The impact of R&D spending on stock performance in the U.S.

(1)

The impact of R&D spending on

stock performance in the U.S.

Bachelor thesis

Mick van Bemmel

10763740

Universiteit van Amsterdam

Faculty Economics and Business Finance

(2)

Statement of Originality

This document is written by student Mick van Bemmel who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the content

(3)

Abstract

This research aimed to identify the impact of R&D on stock performance in the U.S. in the period from January 1990 till December 2019. We have done this by constructing 4 different portfolios based on different measure of R&D. The market proxy we have used was the S&P500 in the same period, from January 1990 till December 2019. The hypothesis 1 till 4 are tested using the Fama-French model and the Jensen’s index to make an OLS regression with a significance of 5%. Based on a quantitative and qualitive analysis of the 4 different and self-constructed portfolios that were based on innovation, it can be concluded that we have found significant evidence that investing in portfolios based on R&D stock performance will have a positive and significant effect on the stock returns compared with the market. We also found in previous research that the competitive landscape has changed since the technological revolution and we wanted to test this to get a better understanding of the changes in the importance of innovation and thus in R&D. We tested this by performing a Chow test (in our case Wald because the error term was not constant) and we can conclude that there has been a structural break in alpha somewhere in between the period 12-31-1996 until 07-31-1998.

(4)

1. Introduction

The need of innovation has never been bigger. In today’s world companies have to deal with an increasing amount of competitors and decreasing product life cycles. It is now more important than ever that firms have a continuous stream of innovations that will improve profitability and create or maintain a competitive advantage (Artz, Norman et al., 2010).

This is due to the fact that the competitive landscape has changed since the

“technological revolution”. This revolution is rapidly altering the nature of competition and strategy (Betis and Hitt, 1995). Managers are faced with new technologies and new problems but also with new possibilities and new ways to earn a potential profits and new tools to make this happen.

Many studies have been carried out regarding the relationship between firm

innovativeness and firm value, noting its effects on market and financial position. Evidence has been found showing that innovation has direct positive effects on financial position, firm value and thereby stock performance (Ruberta and Kirca, 2012). Empirical research also shows that spending in research and development (R&D) increases firm value and new product development (NPD). Subsequently research shows a positive relation between NPD and an increased firm performance. (Song, Im et al.,2011)

The potential of finding new ways to create, change, renew and improve is unlimited. Innovation is one of the most important factors for companies. Through innovation,

companies can research and acquire information that has the potential to benefit them in a range of ways (Yanadori and Cui, 2013). Innovation can lead to competitive advantages for companies, and can benefit whole societies (Wetter, 2011).

The objective of this paper is to acquire a better understanding of how innovation will impact stock returns in the past 30 years in the U.S. with the S&P500 as proxy. Based on previous research, the expectation of this paper is that the portfolio with the most innovative firms has a higher return than the market and therefore will beat the market.

In addition this paper also expects that that the impact of innovation will have changed in the last 30 years due to technological changes and thus innovation now will have a more substantial impact on stock returns than 30 years ago. We will examine the degree of innovation in firms and the different measures of innovation.

(6)

To find the difference between innovative firms and non-innovative firms we will make 4 different portfolios where companies are being ranked on the degree of innovation. We have made calculations with 500 companies over 30 years. We have made calculations at several measures including revenues, historical R&D expenditures, and market capitalization.

Subsequently we will examine if the most innovative companies are also the ones with the highest stock returns using different innovation degree variables. We will do this

regressing the 4 different portfolios against a proxy by using the OLS method. To test this the Jensen’s index and the Fama-French model are used. Jensen’s alpha denotes the difference between the expected portfolio return and the actual portfolio return at a given level of risk. The Jensen’s alpha is positive if the actual portfolio return exceeds the expected return (Miljan, 2017). So if alpha is positive it outperforms the market because the expected returns are equal to the market with its corresponding risk.

This means that the Jensen’s alpha estimation is important to investors. (Phuoc, 2018) In particular, we will investigate if a positive Jensen’s alpha will be present in the most innovative companies and a negative Jensen’s alpha will be present in the least innovative companies. When a positive Jensen’s alpha is present this means that companies have an abnormal amount of return in comparison with the market. By comparing the most innovative companies in the S&P500 with the S&P500 as a whole, the research will reveal if a positive Jensen’s alpha is present and thus abnormal excess returns.

