Previous research has shown that industries have a tendency to evolve in an S-curve manner. Evolution of an industry is often measured through knowledge accumulation and innovation in that particular industry, given the growing importance of information in a modern society. However, mere number of registered inventions, also known as patents, has been shown by previous research to be an inaccurate measure of value of knowledge. Based on this we decided to carry out a three level analysis of construction toy industry, due to relative importance of the product and recent developments in the industry. Firstly, we analyze patenting history to determine the evolution stage. Secondly, we try to determine if knowledge developed by the leading company- the Lego Group, has a significant effect on its performance. Thirdly, we try to determine factors that influence the value of patents. Data was collected from United States Patent and Trademark Office and analyzed using graphs and regression analysis. According to the finding it was determined that construction toy industry is currently in the mature phase and the recent success of the Lego Group was not related to patented knowledge. Lastly, we found forward citations, the number of national classes and the number of years to be positive indicators of value, while backward self-citation and the number of classes to be negative indicators of patent value.
5
1. Introduction
It’s in human nature to wonder about the future and how it can be changed. Unfortunately, the gift of precognition is not available to us. In the business world too, firms try and predict the future in order to stay ahead of their competitors. A theory often used to help firms predict the future evolution of the industry is the theory of S-Curve industry development. However, the question at hand remains, do all the industries follow S-Curve shape and if they do, how can we determine the stage in which a given industry is now?
Our society is quite often referred to as information society, where competitive advantage can be gained through possession and exploitation of information. Possession of information or knowledge in case of a company can refer to the information that was patented by the organization. This notion was initially developed by Fritz Machlup (1962) in his book The Production and Distribution of
Knowledge in the United States and since then has been highly discussed in academic literature. On a
different note, Barney (1991) and other resource-minded academics (e.g. Wernerfelt, 1984) believe that resources are the key to gaining an upper hand in a competitive industry, thus making patents an intangible asset that can create a competitive advantage for the company. Many researchers believe that patenting history of an industry/company can be a good tool in determination of evolution stage (Malerba & Orsenigo, 1996; Andersen, 1999). However, there are also certain drawbacks to using patenting history; for example, not all inventions are patented. Given this, we firstly would like try to determine evolution stage of a selected industry using the patenting history method.
We believe that construction toy industry can be a good choice for this study. Toys are one of the items that have been used since the ancient times to this day. Given their vast history, it is interesting to analyze whether this product follows the patterns of industry evolution cycle and if not, what are the reasons for that. Due to a great variety of toys, in this study we will mostly concentrate on construction toys and one specific company within this industry. Construction toy industry, as many other industries, faces a lot of difficulties in the modern market due to technological developments of 21st
6
century. Toys that require physical activity are being replaced by electronic games, which include video games, virtual toys and others. Nevertheless, construction toys remain an important tool in child development, especially in development of cooperation skills and spatial ability (Brosnan, 1998; Kato, Hattori, Iwai & Morita, 2012).
Furthermore, the recent news about the construction toy giant- The Lego Group, about its positioning as the 2nd largest toy manufacturer (2013) and four and a half fold increase in net income in the past 5 years (2008-2013) has surprised many (Lego Group Annual Report 2012, 2013; The Lego Group Annual Report 2013, 2014; Wienberg, 2013). This superior performance of The Lego Group makes us wonder about the underlying milestones that led to this achievement. Was this achievement a product of successful internal development or was it due to overall development of construction toy industry? Developed and registered knowledge of Lego could have played an important role in its success. Knowledge management, intellectual capital and related topics have received a great attention from researches. Studies have shown support for knowledge being an important intangible and not easily imitable asset in a variety of industries, including IT firms, electrical firms, sports associations and others (Berman, Down & Hill, 2002; Mata, Fuerst & Barney, 1995; Ndlela & Du Toit, 2001). Unlike any other asset, knowledge requires careful management and is difficult to measure, store and transfer both within and outside the company (Lubit, 2001; Argote & Ingram, 2000). Thus, in order to answer this question, a special tool is need to be developed, that would allow us to easily measure the value of The Lego Group existing knowledge, i.e. patents.
Overall, in this study we would like to carry out a three level analysis focused on the industry, company and patent level, because these levels of analysis complement each other. Industry analysis through patenting history would allow us to plot the evolution curve and show us the current development stage of the industry, but would not explain why it is in that stage and if the measures used for analysis are accurate. Analysis of the leading company through patents would demonstrate the possible reasons for industry’s recent performance and the role of patented knowledge in it, however it
7
would not show the factors that influence the significance of a certain invention. Lastly, patent level would unravel the factors that influence the significance of an invention and also show us if the measures used to record industry evolution are accurate. All of this should enable us to answer our main research question: To what extent is knowledge developed and patented by The Lego Group
responsible for the success of the company in the recent years?
This study is structured in the following way. Firstly, we will review the current knowledge on the three main topics of this study: industry evolution, construction toy industry and patenting. Secondly, theoretical framework will be developed. Thirdly, methodology and data collection process will be explained for all three levels of analysis. After that, results will be presented. The study will end with a discussion and conclusions section.
2. Literature review
In this section we will present and compare existing finding and theories in the area of our interest, in order to see on which areas the light has not been shed yet. Firstly, we will discuss the existing knowledge in the area of industry life cycle and different types of industry evolution. Following that, we will describe the industry and company of our choice; briefly discuss its history and relevance to the current research. The last two sections will cover the current knowledge about patents and discuss their relevance in the analysis of industry life cycle and company performance.
2.1 Industry life cycle
First of all it is important to understand what industry life cycle implies. By industry life cycle we understand a process where there are entries, exists, innovation and restructuration within the industry, which happen in regular cycles as industry becomes older (Klepper, 1997; Malerba & Orsenigo, 1996b). As explained by Schumpeter (1942), industry evolution is an inevitable product of modern economy. New products, processes and markets create impulses that keep “the capitalism engine in motion” (Schumpeter, 1942). Evolution of an industry can have different dimensions, for example
8
Schumpeter Mark I and Mark II type of evolution (Malerba & Orsenigo, 1996b); different level evolution, such as the ones identified by Malerba and Orsenigo (1996a), which include specific dynamics of industry, structural dynamism and structural evolution; and others. These characteristics are not related to the current research, however we believe that reader should know that evolution may differ from industry to industry.
