Analysis of patent portfolio and financial of information technology management

Using data of 479 firms from 1990 to 1997 based on the DTI-Scoreboard, patent data from the “EPO Worldwide Patent Statistical Database” PATSTAT and financial data from Standard & Poor's

ANALYSIS OF PATENT PORTFOLIO AND FINANCIAL PERFORMANCE OF FIRMS

In Partial Fulfillment of the Requirements of the Degree of

MASTER OF INFORMATION TECHNOLOGY MANAGEMENT

In Computer Science and Engineer

ANALYSIS OF PATENT PORTFOLIO AND FINANCIAL PERFORMANCE OF FIRMS

In Partial Fulfillment of the Requirements of the Degree of

MASTER OF INFORMATION TECHNOLOGY MANAGEMENT

In Computer Science and Engineer

By

Mr Vu Ba Quang ID: MITM03010International University - Vietnam National University HCMC

Acknowledgements

First of all, I would like to express my deepest gratitude to my advisor, Dr Nguyen Hong Quang for his support and guidance throughout the research His valuable advices led me to the right way to complete the thesis

During my time of studying at International University, I received lot of useful knowledge and sharing as well as guidance from my professors and good support from the Registrar Office Therefore, I would also like to thank them

Plagiarism Statements

I would like to declare that, apart from the acknowledged references, this thesis either does not use language, ideas, or other original material from anyone; or has not been previously submitted to any other educational and research programs or institutions I fully understand that any writings in this thesis contradicted to the above statement will automatically lead to the rejection from the MITM program at the International University – Vietnam National University Ho Chi Minh City

Table of Contents

Acknowledgements i

Plagiarism Statements ii

Copyright Statement iii

Table of Contents iv

List of Tables vii

List of Figures viii

Abstract ix

Chapter One - Introduction 1

1 Motivation 1

2 Problem Formulation 5

2.1 Unclear relationship between patent portfolio and its performance 6

2.2 Unclear relationship between patent portfolio and financial performance 7

2.3 Limited use of patent indicators to predict financial performance 8

3 Objectives 8

4 Scope 9

5 Thesis Structure 9

Chapter Two - Literature Review 10

1 Background 10

1.1 Patent and technology base of a company 10

1.2 Financial performance 13

1.3 Correlation test 16

2 Related Work 17

3 Comparative Analysis of Related Work 19

Chapter Three – Our Proposed Solutions 22

1 Overview 22

1.1 Methodology 22

1.2 Building patent portfolio database 23

1.2.1 USPTO Patent data 24

1.2.2 Patent database from UC Berkeley 25

1.2.3 Integrate financial data with patent portfolio data … 25

1.3 Patent portfolio indicators 28

2 Spearman correlation between patent portfolio and performance 30

2.1 Correlation calculation 31

2.2 Correlation in yearly lags 38

2.2.1 Ability to create new technology 38

2.2.2 Innovation history 39

2.2.3 Innovation rate 40

2.2.4 R&D force 40

2.2.5 Summary 41

3 Spearman correlation between patent portfolio and financial performance 41

3.1 Correlation calculation 41

3.2 Correlation in yearly lags 44

3.2.1 Ability to create new technology 45

3.2.2 Ability to recognize and acquire existing technology 46

3.2.3 Innovation history 47

3.2.4 Patent portfolio performance 47

3.2.5 R&D force 48

3.2.6 Technology protection 49

3.2.7 Summary 49

4 Artificial Neural Network to predict financial ratio 50

4.1 Demonstration of using patent data to predict financial ratios 50

4.1.1 Training 51

4.1.2 Network information 51

4.1.3 Prediction by Observed Chart… 54

4.1.4 Summary 55

Chapter Four - Conclusion 56

1 Summary 56

2 Future works 57

Appendix 58

Appendix A: List of companies in our dataset 58

Appendix B: Sample of patent assignment file 58

Appendix C: Correlation between patent indicators with patent performance 61

Appendix D: Correlation between patent indicators with financial performance 66

References 71

List of Tables

Table 1 Correlation result classification 16

Table 2 Methods used in recent researches and our study 19

Table 3 Indicators used in recent researches and our study 19

Table 4 Example of different names of one company 27

Table 5 Patent portfolio indicators 28

Table 6 Correlation result with 2 year lag 34

Table 7 Correlation between inventor indicators with citation indicators 36

Table 8 Correlation between claim indicators with citation indicators 37

Table 9 Correlation between patent portfolio and financial performance in 2 year lag 42