Furthermore, we will also examine if the impact of innovation on stock returns has changed in the last 30 years due to the technological changes in the company landscape. Lastly we will see if there are structural break in the data and what could have caused these breaks to appear. All the used data for the portfolios is collected from WRDS historical data in the period January 1990 till December 2019.

The subsequent chapter, chapter 2, will be the theoretical review where the relevant literature, definitions, concepts and typologies will be discussed. Besides that some previous research about innovation and firm performance will be discussed to a further degree. In chapter 3, the methodology with the research design, relevant variables and the datasets will be discussed. Following that chapter 4 will explain the results that are found in this paper. Chapter 5 will be a discussion about the results and the possible biases in this paper or in previous research that is used for this paper. Then this paper is rounded up with a conclusion in chapter 6.

(7)

The added scientific value of this research will be that the direct relationship between R&D spending and stock returns is being examined. Most previous literature is based on a more broad understanding of performance, like firm performance. The link between R&D or innovation and the productivity factor, patents, firm performance and profitability is proven many times in previous research. These factors should all contribute the stock performance and we want to know if R&D can actually create abnormal returns. The research method used has been done many times in previous research but most often on sustainable firms but never on innovative firms.

2. Theoretical review and hypothesis

2.1 Innovation and R&D

Knowledge in the form of science and technology is viewed as a key driver for growth and success for firms and this has even become more important in today’s world (Wetter, 2011; Artz, Norman et al., 2010; Betis and Hitt, 1995). The capabilities of finding or creating new internal relevant knowledge in firms has been crucial in succeeding in finding new products or processes and can even be seen as a predictor of the firm’s future ability to create inventive outputs (Chiesa, 1996; Cardinal and Hatfield, 2000). One of the most important measures of innovation in firms is research and development (R&D). It refers to innovative activities undertaken by corporations or governments in developing or researching new products, services or improving existing ones (Helfat and Peteraf, 2003).

R&D spending and its outcomes have been explored in many empirical studies by renown economists (e.g., Schumpeter 1934, Mansfield 1962). They found in general that R&D spending had a positive outcome on firm performance. For example Mansfield (1980) found that when R&D spending kept constant, there was an increase of the total productivity factor. This was the first systematic evidence that was found for a positive relationship between R&D spending and a firm’s ability to increase productivity growth. This has been proved again in 2000 by Cardinal and Hatfield (2000) in the pharmaceutical industry.

Another finding of R&D spending is that it has a positive relationship the number of newly created patents. There has also been a lot of research about R&D spending and acquiring new patents. A statistically significant relationship has been found between the fluctuations in R&D spending within a firm and the current and subsequent changes in its

(8)

level of acquiring new patents. The relationship between R&D spending is especially strong on cross-sectional levels where it reflects differences between firms (Pakes and

Griliches, 1984). This positive relationship between patents and R&D spending has also been

found in the computer industries byHagedoorn and Duysters (2002) and a year earlier in the

chemical industry by Ahuja and Katila (2001).

In addition, R&D spending can have a positive outcome on firm performance. Artz, Kendall et. Al (2010) have done a longitudinal study about the impact of R&D spending on outputs in terms of inventions and innovations and the impact of those outputs on firm performance. They had several expected but also some surprising findings. For example they found out that R&D spending is a good predictor of new product announcements and patents but not always in a positive way. The finding that R&D spending has a positive effect on returns to scale was also a finding that was not directly thought of. They also found that patents can have a negative relationship on firm performance. And expected R&D spending was positively correlated with both product announcements and patents as the previous research mentioned already stated.

Artz, Kendall et. al (2010) state that there is a difference between the size of the firms and the outcome of R&D spending. They found that smaller firms usually have lower levels of R&D spending but the smaller firms can have an advantage in R&D productivity relative to bigger firms. This has probably to do with the fact that smaller firms can make decisions in a faster and more effective way compared to bigger firms with a lot of hierarchy. At the other end of the spectrum can firms who have relatively high levels of R&D spending who take advantage of the economies of scope to spread costs. Bigger firms can also sustain a more sustainable diverse portfolio of new research projects findings in different markets and can also excel in their field because of the newly extra information they have gathered.