According to Schumpeter there are four stages in economic development. These are: depression, revival, prosperity and recession and all together this stages form a well-known S-shaped evolution pattern (Schumpeter, 1939). Steven Klepper (1997) puts this cycle into a product lifecycle context and provides the following description: “…initially the market grows rapidly, many firms enter, and product innovation is fundamental, and then as the industry evolves output growth slows, entry declines, the number of producers undergoes a shakeout, product innovation becomes less significant and process innovation rises”(p. 149). Since then the S-shaped evolution theory has expanded and now is used to explain evolution of products, processes, technology and even industries. Schumpeter’s evolution cycle is a well-researched topic, which found support in many industries, for example chemical, pharmaceutical, mechanical and electrical (Andersen, 1999; Achilladelis, Schwarzkopf, & Cines, 1990; Achilladelis, 1993).
To underline the importance of evolution, Lumpkin and Dess (2001) show that certain evolution stages are more favorable for entrepreneurs than others, for example a startup performs best if it enters the industry in the its growth (revival) stage. Furthermore, knowing the evolution stage of your industry, may allow you to take preventive actions, develop different scenarios and appropriate plans of action (Schoemaker, 1995). Andersen (1999) also shows that S-shaped growth cycle is not significantly dependent on the external shocks, meaning that evolution of industry is more dependent on internal developments. This means that it is up to companies to take initiative actions. All of these show that industry lifecycle is indeed an important factor to consider.
9
The S-shaped evolution curve is a generally accepted theory, with minor exceptions. For example, some technological innovations experience one growth wave after another, with not periods of decline (Andersen, 1999), or the fact that s-shaped evolution can be only used in a fully developed capitalism (Kleinknecht, 1990), or the presence of random fluctuation (Haupt, Kloyer & Lange, 2007; Andersen, 1990).
2.2 Construction toy industry and The Lego group
Toys are objects used in playing. First toys were made from materials available in the immediate environment and manufactured by parents or by children themselves (UNESCO, 1988). Toys have significantly evolved since then, especially with development of mechanics and electronics. The number of different toys in modern market is immense, however in this study we would like to specifically look at the construction toy industry, for two main reasons. Firstly, the scope of this study does not allow for a broader range of toys to be studied and secondly, construction toys play an important role in child development. As was already mentioned, construction toys allow children to develop social and cooperative skills (Brosnan, 1998; Kato, Hattori, Iwai & Morita, 2012). Furthermore, they promote development of 3-dimensional spatial ability not only in children, but also in adults with low visualization skills (Sorby & Baartmans, 2000).
However, importance of this industry does not necessarily make it an attractive industry to enter. As Haupt et al. (2007) mention, attractiveness of technology (or industry, in our case), is dependent on its current life cycle stage. Unlike other industries, such as tobacco, where overall consumption increases, due to growing population, but per capita consumption has decreased in recent year due to populations’ healthier life choices; development in construction toy industry are not as easy to explain (Eriksen, Mackay & Ross, 2012). According to the readily available information it is difficult to determine its stage of evolution. On one hand we can see growing population (thus higher demand for toys) and relative success of The Lego Group in the recent year (Wienberg, 2013). On the other hand, we also know that The Lego Group has developed its dominant design of interlocking blocks more
10
than 50 years ago, which has not radically changed since then, thus should sooner or later enter the decline stage of its development. In addition to that construction toys are being forced out of the market by electronic toys. Thus, this research will be able to clarify in which stage this industry is.
Since we have presented the reasons to study construction toy industry, we will now explain our particular interest in The Lego Group and will present some background information about it. The Lego Group (from now on: Lego) was establish in 1932 in Billund, Denmark, where the main headquarter still remains. The first “Lego Brick” was produced back in 1949 and the design developed by Lego was patented in 1958. Our interest in Lego is based on several factors. First of all, Lego is a construction toy company that regularly patents its inventions and is the most well-known construction toy company. Lego’s presence and performance has further escalated in the recent years, making it 2nd biggest toy manufacturer and an interesting company to study (Wienberg, 2013). However, the reasons for Lego’s success are not that obvious. On the one hand, it is possible that construction toy industry is in the growth stage, so Lego is naturally growing as the market expands. On the other hand, it is also possible that Lego has superior competitive advantage that allows it to conquer the mighty heights of the toy industry.
2.3 Patents
As was previously mentioned, today’s era is information era, where knowledge and information can be a stronger competitive advantage than any other capability. One of the ways that companies ensure that their knowledge is preserved is patenting. As defined by World Intellectual Property organization: “a patent is a document, issued, upon application, by a government office, which describes an invention and creates a legal situation in which the patented invention can normally only be exploited with the authorization of the owner of the patent…. a patent is the right granted by the State to an inventor to exclude others from commercially exploiting the invention for a limited period, in return for the disclosure of the invention, so that others may gain the benefit of the invention” (WIPO, 2008). Other definitions of patent exist, for example United States Patenting and Trademark
11
Office (USPTO) definition, however all of them carry more or less the same meaning. Patenting to some extent ensures inimitability of resource, by protecting company’s ownership of it. It ensures that the same research is not carried out twice and that industry can move forward more quickly. Lastly, a company can benefit from patenting by selling or licensing its invention (Bessant & Tidd, 2011). However, patenting also make companies’ knowledge available to the wide public, allowing competitors to find loopholes and ways around to copy their inventions. And lastly, there a certain costs associated with patenting (Bessant & Tidd, 2011). Nevertheless, patenting is widely used in construction toy industry, making it a suitable measure for us to use.
There are a certain benefits to using patents as a measure of industry performance. First of all they present a homogeneous measure of technological development and patents are available for a long time period. Secondly, patents provide detailed information about each invention, which are easily accessible to everyone (Malerba & Orsenigo, 1996b). Patenting history has been shown to be a reliable source of information for determining industry evolution cycles (Malerba & Orsenigo, 1996b; Haupt, Kloyer & Lange, 2007). Indeed, for a long time the number of patents has been used as a primary indicator of industry development and performance. Recent studies however shown, that number of patents may be an inaccurate representation of industry performance (Harhoff, Scherer & Vopel, 2003). Malerba and Orsenigo (1996) further highlight this, by saying that patents cannot be distinguished in terms of relevance unless specific analysis of patent characteristics is carried out. Furthermore, it is important to remember that not all inventions are patented and different companies have different patenting tendencies (Melrba orsenigo 1996). Some companies, such as Boeing prefer secrecy over patenting (Glickert, 2010). Nonetheless, patenting history is still widely used and supported in academic literature.