Table 10 Neural Network information 51

Table 11 MLP Model summary 53

Table 12 MLP Parameter estimates 53

List of Figures

Figure 1: Tangible versus intangible value of the S&P 500 companies 2

Figure 2: First page of patent “Method for node ranking in a linked” 11

Figure 3: Analysis steps 23

Figure 4: Our data preprocessing 24

Figure 5: Correlation between the numbers of new patent with citation indicators 38

Figure 6: Correlation between the patent age standard deviation with citation indicators 39

Figure 7: Correlation between patent growth rate with citation indicators 40

Figure 8: Correlation between the average of inventor indicators with citation indicators 41

Figure 9: Correlation between number of new patents and financial indicator 46

Figure 10: Correlation between number of purchased patents and financial indicator 46 Figure 11: Correlation between patent age standard deviation and financial indicator 47

Figure 12: Correlation between total citation and financial indicator 48

Figure 13: Correlation between number of inventor and financial indicator 48

Figure 14: Correlation between number of claims and financial indicator 49

Figure 15: Multi-layer perception network with EPS as output 52

Figure 16: Observed values versus predicted values 55

Abstract

Meaningful correlation between technological and financial performances is important to management of technology and innovation The technological performance of a firm could be represented by its patent portfolio since the patented inventions give their owners an exclusive right to exclude others from exploiting and commercializing them on the market, which highly influences the financial performance In this thesis, a new approach is proposed to analyze such a correlation between the technological and financial performances Our contributions are three-fold First, our approach proposes that four patent-portfolio indicators highly correlated to the technological performance of a firm include: patent age, patent claims, the number of inventors, and the number of patents newly applied for or purchased Second, these four indicators give a strong correlation with financial performance of a firm represented by price to earnings, earning per share, stock price

on the market and other four key financial indicators (liquidity, leverage, profitability, and valuation ratios) Third, our analysis takes into account the yearly lags of the technology-finance correlation that happen in reality Our proposed approach adopts Spearman correlation coefficient, artificial neuron network and financial ratio analysis We experimented on two kinds of datasets: (i) the technology datasets, including USPTO patents and UC-Berkeley patent datasets, and (ii) the financial datasets of NASDAQ, AMEX and NYSE stock markets Such datasets include 322,095 patents from 259 companies specialized in computer technologies in the 35-year period (1981 – 2013) Our research outcomes could benefit CEOs, investors and other stakeholders to design better R&D strategies for increasing their technology values or to find their investment opportunities

Chapter One - Introduction

Chapter 1 introduces the foundation of the thesis and provides a background and the current problems and authors’ objectives to solve them

1 Motivation

In this new era of technologies, modern companies are moving from

competing with others by decreasing prices or offering additional gifts to researching and innovating new products and services which can help them to exploit for

commercial advantage These products are results of practical application of

knowledge which is gained through research and development (R&D) activities Utilization of the knowledge during the research phase not only can be used to

introduce new products but also to improve existing ones or optimize processes to save time and cost The ultimate aim is to increase businesses’ values and market share

In technology industries such as computer hardware, software, and internet which are evolving day by day due to continuous technological advancements,

businesses must constantly revise their product offerings in order to meet the demands

of consumers and stay competitive R&D is one solution for them to achieve and maintain their competitive positions Taking mobile-phone industry as an example, since the first iPhone was introduced in the United State, the industry has been

changing significantly and our phones have been evolved from the one with small screen and keypad to a modern one with large and touchable screen The top

manufacturers like Apple, Samsung, LG, and Sony are investing their money in developing new technologies and products to conquer the market

Companies may possess many different types of assets including tangible and intangible assets Tangible assets can be real estate, office equipment, machines, cash

and account receivable These kinds of assets are quite easy to evaluate and often

included clearly on financial reports However, beside tangible assets, companies also

possess a majority of different intangible assets which have real values and are

important to their success These can be organizational ability, brand name, trade

mark, and patented technologies or processes

Figure 1 shows the increasing of percentage of intangible asset of companies

in S&P 500 Its value reached 85% at the beginning of this year, an all-time high for

the years covered by the firm’s research, which extends back to 1975 “Within the last

quarter century, the market value of the S&P 500 companies has deviated greatly

from their book value This ‘value gap’ indicates that physical and financial

accountable assets reflected on a company’s balance sheet comprises less than 20% of

the true value of the average firm” (James E Malackowski, personal communication,

June 15, 2010)

Figure 1: Tangible versus intangible value of the S&P 500 companies

(http://www.oceantomo.com/intellectual-capital-equity)

Not only increasing company assets, intangible assets also play an important role to strengthen owner’s competitiveness and are considered as an early indicator of stock price performance as Louis Basenese (2012), the founder of the Wall Street Daily, said “While earnings growth remains a reliable indicator, we’d be well served

to add patent filings to our repertoire, too It’s an even earlier indicator of stock price performance.”