Findings of the research were also that product announcements are positively related to a firm’s performance and this suggests that firms who can deliver a stable flow of newly created products can generate higher levels of performance. Higher levels of product announcements can also act in a way as marketing because consumers find it exciting when new products are released and they will have conversations about this. The relationship between product announcements and sales growth has been found in this study hence leads to higher profitability. Which confirms the study done by Hua and Wemmerlöv (2006) which states that product change frequency is linked to market growth and thereby firm

(9)

Besides having a positive outcome on the productivity factor, patents and firm performance innovation also has a positive outcome on profitability. The intention of

innovative action influences the extent to which abnormal profit outcomes persist over time. This can be explained because high intensive innovative propensity yields a series of

temporary monopolistic positions at the product level. A monopolistic position can be translated into persistent abnormal profitability. However the length of the period where the abnormal profits will occur decreases very quickly over time (Roberts, 1999). This has to do

with the finite period of product patents and imitations. HoweverCho and Pucik (2005) has

found that innovation alone is not sufficient to increase a firm’s profitability. They claim that only a combination of quality and innovation can cause a firm to grow and to increase

profitability. And neither of these variables individually can make a statistically significant difference in growth or in profitability.

2.2 Innovation and stock performance

The relationship between innovation and stock returns is measured in a event study by Sood an Tellis (2009). They have concluded that the market returns of an innovation project are more than 13 times higher than an average innovation event. They found out that returns on negative events are higher in absolute value than innovation events that are positive and that the returns to initiation occur 4.7 years ahead of launch. Furthermore returns are higher for small firms than for bigger firms and the returns from the announcing firms are

substantially bigger than those to competitors in the same markets and this seems in line with more recent research. Firms introducing new products realise significant abnormal returns gains following the announcement, however product announcements that receive neutral coverage in the financial press do not realize these significant abnormal returns (Doukas, Guo et. al. 2016).

The benefits of innovation and R&D seem very consistent in the research that is discussed above. Therefor we expect that R&D and innovation will have a positive impact on stock returns. The value that this research will add to previous corresponding research is the impact of R&D in specific on stock returns. And the research question of this research is if spending on R&D will substantially increase stock returns.

(10)

3. Research design and data

3.1 Innovation portfolios

We first need to select the most innovative companies within the S&P500 and then calculate the monthly average returns of this portfolio. To distinguish the difference between the innovative companies and the non-innovative companies, we will make use of variables that measure innovation. We have chosen to use the accumulated R&D expenses in the last 30 years, annual R&D expenses and the annual R&D intensity. R&D expenses are all the costs that occur for research and development and they are the most significant factor in

improvement and new discoveries. R&D intensity is the amount of total R&D spending divided by its total revenues in the corresponding period. All the companies of the S&P500 were examined regarding the R&D expenditures and revenues gathered from WRDS

historical data in the period from January 1990 until January 2019.

Because the data was from 30 years and from 500 companies, we ended up with more than 15000 variables for annual R&D expenditures, annual revenues and R&D intensity. To make the data more manageable the average was calculated for every firm. First the number of years a company was active was looked at, then the annual R&D expenditures and the annual revenues were accumulated. After this, the accumulated results were divided by the number of years the companies were active. For the calculations of the R&D intensity the accumulated R&D expenses were divided by the accumulated Revenues. For 257 of the companies there were no R&D expenses available these are not included. In table 1 the descriptive statistics of these 243 companies are displayed.

Table 1, descriptive statistics of innovation measures.

Variable N Mean Min Max

Years 243 25.3 2 30 CURV 243 $400,728,900 $2,626,188 $7,044,971,000 AVRV 243 $15,283,340 $117,077 $234,832,400 CURD 243 $17,951,880 $1780 $211,204,400 AVRD 243 $694,134 $59.33 $7,446,809 AVRDI 243 7.42% 0.007% 76.467%

The variables years (years a company was active), CURV (accumulated revenues of years), AVRV (average revenue per year), CURD (accumulated R&D expenses of years), AVRD(average R&D expenses per year), and AVRDI (average R&D intensity per year) were used.