This vast use of patenting history in research can be explained by development of new frameworks for patent data analysis. Academics these days do not merely use the number of patents as an indicator of industry performance. Instead they use a range of characteristics, for example citation,
12
to measure the value of patents. Given this, we decided to include the third level of analysis, where we would like to analyze the effect of different patent characteristics on the value of the Lego’s patents. USPTO website shows certain characteristics of patents that are available for all registered patents. These characteristics include (only relevant characteristics will be listed): date filed, date granted, inventors and/or assignee, applicant, current U.S. class, forward citation, backward citation, foreign citation and inventors’, applicants’ and assignees’ countries of origin. The amount of research on these characteristic differs from variable to variable, citation being the most well researched characteristic.
3. Conceptual Framework
In this section we will present the conceptual framework of the 3rd level of analysis, where we hypothesize about the effect of patent characteristics on the value of patent.
3.1 Patent value indicators
3.1.1Forward Citation
Forward citation is citations of the original patent by subsequent patents (Lanjouw & Schankerman, 2004). Trajtenberg (1990) showed in his research that value of patents, just like the value of scientific publication can be measured through the number of forward citation received. Informative value of forward citations as an indicator of patent usefulness was further supported by researches such as: Harhoff et al. (2002); Harhoff, Narin, Scherer & Vopel (1999); and Lanjouw & Schankerman (2004). Lanjouw and Schankerman (1999) also tested forward citation (along with backward citation, number of claims and family size) as an indicator of patents knowledge contribution, and have shown it to be the least noisy and most reliable indicator and later they have shown that number of forward citations increases litigation, which is a good indicator of its importance (Lanjouw & Shankerman, 2001). Furthermore Jaffe, Trajtenberg, Fogarty (2000) found that forward citation was an indicator of both technological and economic value of patent. Overall, it seems that
13
measurement of patent value through forward citation is well grounded, especially in the pharmaceutical, chemical and technological areas. Last, but not least, forward citation by different assignees/inventors (not citation) are the most useful indicator of patent value, because self-citation increases the chances that the new invention is more “derivative” in nature (Lanjouw and Schnakerman, 2001). Assignee, according to USPTO, is the entity that is the recipient of a transfer of a patent application and patent from its owner of record (Assignee, n. d.). Given this unanimous theoretical support, we use forward citation by other assignees/inventors as an indicator of its value.
3.1.2. Backward Citation.
Backward citation, or in other words the number of prior patents cited by the patent, is also well researched topic; however unlike forward citation it has received mixed opinions (Lanjouw & Shankerman, 2001). Pioneered by Carpenter, M., Cooper, M., et al backward citation to scientific literature has been to found to increase usefulness of a patent. Lanjouw and Schankerman (2001; 2004) how found some support for usefulness of backward citation as an indicator of value and have discussed the fact that backward self-citation can be an indicator of companies’ investment and dedication to the research field, however these were more theoretical arguments, rather than finding backed by data. Furthermore, Harhoff et al. (2002) have found that backward citation positively correlates with technological value of patent. It was also found that backward self-citation decreases the probability of litigation, meaning that as the number of backward self-citation increases, usefulness of patent decreases, because patent becomes more derivative in nature (Lanjouw & Schankerman 2001, Lanjouw & Schankerman, 1999). Given this, we hypothesize that backward self-citation decreases usefulness of patent, while backward citation by or of others can either increase usefulness or not affect usefulness at all.
Hypothesis 1: Backward self-citation by Lego would decrease the usefulness of patent, in other words,
14
Hypothesis 2a: Backward citation of others would increase the usefulness of patent, in other words, as
proportion of backward citation increases, forward citation should increase as well.
Hypothesis 2b: Backward citation of patents would not affect the usefulness of patents.
3.1.3. Number of claims
Another factor that can potentially be used to measure the value of patent is the number of claims that a patent has. Claims, according to USPTO: “…define the invention and which aspects are legally enforceable. The specification must conclude with a claim particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention or discovery” (Claims, n. d.). Supporters of claims being an indicator of usefulness preach that as the number of claims increases, the breadth of invention increase, as a result increasing its both economic and technological value potential (Markus Reitzig, 2001). Lanjouw and Schankerman (2001) show that as the number of claims increases, the probability of litigation increases as well, indicating it being more valuable. This is further supported by study based on R&D and patent data of 100 US manufacturers, which shows claims being the second best indicator of usefulness (Lanjouw & Schankerman, 1999). In 2004 Lanjouw and Schankerman further support this ideas by identifying that claims are the best indicators of patent usefulness in all the studies industries, except drug (forward citation). However, there is also evidence presented by Jaffe et al. (2000), which shows that the number of claims is actually a bad indicator of patent value, possibly because the greater is the number of claims, the broader and more general it becomes, making it less useful. Given the above finding, we incline more towards the belief that greater number of claims increases the value of patent.
Hypothesis 3: Number of claims will have a positive effect on usefulness of a patent, in other words, as
the number of claims increases, the number of forward citation should increase as
15
3.1.4. Inventors
Yet another potential indicator of patent value is the inventors of the patents, because it is their knowledge and ideas that go into the core of a patent. To be more specific, in our study we are interested in the number of new inventors (meaning, inventors that have not been previously involved in the patented inventions by this company) with every patents. We will apply theories on external knowledge versus internal knowledge development and external versus internal recruitment to build our hypothesis. The research on both of these topics is similar: internal development/recruitment is important; however external knowledge/recruitment brings diversity, new outlook and ideas to the organization. An organization has to manage to create equilibrium between these two worlds (Gittelman & Kogut, 2003). De Clercq and Dimov (2008) highlight the importance of external knowledge and creating harmony between the two sources of knowledge, especially in cases when there is an incongruity between existing company knowledge and what company wants to do. William Chan (1996) believes that external recruitment should take place only when external candidate has a superior knowledge to internal candidates. Furthermore the book by Noe, Hollenbeck, Gerhart & Wright (2012) shows that pros of external recruitment can be: increased diversity and talent pool, new insights brought from outside and facilitation of faster growth. On the other hand the con of external hiring is the fact that new employees may need time to get settled and figure out how things work. Furthermore, external hiring can also hurt existing employee loyalty and morale. If this theory is to be applied in context of new inventors, we believe that new inventors can bring new vision and ideas to the table, which would create an invention that is different from the previous invention. This would mean that invention by new inventor would be more original and less derivative in nature, i.e. more valuable (Lanjouw and Schnakerman, 2001). Consequently, we base our hypothesis on these assumptions.