In a flat world today, being the first mover to introduce innovative products or services to market is not enough to gain a competitive advantage Companies have to protect their ideas by applying for patents as a legal tool This tool gives the owners exclusive rights to solely exploit the patents or allow others to utilize by licensing

R&D activities indicate that a company is making effort to gain new

advantageous and to make profit, which eventually reflect in the financial

performance measures There have been already researches on patents as a measure for technology bases as well as on how to evaluate patent value However, research on how the patent portfolio and financial performance is not very large and the utilization

of this correlation is not popular neither

There are several studies have reported that there is a positive relationship between innovation and firm performance Hall et al (2005), focus on patent citations

to explore whether they can be considered as a measure the market value They estimate Tobin’s q equations on the ratios of R&D to assets stocks, patents to R&D, and citations to patents And they find that each ratio significantly impacts market value, with an extra citation per patent boosting market value by 3% Another study of Chang, Chen, and Huang (2012) calculated patent H index, current impact index (CII) and essential patent index (EPI) Then they use panel fixed-effect model to verify if these indices are positively associated with corporate performance which are

represented by market value, sales and Return on Equity (ROE) The empirical results

of the fixed effect model indicated that patent H index and EPI were positively

associated with its market value, sales and ROE That meant that the higher the patent

H index and EPI, the more was its market value, sales and ROE In contrast, CII was not positively associated with its market value, sales and ROE Using data of 479 firms from 1990 to 1997 based on the DTI-Scoreboard, patent data from the “EPO Worldwide Patent Statistical Database” (PATSTAT) and financial data from Standard

& Poor's COMPUSTAT Global and COMPUSTAT North America databases,

Neuhäusler, Frietsch, and Blind (2011) found that number of patent applications is not

a good predictor of firm performance while family size has a positive association with Tobin’s q and ROI and average number of forward citations seems to affect market value positively but not on ROI Beside the studies of researchers, there are many products to help users analyze patents data for their very specific purposes One of them is Patent Research and Analysis tool of Thomson Reutuers which provide powerful analysis and visualization tools to gain greater insight Another famous patent research and analysis platform is Patent iNSIGHT Pro which includes

advanced text mining algorithms to bring out those insights in minutes which would erstwhile take days for a researcher However, most of the tools focus their strength

on patent data analysis to view technology trends, generate patent map report or identify licensing, research or acquisition opportunities Little of them can provide an insight on how the patent portfolio correlates to financial figures or ratios

In this thesis, the authors try to analyze the relationships between patent portfolio and financial performance of firms to prove that such relationships do exists They will be the framework to build a prediction tools for normal users such as investors to utilize the patent data in their decision making process

2 Problem Formulation

There are a large number of factors can affect the value and competitive advantage of firms such as business strategy, human resources, market position, products and services,…Nowadays, people tend to focus on a relatively new factor which is innovative capacity because it allows the companies to implement new products or improve existing ones to meet new requirements from customers as well

as to adapt with new market change However, the successful completion of the innovation process alone is not enough to secure the benefits gained from R&D A firm has to think of how to prevent other competitors to enter its market or mimic its products and services In other to do that, it must have protection mechanism provided

by government which patent is one of the most important instruments

The thesis proposes to solve the problem of analyzing meaningful correlation between patent portfolio and financial performance by examining the relationship between indicators of number of patents, patent ages, inventors and patent protection (patent claims) and patent performance (patent citation) and then between those indicators and financial performance

To assess company performance, financial ratio analysis method has been conducted It is a method to analyze at company’s financial statements to gain an insight in financial position of a company in order to form the basis of all investment decisions If we find some relationships between patent information and financial ratios, it means that we can also predict the financial health in the future by using public patent data

The following analysis tries to answer the question of how far the result of R&D and the protection which patents bring can influence the financial performance and market value of a firm We will use patent data as a representative of technology

base and market protection of a firm Because the value of each patent is different, not only number of patents in portfolio but also other indicators such as number of

forward citations, number of inventors are employed in the model In other to

determine the health of a company, we will use stock price and financial ratios which measure its liquidity, leverage, profitability, and valuation ratios These ratios can help to shed a light on how a company is performing in relation to key measures of business success