(11)

Since this research now has the data to select innovative and non-innovative companies 4 portfolios can be constructed with the innovative companies. We will make 4 portfolios of the 243 companies. First we will construct a portfolio with the cumulative R&D spending as the indicator of innovation. This is an interesting indicator that shows the total amount of R&D expense over the last 30 years. This is also the most broad indicator because the time period of expenses and the amount per year are not specified with this indicator. Then we will construct a portfolio with the average R&D spending as the indicator of innovation. This is an interesting indicator that shows the average annual amount of R&D expense over the last 30 years. This is a narrower indicator because it shows what a company spends on average on R&D per year. In addition we will construct a portfolio with R&D intensity as the indicator of innovation. This is the narrowest indicator because it shows annual R&D expenses as the percentage of annual revenues. And lastly a portfolio will be constructed looking at cumulative R&D spending, average R&D spending and the R&D intensity. This portfolio will be constructed by ranking all the above indicators and then adding the ranks per company which will give us an overall picture regarding the different indicators.

From all the portfolio’s the companies from the 75th percentile will be labeled

innovative. These portfolio’s consist of 243 companies and therefore the top 25% will be 60 companies. These portfolios will have a higher variance than the S&P500 because the

S&P500 has 500 companies in its portfolio. More companies in a portfolio will make sure that the unsystematic risk will decrease more than with 60 companies in a portfolio. The portfolio is equally weighted to minimize the variance. The different portfolios including company names and the values where the the ranking is based on can be found in appendix 5 for the CURD portfolio, appendix 6 for the AVRD portfolio, appendix 7 for the AVRDI portfolio and appendix 8 for the OVER portfolio.

3.2 Sample selection

To determine if companies with high innovation have abnormal returns, this research includes data from January 1990 until December 2019 which equals a 30 year period. This time frame has been chosen because the time between the discovery of an innovation and the actual benefits of this innovation will take on average 4.7 years (Sood an Tellis, 2009). Another

(12)

reason for choosing the time period of 30 years is the cyclical nature of the economy and a longer time period cancels out these inaccuracies.

As described above, 4 different innovative portfolios has been chosen. The sample sizes of these portfolios was relatively high because 60 companies per portfolio were which had their monthly returns over a period of 30 years which equal to 360 months. If we calculate 60 times 360 months we should get 21600 observations but because a lot companies did not exist for 30 years the number of observations is smaller than 21600. The monthly average was then calculated by taking the average of all the different companies in the corresponding month.

The S&P500 will be the control group in this research. The S&P500 or the Standard & Poor’s 500 index is a market-capitalization-weighted index of the 500 largest publicly listed and traded companies in The United States. This control group has been chosen as benchmark because it the most suitable for this research because it is the most widely accepted and used proxy. A lot of index funds and Exchange Traded Funds (ETF) have been constructed with the S&P500 as base. Furthermore a lot of researcher, academics and analysts use the S&P500 as their proxy to perform research on various studies and on market behavioral patterns. For this reason, it will also ensure that this research is more comparable to previous conducted research in the same field. Because the S&P 500 is calculated as one entity the sample size is 30 times 12, which equals 360 observations.

The risk-free rate will be the monthly rate on 30-year U.S. treasury bill. A long term U.S. treasury bill (T-bills) is usually seen as the most safe way to invest and can therefore be seen as the risk free rate. This is because T-bills are assumed to have a zero percent default rate because they are backed up and represent the U.S. government. Because the Risk free rate is calculated as one entity the sample size is 30 times 12, which equals 360 observations. The most important statistics of the variables can be found in table 2.

Table 2, Descriptive statistics portfolios, benchmark and risk free rate

Variable N Mean Std. Dev.

Return IP CURD 360 0.01202 0.03885 Return IP AVRD 360 0.01208 0.03815 Return IP AVRDI 360 0.01469 0.05153 Return IP Overall 360 0.01270 0.04175 Return S&P500 360 0.00889 0.03961 Risk Free 360 0.00224 0.00188

CURD (accumulated R&D expenses of years), AVRD(average R&D expenses per year), and AVRDI (average R&D intensity per year) Overall is based on the rankings from CURD, AVRD and AVRDI.