16
Hypothesis 4: Number of new inventors will have a positive effect on usefulness of a patent. In other
words, as the number of new inventors increases, the number of forward citation
should increase as well.
4. Methodology
The following section will demonstrate the design and methodology of this study for all three levels. Firstly, the design will be discussed and then pros and cons of chosen design will be mentioned. Secondly we will present the chosen sample and explain the measures used. Lastly, we will describe the actual procedure and data analysis steps.
4.1. Research design
For this study we chose a quantitative longitudinal design method. Firstly, it is clear that for this research we use deductive approach, because a certain level of knowledge on topic of our study already exists. Given this, we test proposed hypotheses, which are based on previous research. In order to do so, we use quantitative research method, because unlike qualitative method, this method allows us to test the proposed hypotheses on a large scale of data, which would be impossible to do with qualitative research. Furthermore, the chosen variable are quantifiable and at this point we try to establish what relationships exist between patent value and its indicators, rather than trying to understand why this relationships exist, making quantitative method once again more applicable (Kumar & Phrommathed, 2005; Saunders, Lewis, & Thornhill, 2011). Secondly, our research is longitudinal, because we collect the same indicators and patent data for the same industry/company over a long period of time (Saunders, Lewis, & Thornhill, 2011). Longitudinal study is necessary in our case, because unlike other industries, such as pharmaceutical industry, construction toy industry does not have enough patents per year in order for us to conduct appropriate testing of hypothesis (Saunders, Lewis, & Thornhill, 2011). Furthermore, studying evolution of an industry and company
17
inevitably entails collection of data over a certain period of time, making this method once again the most appropriate one.
Of course there are certain drawbacks regarding this method. Firstly, as it was mentioned before it does not help to answer the why questions, for example it does not help us to understand why certain changes occur in the industry. Furthermore, external and internal validity might also be an issue in our research, for example because we assume that forward citation is an indicator of its value (Saunders, Lewis, & Thornhill, 2011). Nevertheless, we believe that this design is he most appropriate for this study.
4.2. Sample
Due to the fact that our study analyses both company and industry level data, we will discuss each one separately.
4.2.1. Construction toy industry
The focus of this level is to study the history of construction toy industry and identification of evolution stage of industry at the moment. For this study convenience sampling was used. We have used only patents available on the website of USPTO, excluding other patenting offices, mainly due to convenience, time constrains and USPTO being the biggest online database of patents. Furthermore, USPTO has a specific patent classification, according to which patents that are categorized in one of the sub-groups of construction toy patents, are no longer necessarily listed under general classification as construction toy, but rather as, for example, wheeled construction toy. As a result, we decided to analyze the most general category: Construction toy (USPTO Class: 446, Subclass: 85). Data was collected for all the patents present in this category for the time period from 1 January 1976 to 1 May 2014. 1976 was chosen because patents starting from 1976 have all the necessary data; patents registered before this date might be missing some of the important data. In total, data for 221 patents were recorded. Out of those only 95 were used in the analysis. 126 patents were eliminated due to the fact that no assignee was present for this patents, thus they were categorized as “garage inventors”,
18
and focused on the company assigned patents. . In this study we concentrate only on existing companies. In total, 64 companies have been identified. Out of 64 companies, 10 companies had more than 1 patent in the period from 1976 to 2014. The company that patented the most was InterLego AG (The Lego Group) with 13 patents in this category, in the chosen time span. The average number of patents per company is 1.4 patents; the average of patents per company among the top 10 patenting companies is 3.8. There were in total 13 national classes, the most popular being USA, with 50 patents and China, with 17 patents. Latest patent was granted in 2014.
4.2.2. The Lego Group and patent indicators
The focus of this level is to study patenting history of Lego; identification of its evolution pattern and measuring the patent indicators necessary for hypotheses testing. There are some sampling similarities with the industry level, this includes: time period 1 January 1976-1May 2014 (chosen for the same reasons, as previously). Here we concentrated more on the company, rather than a patenting category. We have chosen two search items for this sample. Firstly, it had to be in Amusement
Devices: Toys category (USPTO class: 446); secondly, assignee name has to be InterLego AG (The
Lego Group patenting name). As it was later checked, all the patents were registered under one of the construction toy classes. In total 109 patents were recorded. None of the patents were eliminated. Latest patent was granted in 2009.
4.3. Data collection
Mono method was used to collect all the data. Database was used to collect data for this study. We used United States Patent and Trade patent database, because it is the largest patent and trademark database available and it provides free access to electronic copies of issued patents. Other patent and trademark offices exist, such as European Patent Office or Eurasian Patent Organization, however USPTO provides the most information, which is available online for free. Patent information was collected manually for every patent and recorded in Microsoft Excel, because direct extraction of the necessary data from the website is not possible. Firstly, data for industry was collected, and then
19
information for Lego was collected. Data collected for Lego evolution analysis and patent indicators were collected simultaneously, due to convenience. The data was later transported to IBM SPSS software for analysis.