The data analysis comprises 3 problems as following:

1 Unclear relationship between patent portfolio and its performance

2 Unclear relationship between patent portfolio and financial performance

3 Limited use of patent portfolio to predict financial performance

The technological performance of a firm could be represented by its patent portfolio since the patented inventions give their owners an exclusive right to exclude others from exploiting and commercializing them on the market, which highly

influences the financial performance

2.1 Unclear relationship between patent portfolio and its performance

Patent portfolio is the result of R&D activities of companies However, the

relationship between these indicators with patent performance is not very clear Knowing this relationship, the board of management can propose a strategic to

improve their portfolio performance and its value

Citation to prior art is an indicator of the importance of the prior art to

subsequent inventions The more citation a patent receives, the more significant it is measured In order to evaluate the efficiency of R&D activities, researchers and management board usually use patent data The innovation capabilities of companies are often measured by some indicators Basically, we have patent count as an outcome

of R&D performance However, this figure is somewhat noisy because not all patents have the same value or technology strength Some researchers suggest that among various indicators, patent citation is one of the better to demonstrate patent portfolio’s value and patent quality Because when applicants submit new patents, they and examiners must find and cite older patents which anticipate or be similar to the new inventions If we stand at the site of cited patents, these citations are forward citations

If a patent is highly cited (i.e cited in 5, 10, or more subsequent patents), then that patent is likely to contain an important technology which later patents are built on

Among patent indicators, there are some depends on R&D activities, R&D team or strategy of companies such as number of patents, patent age, number of inventors, and number of claims These indicators are dependent on companies and might not reflect the values of patent portfolio Solving this problem may give us a better understanding on how patent portfolio performance or value is related to the portfolio characteristics

2.2 Unclear relationship between patent portfolio and financial

performance Companies spend money in R&D activities which is transformed into

new products, processes, and services in the future The ultimate purpose of these activities is to gain more revenue and benefit and that is also what most investors want To choose a stock to invest, they normally consider price and valuation or evaluate financial health but little of them pay attention to the innovation aspect

The value and performance of firms depend on a various factors such as business strategy, human resources, market, products and services In addition to this, innovative ability is also very important because it allows firms to renew their

products and services to adapt with constantly changing market or to compete with other companies It is said that increased innovative capability can help to improve the

competitive ability and as a result, leads to an increasing in company revenue and value

Calculating the correlation coefficients between patent portfolio figures and financial ratios can provide us an insight to the relationship between the 2 sets of information Although the correlation does not prove that it is a cause-effect

relationship, knowing the trend of the patent portfolio figures which can be calculated using public data can help to predict the trend of financial performance

2.3 Limited use of patent indicators to predict financial performance

Solving problem 2 may give us the results that there are indicators have strong, medium, weak or no relationship with financial performance It means that we can use indicators which strong, medium and even weak indicators to build a model to predict future values of them If we can do that, investors now have additional tool to help them to make decision on choosing the right companies

For average investors, it can be a challenge to select the right stocks on the market to buy There are some characteristics which they can pay attention to such as business model, financial performance, dividend paid, and market trend (Alexander, Raznick, & Bedigian, 2012) …However, little of them analyze patent data to invest because the lack of an appropriate tool to give such information and help them to make decision

3 Objectives

The aim of our study is to solve above 3 problems We will focus on

information technology firms which are listed on NASDAQ and NYSE stock markets

of USA because in the new “knowledge economy” era, they are among the fastest and the most innovative companies and the way this industry has changed over the last half century Specifically, we test for the relationship among patent portfolio variables

such as patent count, patent citations, patent age, citation age, and inventor and firm financial positions

4 Scope

The scope of this thesis focuses on analysis of patent portfolio and financial performance of firms First, we analyze the correlation of patent count, patent age, patent claims, and inventor figures to the patent performance which can be indicated

by citations Then we continue to analyze the correlation of those patent portfolio indicators with financial ratios including liquidity, leverage, profitability and valuation ratios However, the analysis does not aim to explain a cause-and-effect relationship between them because patent portfolio may not have a direct influence on financial performance and we need further tools and analysis to explore it

Chapter 3 is where we discuss our solutions and results when we solve problem 1, problem 2, and problem 3

Finally, chapter 7 presents the conclusion of my thesis report and proposes future work

Chapter Two - Literature Review

1 Background

1.1 Patent and technology base of a company A patent, firstly, is a legal

tool having technical claim(s) which describes a technical invention such as a new device, process, system, or method According to The United States Patent and