4.4. Variables: Dependent variable
4.4.1. Forward Other Assignee Citation
In order to collect this variable, USPTO website was used. Data was collected for all InterLego AG patents from 1 January 1976 till 1 May 2014 and that are classified under Construction Toy class. Forward citation was recorded using three main parameters: total forward citation, forward citation by Lego (self-citation) and citation by other assignees. In order to arrive at the final variable, the number of forward citations will be adjusted for time and total number of patents. Time adjustment is necessary because the old patents had more time to receive forward citation, unlike the newer patents, which had fewer years to get cited by other patents. Thus number of forward citations received will be divided by the time difference between current year (2014) and year the patent was granted. This will be used in first round of analysis. For this variable we use year granted, because that is the year that invention was officially registered as a patent. For the second round of analysis we will use the proportion of forward citation by other from the total number of forward citation. The number of forward citations by Lego Group itself, was measured for the purposed of accuracy and to avoid making counting mistakes.
4.5. Variables: Independent variable
4.5.1. Backward Citation
This variable was also collected on USPTO website, following the same conditions we used for dependent variable. For backward citation three parameters were recorded: total backward citation, backward citation by Lego and backward citation by other assignees. From the collected data, two main variables were created: proportion of Lego patents from the total number of patents that were
20
referenced and proportion of other assignee patents from the total number of patents that were referenced.
4.5.2. Number of claims
As previous research shows, numbers of claims indicate breadth of patent and its potential contribution that is why it is classified as one of the independent variables. As other variables it is collected from USPTO website, following the same conditions. The number of claims is always numbered, so for this variable the number of claims were just counted and recorded. This makes it very convenient, however also raises a certain problem. What if the actual number of claims is not representative of the number of contribution it makes, what if a claim under one number actually consists of more than one claim. For now, we assume accuracy of claim numbering; however this is indeed one of the limitations of this study.
4.5.3. Number of Inventors
Once again, data for this variable was collected from USPTO, following the same conditions as the variables described above. Firstly, for each patent the names of all the inventors were recorded. After that, using Microsoft Excel formulas, the number of new inventors, meaning the inventors who previously have not been registered as inventors, and the cumulative number of new inventors were counted. There is a certain limitation to this variable, which comes from the fact that we only look at the patents since 1976, this means that it is possible that new inventors have been registered before, but due to our timeline cut of we are not aware of that. Nevertheless, the number of patents prior to 1976 is not that big and neither is the number of inventors prior to 1976.
4.6. Variables: Control variable
4.6.1. Year Patented
For this variable we simply record the year in which the patent was granted. We chose to include the year granted over year filed, due to several reasons. In this study we used only granted patents and excluded filed patents, making the grant date more relevant than filed. We also have to consider the
21
time lag between the application and issue date. The patent does not get cited during that period, making the grant year a more accurate measure given that our dependent variable is based on citation.
4.6.2. Time to Grant
This variable represents the time difference between the date the patent was filled and the date it was granted. It is possible that more complex, important and technologically novel patents take longer to be granted, due to greater amount of work needed to be put in to make sure no overlap with previous patents exists (Popp, Juhl & Johnson, 2003).
4.6.3. Number of Patents per year
By number of patents we mean the number of patents filed in a given year. We believe that as the number of patents per year increases, there is an increasing chance that the value of those patents will also increase, because as the famous idiom says: practice makes perfect. Some however may argue that number of patents can also have a negative effect, because as the number of patents increase, their contribution per patent may decrease. High patenting rate might signify derivative nature of patents, because they take less time to develop, since it is easier to build on the previous knowledge rather than coming up with something completely new. Nevertheless, the number of patents is still used by many as an indicator of the development of an industry/company that is why we need to control this variable
4.6.4. Number of foreign backward citations
The number of foreign backward citations was originally collected as one of the independent variables, but was later reclassified due to access problem to some of the patents. In general we believe that foreign forward citations increase the value of patent, because Lego group, as it was later discovered has only limited number of patents patented not in USPTO. This means that most of patents that were referenced would belong to other assignees, which either would have no and positive effect on patent value, as if was discussed in theoretical framework, which means that this variable needs to be controlled.
22
4.6.5. Number of National Classes
The number of national classes indicates the number of countries from which forward citing patents originate from. This is an indicator of international spillover of knowledge, which can also indicate the overall value of the patent; because the better is the patent the greater is the incentive for their overseas colleagues to build on this knowledge (Lerner, 1994).
4.6.6. Number of Patent Classes
Research by Joshua Lerner (1994) shows that number of patent classes under which a certain invention is classified can have a positive effect on the value of patent. Even though Lerner used the international patenting classification for his research, we believe that his findings can also be transferred to the US patent classification. This is why we control this variable.
4.6.7. Number of Attorneys
The last control variable is the number of attorneys that have been involved in the patent. Research by Bessen and Meurer (2004) has shown that companies that patent a lot usually have a greater number of attorneys involved. The attorney team allows the company to inspect all the possible patent breaches and sources of future litigation. The bigger the team, the more complicated and broad the patent may be. More difficult patent would require bigger team, where each lawyer brings his own experience and knowledge to the table and that is why we control for this variable. In addition to that, experienced attorneys can also contribute their knowledge and information to the innovation process within the company. For our research this means that the greater is the number of attorneys, the more valuable is the patent.
4.7. Analysis
In order to carry out necessary analysis of the data we used both Microsoft Excel 2010 and SPSS IBM, 20th and 22nd version. The data for the evolution of both the industry and Lego were visualized using graphs and analyzed using linear regression in order to see the evolution stage and determine either decline or rise of the industry. The data was then divided into three periods to determine at
23
which stage the industry is declining or rising according to patenting history. In order to analyze numerous variable of Lego’s patenting history, we use linear regression analysis to establish the relationship between dependent and independent variables. 4 Models will be tested. First model will consist of dependent and control variables; independent variables will be added one by one to the consequent models, resulting in 4 models.
5. Results 5.1. Industry and Lego evolution
In order to determine the evolution stage of both the industry and the company the grant and filed years were recorded and analyzed. The analysis was carried mainly through use of graphs, linear regressions and correlation.
In Table 1 correlations between number of patents for the industry and Lego are presented, for both year filed and granted. According to the table, there are four significant correlations. There is a positive correlation between number of years and number of granted patents in industry (r(39)=0.496,
p<0.01), between Lego number of granted patents and number of granted patents in industry
(r(39)=0.475, p<0.01), between Lego number of filed patents and two other variables: number of patents filed in industry(r(39)=0.403, p<0.05) and granted to Lego(r(39)=0.335, p<0.05). Presence
Table 1, Correlations between the variables
1 2 3 4 5
Year (1976-2014) (IV) 1
Industry Granted (DV) 0.496** 1
Industry Filed (DV) 0.312 0.134 1
The Lego group Granted (DV) 0.059 0.475** 0.295 1
The Lego group Filed (DV) -0.065 0.191 0.403* 0.335* 1 Note. N=39. *p<0,05. **p<0,01. DV=dependent variable, IV=independent variable.
24
of these correlations signify the possibility of presence of certain relationship between the variable that would later by analyzed in more detail using regression analysis.
We continue our analysis by plotting a graph (see Appendix) of the patenting history for both data sets, including both filed and granted data. According to the trend line for both Lego and the industry as a whole, the number of patents per year is increasing, because the trend line has a positive slope. We conduct regression analyses of relationship between years and the number of patents granted. Patents granted were used in this case, because on the whole this data yield more significant relationships than patents filed, meaning it is more informative.
Table 2, Regression of Industry and The Lego Group patenting history in different time sectors
1976-2014 1976-1988 1980-2000 2000-2012 2000-2014 Industry (DV) Coefficients Std. Error Beta 0.102** 0.115 0.418* -0.005 0.000 0.029 0.087 0.195 0.128 0.101 0.496 0.370 0.542 -0.013 0.000 R2 0.246 0.137 0.294 0.000 0.000 Lego (DV) Coefficients Std. Error Beta 0.017 0.055 0.418 -0.670** -0.571** 0.047 0.75 0.258 0.160 0.124 0.059 0.217 0.438 -0.784 -0.787 R2 0.004 0.047 0.192 0.615 0.619
Note. Dependent Variable is Industry and The Lego Group patents number. N=39. *p<0,05. **p<0,01 DV=dependent variable
Table 2 shows regression analysis data for the industry and Lego for five different time periods. The time period 2000-2012 was included for the following reason: according to our calculations the average time to grant a filed patent for both industry as a whole and Lego is 2 years. Thus, in order to adjust our data for the granting time, we eliminated years 2013 and 2014 for two regressions.
25
According to the table there are several significant relationships. Firstly, in total (meaning time period 1976-2014) the number of the patents per year is increasing in the industry as a whole (β=0.496,
p<0.01, R2=0.246), meaning that variance in the number of patents per year is to 24.6% explained by
change in year. Furthermore, a positive relationship was also established between industry and the time period 1976-1988 (β=0.542, p<0.05, R2=0.294). Lastly, a negative relationship was established for
Lego and time periods 2000-2012(β=-0.784, p<0.01, R2=0.615) and 2000-2014 (β=-0.787, p<0.01,
R2=0.619). The difference between the findings for the two year periods do not appear extremely
different.
Lastly, we would like to look at the development of 10 most patenting companies in the industry. According to the diagram 3 (see Appendix), the major 10 companies had a significant increase in number of patent of the period from 1992 to 2005. This to a certain extent agrees with our finding in previous section, because there a significant positive beta of 0.418 in the period from 1980 to 2000. It is also visible that Lego is the major patentee in this class. It should also be mentioned that approximately 57% if our data consists of companies that patented only once in this category in the period from 1976 to 2014. Implication of this would be discussed in discussion.
5.2. Patent indicators results
Regression analysis was carried out for four models. First model includes all the control variables, second model adds backward citation by other assignees and Lego to the first model, third model further adds number of new inventors, and last model incorporates all the variables including number of claims. This analysis was carried out twice, for two differently adjusted forwards citations by other assignees. In the first round we used number of forward citation by other assignees divided by the year difference between the year it was granted and current year (2014). This way we eliminate the issue of time available for the patent to be cited. In the second round we use proportion of forward citations by other assignees from the total number of forward citations.
26
5.2.1. Forward Citation adjusted for time
The first model- control variable model explains 22.8% of variance in forward citation by other assignees. However, out of six variables only two were found significant: national classes (β=0.459,
p≤0.001, R2=0.228) and year it was granted (β=0.223, p<0.05, R2=0.228). The second model, which
adds backward citation to the equation increases explanatory value to 24.6%, however neither proportion of backward citations by Lego (β=-0.092, ns) nor proportion of backward citation by other assignees (β=0.047, ns) made any significant contribution to the model. The increase in R squares can be attributed to its general tendency to increase when extra variable is added. The third and the fourth model do not add and significant new indicators to the model either, accounting for 24.7% and 25.4% of variation in forward citation by other assignees respectively. Neither number of new inventors
(β=-0.036, ns), nor the number of claims (β=-.095, ns) have a significant effect.