Trademark Office (USPTO), an agency of the U.S Department of Commerce, patent for an invention is the grant of a property right to the inventor and the right conferred

by the patent grant is, in the language of the statute and of the grant itself, “the right to exclude others from making, using, offering for sale, or selling” the invention in the United States or “importing” the invention into the United States

In order to register for a patent, inventor(s) normally start with filing an application The application should include description of the invention, the

implementation, and a collection of claims with the inventor(s) want to have The claims, which are one of the most important parts of patent, define, in technical terms, the extent, i.e the scope, of the protection conferred by a patent This application then

is filed in the country where inventor(s) apply patent After that, they have one year decide whether they want to expand the application internationally (it is called patent family) At the lasted 18 months after first filing, the patent application can be

published It means that it is widely and freely accessible by anyone

Figure 2 is an example of a patent named “METHOD FOR NODE RANKING

IN A LINKED DATABASE” In the first page, there are patent number, publication date, name, inventor(s), assignee(s), prior patents which it cites and the abstract

Figure 2: First page of patent “Method for node ranking in a linked”

The patent application and patent publication include a header with show

name and address of the inventor(s), the assignee(s), the country of origin, filing date

and the state of the art citations (citations which the inventor(s) or the of lawyer of Patent Office make) In addition to this, there are the details of innovation so everyone can access for free All the patent data make up a very rich resource to study about company technology development, technology strategic as well as result or R & D Moreover, the data are free and available for being analyzed

The product cycle model places a foundation for the idea that technology can drive the long-term development of market shares This theory assumes a variety of ability to exploit new technologies among different entities (Dosi, Pavitt, & Soete, 1991) In addition, it implies that a follower will need time and costs to imitate and absorb new technology to apply for his products or services These conditions mean that innovative products will make monopoly of the market in a period of time before the followers can catch up Consequently, firms developing new products or services using superior technology can take a large share of the market and gain more benefits than others To protect themselves from being imitated by other competitors, firms usually public their technologies to apply for patents Therefore, patent is one of the most important intangible assets which can be related to financial performance

This thesis aims to find out which of these indicators are related to financial ratios and are applicable for the evaluation of a company value Although total patents

in a portfolio is direct result of R&D activities, not all patents have the same

economic or technological value so only patent counts does not give us an accurate view on firm’s technological basis Therefore, many other indicators have been calculated and proposed to asset many aspects of patent portfolio Of course not all of them have the same impacts on financial performance, some may have strong impact, some may have slight impact or not at all

1.2 Financial performance When people started to have share markets to

buy and sell securities of listed or unlisted companies, they also began to assess company performance by analyzing financial ratio

Ratio analysis is one of the most popular and widely used tools of financial analysis A ratio is a relation between two quantities Although it is simple to

calculate a ratio, it may be complex to interpret the outcome To be meaningful, a ratio must refer to an economically important relation For example, there is a direct and crucial relation between an item’s sales price and its cost Accordingly, the ratio

of cost of goods sold to sales is important

Analysis of financial ratios can help stakeholder like creditors, investors, regulator, or manager to find out the financial soundness of an organization For example, CEOs may look into financial ratio reports to get clues for their strategic changes in business investment or financial activities They also analyze competitors

to evaluate profitability and risk

D’Amato (2010) proposed top 15 financial ratios for investors to consider Below are selected ratios to be used for the analysis:

Liquidity ratio: liquidity ratios indicate whether a company has the ability to

pay off short-term debt obligations (debts due to be paid within one year) as they fall due Generally, a higher value is desired as this indicates greater capacity to meet debt obligations

• f1 = Current Ratio: The Current ratio measures a company’s ability to repay short-term liabilities such as accounts payable and current debt using short-term assets such as cash, inventory and receivables Another way to look at it would be the value of a company’s current assets that will be converted to cash over the next

twelve months compared to the value of liabilities that will mature over the same period

  

• f2 = Cash balance to total liabilities (CBTL): this ratio shows a company’s cash balance in relation to its total liabilities Cash is the most liquid asset a business has A negative cash balance (caused by overdrafts) raises a warning signal and failure to address such an issue will likely result in liquidity problems

Lower risk firms typically have a higher value CBTL, because they have more cash that can be used to pay suppliers, banks or any other party that has provided the company with a product or service Higher risk companies typically have a lower value CBTL, which means the company’s ability to meet its debt obligations is

significantly hampered

 =

Leverage ratio: leverage ratios, also referred to as gearing ratios, measure the

extent to which a company utilizes debt to finance growth Leverage ratios can

provide an indication of a company’s long-term solvency Whilst most financial experts will acknowledge that debt is a cheaper form of financing than equity, debt carries risks and investors need to be aware of the extent of this risk