5.2.2. Forward citations as a proportion
According to model 1, there is only one significant explanatory patent indicator: number of patent classes (β=-0.215, p<0.05, R2=0.076). The overall explanatory value of the model has decreased in
comparison to the first model of the previous round of regressions. The second mode has R squared of 16.7% which is almost twice greater than explanatory value of first model. According to it, backward citation by other assignees does not have a significant effect (β=-0.137, ns), while backward citation by Lego has a significant negative relationship with proportion of forward citations (β=-0.435, p<0.05,
R2=0.167). In addition it supports the finding of model 1; number of patent classes once again has a
negative relationship with dependent variable (β=-0.256, p<0.01, R2=0.167). The third and the fourth
model add new inventors (β=0.042, ns) and number of claims (β=-0.123, ns) to the models. However, neither independent variable has a significant effect. Previously mentioned significant variables, remain as such: patent classes (3rd (β=-0.256, p<0.01, R2=0.169) and 4th (β=-0.255, p<0.01,
27
p<0.05, R2=0.181)). Furthermore, number of national classes become a significant variable (β=0.198,
28 Table 3, Regression of The Lego Group patent indicators in 4 models
Forward Citation (Year adjusted) (DV)
Model 1 Model 2 Model 3 Model 4
Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta
Constant -34.945* 17.920 -38.554* 18.061 -38.753* 18.146 -44.289* 19.080 Year Granted (CV) 0.018* .009 .223 .019* .009 .246 .019* .009 .248 .022* .010 .283 Time to Grant (CV) -0.032 .059 -.052 -.043 .059 -.069 -.049 .062 -.079 -.047 .062 -.075 Patent classes (CV) 0.013 .032 .038 .008 .032 .023 .008 .033 .024 .009 .033 .025 Attorneys (CV) 0.045 .033 .142 .041 .034 .131 .039 .034 .125 .038 .034 .120 National Classes (CV) 0.164*** .033 .459 .169*** .034 .474 .168*** .034 .471 .178*** .036 .499
Backward Foreign Citations (CV)
-0.009 .021 -.041 -.010 .022 -.047 -.009 .022 -.040 -.008 .022 -.036
Patents per year (CV) -0.008 .018 -.053 -.010 .018 -.065 -.010 .018 -.060 -.011 .018 -.071
Backward Citation Other Assignee (IV)
.135 .564 .047 .159 .570 .056 .150 .570 .052
Backward Citation Lego (IV)
-.285 .628 -.092 -.266 .633 -.086 -.357 .640 -.115
New Inventors (IV) -.024 .063 -.036 -.026 .063 -.039
Claims (IV) -.009 .009 -.095
R2 0.228 0.246 0.247 0.254
Note. Dependent Variable is Forward Citation by Other Assignees Adjusted for Year Difference. N=109. *p<0,05. **p<0,01. ***p≤0,001. DV=dependent
29 Table 4, Regression of The Lego Group patent indicators in 4 models
Forward Citation
(Proportion Of Total) (DV)
Model 1 Model 2 Model 3 Model 4
Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta Coefficient SE Beta
Constant -5.725 8.072 -9.070 7.810 -8.976 7.846 -11.949 8.229 Year Granted (CV) .003 .004 .102 .005 .004 .158 .005 .004 .156 .007 .004 .203 Time to Grant (CV) .018 .027 .071 .006 .026 .024 .009 .027 .035 .010 .027 .039 Patent classes (CV) -.031* .015 -.215 -.037** .014 -.256 -.037** .014 -.256 -.037** .014 -.255 Attorneys (CV) .008 .015 .059 .003 .015 .023 .004 .015 .030 .003 .015 .022 National Classes (CV) .016 .015 .111 .023 .015 .158 .024 .015 .161 .029* .015 .198
Backward Foreign Citations (CV)
.010 .010 .112 .007 .009 .079 .006 .010 .071 .007 .009 .076
Patents per year (CV) -.006 .008 -.086 -.008 .008 -.116 -.008 .008 -.122 -.009 .008 -.136
Backward Citation Other Assignee (IV)
-.161 .244 -.137 -.173 .246 -.147 -.178 .246 -.151
Backward Citation Lego (IV)
-.555* .271 -.435 -.564* .273 -.442 -.613* .276 -.480
New Inventors (IV) .011 .027 .042 .011 .027 .039
Claims (IV) -.005 .004 -.123
R2 0.076 0.167 0.169 0.181
Note. Dependent Variable is Proportion of Forward Citation by Other Assignees from the total Forward Citations. N=109. *p<0,05. **p<0,01. ***p≤0,001.
30
5.3. Results summary
This section will provide a brief summary of acquired results and will define which hypotheses were supported and which were refuted. Firstly, according to our finding the construction toy industry as a whole is still growing given the positive significant value of beta in period 1976-2014; it especially grew in the period from 1988-2000. Unlike the industries recent years, data for which didn’t yield any significant results, Lego’s patenting has been declining in the past 12 years, even when adjusted for time to grant.
According to the analysis of the variables which contribute to the value of a patent several interesting finding were made. Firstly, three of the control variables had a significant effect on the dependent variable, these include: year granted, number of classes and number of national classes. Following is a summary of which hypothesis were supported/refuted according to our findings:
Hypothesis 1: Backward self-citation by Lego would decrease the usefulness of patent, in other words,
as proportion of backward self-citation increases, forward citation should decrease. –
Supported
Hypothesis 2a: Backward citation by others would increase the usefulness of patent, in other words, as
proportion backward citation by others increases, forward citation should increase as
well. - Not supported
Hypothesis 2b: Backward citation by others would not affect the usefulness of patents. - Supported
Hypothesis 3: Number of claims will have a positive effect on usefulness of a patent, in other words, as
the number of claims increases, the number of forward citation should increase as well.
31
Hypothesis 4: Number of new inventors will have a positive effect on usefulness of a patent. In other
words, as the number of new inventors increases, the number of forward citation should
increase as well. – Not supported
6. Discussion
In this section we will discuss theoretical meaning of our results. Firstly, we will discuss industry as a whole and development of the top performing companies according to patenting history. Then, we will discuss performance of Lego and lastly we will discuss the findings from the regression analysis of patent indicators.
6.1. Construction Toy Industry
Several trends could be identified in the industry data. Firstly, there was a significant boom in the number of patents in the period 1988-2000. We hypothesize that this can be the result of several developments during that time period. Firstly, back in the ‘90s and to this day we are facing a growing population, meaning that the overall size of the market is increasing (World DataBank, 2014). Furthermore, this period mostly took takes place in the ‘90s, which is the period of the US well known economic boom. Economic prosperity of that time period increased parents’ ability to purchase more toys for their children, as a result giving the toy companies more revenue to be used for development of new toys and patenting. In addition to that, fall of the Soviet Union, opened new markets for the construction toy industries, which included most of Eastern Europe, Russia and its post-soviet republics. Equally important is the fact that the overall number of patents granted by USPTO has significantly grown in the 1990s and continues to grow to this day (USPTO, 2013). There are however also certain factors that disagree with this finding. Firstly, even though we are facing a growing population, the fertility rate has been falling since the end of the ‘60s, thus even though the population is increasing, it is increasing at a slower rate than it used to, meaning that market growth is decelerating (World DataBank, 2014). Furthermore, Canadian economist David Boom has pointed out
32
that unlike many other industries, toy industry is the one which should have suffered during the ‘90s period, because baby boomers at that time had reached the non-childbearing age (Matthews, n. d.). If we look at the other time segments, the time period 1976-1980 has a positive insignificant slope, while time period 2000-2012/14 has no or slightly negative insignificant slope; overall this finding are inconclusive. It can only be implied that, based on the insignificant slope and close study of graph section for period 2000-2014, that patenting rate of the industry is currently declining. This goes against our finding that overall patenting slope for the period 1976-2014 is positive. We explain it by the very low patenting rate in late 70s and early 80s and rapid increase in 90s. This created a sharp slope that has not been evened out yet by the recent developments in the industry. However, there is one more trend that is worth mentioning. If we study closely the recent data regarding time needed to grant a patent, it has largely increased from 1-2 years in most cases, to 2-5 years starting from year 2000. There is not enough data for us to make a well grounded conclusion based on that; however this may be a possible limitation of this study, which makes studying recent developments in the industry even more difficult.
Furthermore, we would like to mentions that correlation analysis did not show significant correlation between the filled and granted patents. However, if we closely study the graphical representation of the data, we could see that the granted patents line resembles the shape of the filled patents line with two to three years of time-lag.
In addition to the industry as whole, we looked at the companies that were in our data set. Firstly, we would like to mention that almost 60% of patents that we recorded for the industry were individual inventors that were not included in the analysis, due to their rarely significant contribution to the industries knowledge development. We believe this high percentage of private inventors has something to do with private inventors’ inability or unwillingness to identify the specific sub-classes in construction toy patenting class that their invention belongs to. Big construction toy companies, however, which have much larger experience in patenting, place their inventions into the general class,
33
possibly due to its universal applicability of their invention in a variety of construction toy sub-classes. The 10 largest patentees were identified (see Appendix, Fig 3), with Lego as the highest patenting construction toy patentee. In this category, Lego seems to patent the most during the period 1992 to 2005 and has relatively low patenting level since then onwards. This is also in line with our finding for
all Lego patents in all construction toy sub-classes (see Appendix, Fig 2). According to the graph, the
pick of Lego’s patenting was in the period 1994-2007. Out of the top 10 companies 7 are still operating: Lego, Mattel, Minds-I, Takara Co, K’NEX, Canine Genius and 90Degrees; the status of the other three companies is not available. However it turned out that Canine Genius is actually a producer of toys for animals, while 90Degrees is a construction company. Out of the remaining 5 construction toy companies, the two most well-known and well-performing are Mattel and Lego. However, the amount of patenting data for these 10 companies is relatively limited, making it impossible to draw valid conclusions.
If we were to determine the evolution stage of the industry, then according to our findings and the available theory on evolution of industry we would hypothesize that construction toy industry in currently in mature stage. This conclusion is based on two main findings: one, high rate of patenting in period 1988-2000, which can be identified as growth stage; two, relative decline in patenting in period 2000-2014, which signifies maturity stage, where product development is replaced with operational efficiency.
6.2. The Lego Group
Moving on to the analysis of Lego patenting history, which was scarcely mentioned in the previous section. Firstly, we carried out correlation analysis for Lego and industry as a whole. It showed that Lego granted patents correlate with industry granted patents and Lego filed patents, while Lego filed patents correlate with industry filed patents. This shows that the Lego follows similar evolution pattern as the industry as a whole. From the regression analysis of Lego we can see that in the period from 1976 to 2000 the patenting slope, even though insignificant, was positive and that
34
there is no significant direction in the development of Lego patenting for the period overall (1976-2014). However, staring from the year 2000, Lego has experience rapid decline in patenting according to analyses of both 2000-2012 and 2000-2014 periods. This to a certain extent agrees with our findings for the industry as a whole, which has zero/ slightly negative slope for the same period. This also supports our conclusion about construction toy industry evolution stage, in which we conclude that this industry is in mature stage, maybe even already approaching the decline stage, given the recent decline in Lego’s patenting.
From our analysis we can see that patenting was not related to the current success of Lego. It is possible that patenting in construction toy industry is used in order to prevent other companies from copying their products, rather than help the development of toys. Recent articles regarding Lego tell us that the possible reasons for their recent success can be the launch of a new animated movie The Lego
Movie and success of the Chima line in Asian market (The Lego Group Inc., 2013). In other words,
company’s success was a result of successful new market expansion and horizontal penetration. 6.3. Patent value indicators
According to our results, 4 characteristics of a patent have an effect on forward citation by other assignees, in other word on the value of patent. We will now explain the reason for presence/absence of the relationship for each variable.
Our first control variable, identified as year, by which we mean the year in which patent was granted, had a significant effect on the forward citation in the round of regressions with time adjusted forward citations. This shows that as time passes the value, which individual patent brings increases. This means that Lego is becoming better in inventing new products as time passes. In general this is in line with the assumption that “practice makes perfect” and that more experienced patentee may be more innovative. However, this line of thinking may be broken by innovative newcomers who think outside the box and bring in new knowledge and inventions.
35
Next control variable, that had a positive effect on the number of forward citations, is the number of different national classes (countries of origin of citing companies), which was significant in the first round of regressions, but not in the second. This is surprising because we would expect this factor to have an effect in both rounds, because as the number of forward citations increases, the possibility of having a greater variety of nationalities also increases, due to the simple probability theory. Thus, this variable requires further investigation.
Last control variable that had a significant negative effect on the value of patent is number of patent classes. This is surprising for us, because according to the previous research we would expect this variable to have a positive effect (Lerner, 1994). Possible explanation to this phenomenon can be the nature of this industry. It is possible that in construction toy industry the more specific is the finding the better it is. The more particularized patent will be easier to assign to one group, thus, the less number of classes a patent has the better. This theory however is not supported by our findings regarding the number of claims a patent has (dependent variable). According to the results, the relationship between claims and forward citation, even though negative, is not significant. If our theory were to be true, then the less claims a certain inventions has the better. Our finding are inconclusive regarding this matter, thus further research is required.
The next independent variable is the number of backward citations by Lego and other assignees. We found in the second round of regressions that backward citations by Lego indeed has a negative effect on the value of the patent, which supports our hypothesis. It also supports the findings of Lanjouw and Schankerman (2001) and Lanjouw and Schankerman (1999). However, this hypothesis was not supported by our first round of regression with time adjusted forward citation, thus even though hypothesis is supported by the first round, the strength of that relationship is questionable. We also found no effect of backward citation by other assignees on the value of patent, which is also in line with our hypothesis and supports findings and reasoning of Lanjouw & Schankerman (2004) and Harhoff et al. (2002).