• f3 = Debt to equity ratio (DE ratio): The debt to equity ratio provides an indication of a company’s capital structure and whether the company is more reliant

on borrowings (debt) or shareholder capital (equity) to fund assets and activities

Profitability ratio: this type of ratio measures a company’s performance and

provide an indication of its ability to generate profits As profits are used to fund

business development and pay dividends to shareholders, a company’s profitability and how efficient it is at generating profits is an important consideration for shareholders

• f4 = Earnings per share (EPS): EPS ratio measures earnings in relation

to every share on issue It is calculated by dividing the company’s net income by the number of shares on issue

 =

• f5 = Gross profit margin: this ratio tell us what percentage of a company’s sales revenue would remain after deducting the cost of goods sold This is important as it helps to determine whether the company would still have enough funds

to cover operating expenses such as employee benefits, lease payments, advertising, and so forth

Valuation ratio: Ratios belong to this group are used to figure whether the

current share relation to its true value Valuation ratios also help us assess if a company is cheap or expensive relative to earnings, growth prospects and dividend distributions

• f6 = Price to earnings ratio (PE): the price to earnings per share is a valuation ratio of a company's current share price compared to its per-share earnings

1.3 Correlation test Correlation analysis is often utilized to find out the

relationship between 2 variables X and Y It indicates the extent which 2 variable fluctuate together If variable X and variable Y increase or decrease in parallel, we have a positive correlation If variable X increases and variable Y decreases inversely,

we have a negative correlation

Correlation coefficients can range from -1.00 to +1.00 Value -1.00 represents perfect negative correlations while value +1.00 represents a perfect positive

correlation The closer the coefficients are to +1.00 and -1.00, the greater the strength

of the relationship between variables is

We use Spearman’s correlation coefficient in this thesis The formula used to calculate its value is as following (Lovie, 1995):

,= 1 − (6 ∑ 0− 1)0

Where:

Σd2: the sum of the squared differences between the pairs of ranks

n: the number of pairs

In general, the higher the correlation coefficient is, the stronger the

relationship is The following tables present classification of values of correlation coefficients (Dancey & Reidy, 2004)

Table 1 Correlation result classification Value of the Correlation Coefficient Strength of Correlation

0.7 <= | rs | < 1 Strong

0.4 <= | rs | < 0.7 Moderate

0.1 <= | rs | < 0.4 Weak

Beside the rs value, we should consider p-value which is used to measure the

“significance” of the empirical analyses The calculation of the p-value is based on a number of assumptions that are beyond the scope of this discussion, but people who need p-values can simply look them up in standard statistical tables The p-value is a number between 0 and 1 representing the probability that this data would have arisen

if the null hypothesis were true

A low p-value (such as 0.01) is taken as evidence that the null hypothesis can

be “rejected” Statisticians say that a p-value of 0.01 is “highly significant” or say that

“the data is significant at the 0.01 level” Null hypothesis is denoted by H0

H0: there is no correlation in the population

H1: there is correlation in the population

2 Related Work

There research on patent data and corporation performance to estimate market value base on patent indicator or find the relationship between 2 datasets is popular nowadays Some authors related stock market value of firms to some measures of patent data and R&D activities They found that the patent count does not reflect the true knowledge capital value of a company so in later studies, they incorporate more indicators of citation, family size, inventors or other indices

Hall, B H, Jaffe, A., Trajtenberg, M (2005) estimated Tobin’s q equations on the ratios of R&D to assets stocks, patents to R&D, and citations to patents using Patents and citations data from USPTO and financial data from Compustat The authors found that that each ratio significantly impacts market value, with an extra

citation per patent boosting market value by 3%, “unpredictable” citations have a stronger effect than the predictable portion, and that self-citations are more valuable than external citations

Another interesting study was conducted by Neuhäusler, P., Frietsch, R., Schubert, T., & Blind, K (2011) to analyze the relationship between patents and the financial performance of firms using correlation test and fixed effects panel regression model Their data include 479 firms from 1990 to 2007 of the DTI-Scoreboard which was provided by the British Department for Innovation, Universities & Skills (DIUS) and the Department for Business, Enterprise & Regulatory Reform (BERR) The relevant patent data were extracted from the "EPO Worldwide Pa-tent Statistical Database" (PATSTAT), which provides information about published patents collected from 81 patent authorities worldwide Neuhäusler et al introduced hypothesis which describe the correlation between patent portfolio indicators like number of patent, number of citation, family size and average number of inventors and firm

performance Their empirical result supports almost hypotheses of the authors

According to those, the number of patents, number of forward citations, and family size have positive effect while backward citation has negative effect on financial performance

In 2012, Chang et al calculated the Patent H index, Current impact index (CII), and Essential patent index (EPI) and used fixed effect model to verify

association of these indices with corporation performance Their results indicated that patent H index and EPI were positively associated with its market value, sales and ROE That meant that the higher the patent H index and EPI, the more was its market value, sales and ROE However, CII was not positively associated with its market value, sales and ROE

3 Comparative Analysis of Related Work

Table 2 summaries the methods used in 3 researches which we review and ours In this thesis, we use correlation test to find the correlation coefficients between

2 datasets and ANN to explore the probability to predict financial performance base

on patent portfolio indicators

Table 2 Methods used in recent researches and our study

Hall et al Chang et al Neuhäusler et al Our study

Method • Non-linear

least squares

• Fixed effects panel regression model

• Spearman correlation

• Fixed effects panel regression model

• Spearman correlation

• Artificial Neural Network

In addition to the difference in analysis methods, the authors used a variety of patent and finance indicators In our study, we use some indicators which are not presented in previous researches Table 3 shows us more details about that

Table 3 Indicators used in recent researches and our study

Hall et al Chang et

al

Neuhäusler

et al

Our research

Patent portfolio or innovation indicators

R&D investment / book value of

Number of citation / Number of

Essential patent index (EPI) x

Number of new purchased

patents

x

Overall/ internal/external

citation growth rate

x

Average number of claims per

patent

x

Standard deviation of claims x

Financial performance indicators

Chapter Three – Our Proposed Solutions

1 Overview

1.1 Methodology This research is conducted in the United State technology

industry The data to be used in this thesis are collected from 259 companies which are listed on NASDAQ, NYSE and AMEX stock markets The full list of these

companies can be found in table 1 of the Appendix A Patent data are from UPSTO weekly patent releases which are hosted by Google Our financial data are from a company engaged in the business of financial news, GuruFocus, LLC These data span the period from 1999 to 2013 In this study, Spearman correlation coefficient and artificial neural network are employed to explore the linear and nonlinear

relationships of patents and financial performance

Techniques used in this analysis:

• Correlation test: help us to explore the bivariate relationship of patent and finance

• Artificial Neural Network: used to explore the multivariate relationship of patent portfolio and finance

Softwares:

• Visual Studio 2012: used to build tools to process data and visualize the patent portfolio and finance indicators

• IBM SPSS Statistics 20: used to conduct correlation test and ANN

• Microsoft SQL Server 2012: store data for the analysis Our steps to solve the 3 problems:

Figure 3: Analysis steps

1.2 Building patent portfolio database The patent data are taken from 2

main sources: one is the UPSTO weekly patent releases which are hosted by Google

and another is research result of Balsmeier et al (2014) at UC Berkeley Raw data

from Google are downloaded, parsed and inserted to MS SQL database Financial

data are extracted data from GuruFocus, LLC, a Better Business Bureau (BBB)

accredited company with an A+ rating The data are verified again U.S Securities and

Exchange Commission data randomly to make sure they are correct and can be used

for our research

In this study, we want to test the relationship between patent portfolio and

financial performance; therefore we must have a link of the 2 datasets A name of a

company in patent data may differ from name of the same company in financial data

To solve this issue, we use the tool called FRIL (Fine‐grained Records Integration and

Linkage Tool) (http://fril.sourceforge.net) Financial data are combined with patent

data in the same SQL database and our analysis is performed on this

Data

preprocessing

Spearman correlation between patent portfolio and performance

Spearman correlation between patent portfolio and financial performance

The use of patent data

to predict financial ratios

Conclusion

Figure 4: Our data preprocessing

1.2.1 USPTO Patent data The United States Patent and Trademark Office

(USPTO) is the federal agency for granting U.S patents and registering trademarks The USPTO advises the president of the United States, the secretary of commerce, and U.S government agencies on intellectual property (IP) policy, protection, and

enforcement; and promotes the stronger and more effective IP protection around the world On 2 June 2010, USPTO and Google have entered an agreement to allow

Google to provide bulk patent and trademark data to the public The agreement

requires Google to host the data at no charge and without modification to the public Data is being provided in weekly segments, at least for the last ten years or so Older data (pre-1996) is offered in yearly files

USPTO patent products are available on Google:

• Patent Grants

• Patent Application Publications

• Additional Patent Data

UPSTO Pats

data

(XML format)

Pats database Fung Institute

Financial data of USA companies from NASDAQ and NYSE

Extract patent

assignment

information

Consolidate data and calculate indicator

Consolidated Patent and Finance database

In this thesis, we take patent assignment text data in XML format from Jan

1982 to end of 2013 to extract patent purchasing information of companies in our analysis The patent assignment data include the first assignment for the original assignee and other assignments such as patent purchase, security to apply for a mortgage, patent purchase and patent sale In order to build patent portfolio for selected firms, we should parse all assignments and keep only re-assignment records (patent purchase and patent sale information)

Patent assignment file format is described in patent assignment daily XML file description which can be found in Appendix B

1.2.2 Patent database from UC Berkeley Patent data has been used vastly for

over half a century to study invention and innovation as well as technology growth (Hall et al., 2012) However, patent data is not organized in a well-structured way because entities (inventors, assignors, assignees, application law firm, and location) are maintained by their names on patent document and there is no unique identifier from the patent office even in one country This barrier makes each researcher has to spend a large amount of time and resources on the manual job to disambiguate these information

The database is taken from UC, Berkeley using a Patent Database Search tool (rosencrantz.berkeley.edu/batchsql/downloads) We can select a small set submitting

as a job together with our emails and the tool will send us a link to get data when our job is completed on the server

The raw data are imported to SQL server and then filtered to get only data of our companies in the analysis

1.2.3 Integrate financial data with patent portfolio data Because financial

data and patent portfolio data come from different sources and the firms are recorded

by their names, looking at these fields as they appear on both datasets reveals various forms of misspellings or correct but different names This issue makes a barrier for an easy integration from us Let’s take Google, in financial data, its name is Google Incorporation However, in patent portfolio data, its name can be AGGOGLE INC,

GOGGLE INC, OOGLES INC, or GOOLGE INC…

In order to match the company names between 2 datasets, we have to calculate the “distance metrics” to estimate the similarity between values In the paper, we choose the “edit distance” as our “distance metric” which was proposed by Jurczyk et

al (2008) In this approach, input values are treated as raw string and the distance is the cost of best sequence of edit operations need to be applied to the first string so that

it is converted to the second string Typical edit operations are character insertion, deletion, substitution, and or switching two characters that are next to each other Then we choose a approve and disapprove threshold to calculate this metric The 2 parameters should belong to range [0…1] and approve level should be greater than disapprove level The score of edit distance function, when it is calculated, also has value in range of [0…1] Below is the formula for edit distance function

d: disapprove level length(strA), length(strB): the lengths of strA and strB accordingly

Before running the function between each pair of company names, we truncate the common words in name to avoid them to affect the score and make wrong result

For example, score(“ABC Corporation”, “CCC Corporation”) seems to be high

because of the word Corporation appears in 2 names Some common words need to be truncate are: “Technologies”/” Technology” to “ TECH”, “International” to “INT”,

“Semicondutor”/”Semiconductor” to “SEMI”, “Communications” to “COMM”,

“Holding” to “HOLD”, “Limited” to “LTD”, “Incorporated” to “INC”, and

Company name in patent data

Company name in financial data

/ADANVED MICRO DEVICES, INC

Advanced Micro Devices,

/ADAVANCED MICRO DEVICES INC

Advanced Micro Devices,

Low confidence but correct

1.3 Patent portfolio indicators Table 5 lists the indicators which we used in

this analysis:

Table 5 Patent portfolio indicators

Innovation ability

of firm

p2 Number of

purchased patents

New Count the newly

purchased patents per year

Ability to recognize and acquire existing technology p3 Number of sold

patent

New Count the sold patents

per year

Technology transference to outside p4 Patent growth rate New Number of new patents

divided by average new patents of previous 5 years

Innovation rate of a firm

New Standard deviation of

all patent age in portfolio

The spread of patent age in portfolio

Citation p7 Total number of new

citations

Reused Count number of

citations which the

As citations represent the extent

Spearman correlation between patent portfolio and financial performance

The use of patent data

to predict financial ratios... server and then filtered to get only data of our companies in the analysis

1.2.3 Integrate financial data with patent portfolio data Because financial

data and patent portfolio. ..

Innovation rate of a firm

New Standard deviation of

all patent age in portfolio

The spread of patent age in portfolio

Citation p7 Total number of new

citations