(13)

3.3 Variables

The data of the dependent variable will be the monthly returns of the most innovative

companies in the S&P500. The data of the independent variable will be the monthly returns of the S&P500. The data will be gathered from WRDS historical data in the period of January 1990 to December 2019. The returns of both the dependent and independent variable will be calculated with the following formula:

𝑅𝑒𝑡𝑢𝑟𝑛_𝑡 =(𝑃𝑡− 𝑃𝑡−1)

𝑃_𝑡−1

𝒘𝒉𝒆𝒓𝒆:

𝑃_𝑡 = 𝑃𝑟𝑖𝑐𝑒 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑖𝑛 𝑡𝑖𝑚𝑒 𝑡

𝑃_𝑡−1 = 𝑃𝑟𝑖𝑐𝑒 𝑜𝑓 𝑠𝑡𝑜𝑐𝑘 𝑖𝑛 𝑡𝑖𝑚𝑒 𝑡 − 1

Furthermore inconsistencies can occur when dividend is being paid out or when stock splits occur. In this research the prices of the different stocks and the S&P500 value were both calculated by using the adjusted closing values. This will make sure that the prices are corrected for dividends. Furthermore the data is also checked and corrected for stock splits. We also corrected for outliers by calculating the first quartile (Q1) and third quartile (Q3). By subtracting Q1 from Q3 we get the interquartile range (IQR). The upper bound can be

calculated by Q3+IQR and the lower bound is found by Q1-IQR. All values that were outside of the bounds were not included in the regression.

3.4 Hypothesis

The main objective of this research is to examine what the impact of R&D is on the stock returns of the companies that are in the S&P500. From previous literature research regarding innovation and stock performance we expect there will be a positive relation between stock performance and innovation. To examine if the innovative portfolios have a positive impact on stock performance the following 4 hypothesis have been made. The first hypothesis has been made with the CURD variable, the second with the AVRD variable, the third with the AVRDI variable and lastly the OVER variable that combines the 3 before that.

(14)

1) - H0: Investing in a portfolio with innovative firms(measured on CURD) does not outperform the market

- H1: Investing in a portfolio with innovative firms (measured on CURD) does outperform the market

2) - H0: Investing in a portfolio with innovative firms (measured on AVRD) does not outperform the market

- H1: Investing in a portfolio with innovative firms(measured on AVRD) does outperform the market

3) - H0: Investing in a portfolio with innovative firms (measured on AVRDI) does not outperform the market

- H1: Investing in a portfolio with innovative firms(measured on AVRD1) does outperform the market

4) - H0: Investing in a portfolio with innovative firms (measured on OVER) does not outperform the market

- H1: Investing in a portfolio with innovative firms(measured on OVER) does outperform the market

If the results show that one or several of the four innovative portfolios has a positive significant Jensen’s alpha and thus a outperforms the market, than H0 gets rejected and H1 will be accepted. If hypothesis 0 is accepted in one or several portfolios, then that means that the innovative companies in that corresponding portfolio do not have a Jensen’s alpha and thus their returns will not outperform the market. A significance level of 5% will be used.

We will, as already mentioned, also examine if the Jensen’s alphas will change over time. This will be done to see if the positive relation of R&D and stock performance has increased over the period of 30 years like was suggested in previous research. After the Fama-French factor 3 model regression has been done, the constant of the months in the period from January 1990 till December 2019 will be compared with each other, in search of a structural break. After determining where the structural break appeared, we will be able to divide the regression into two different regressions. These two regressions will than be tested against each other to see if they differ significantly. To see if that research holds the following hypothesis has been made.

5) H0: The returns of the innovative companies is not changing over time H1: The returns of the innovative companies is changing over time

(15)

If the results show that the constants (Jensen’s alpha) of the divided regressions differ significantly, then H0 gets rejected and H1 will be accepted. This will show that after the break the impact of R&D on stock performance will have significantly changed. If the results show that the constants do not change significantly there is not enough evidence to prove that the impact of R&D on stock performance has changed over time.

4. Methodology

4.1 Jensen’s index

A central problem in finance has been the evaluation of “performance” of portfolios and their corresponding risk. This concept of portfolio performance has two dimensions. The first dimension is the ability to predict future prices and the second dimension is the ability to minimize the level of “insurable risk”. The major difficulty was the proper understanding of and the measure of risk in attempting to evaluate a portfolio performance. The measures of Sharpe, Lintner and Treynor and their models are focused on the “predictive ability”, and thus do not fully cover the levels of risk (Jensen, 1968).

In his research Michael Jensen was searching for a method to fix this central problem in finance. He wanted to investigate if mutual funds that he was studying were outperforming the market. The returns of the mutual funds were compared to the market return. To measure the mutual funds portfolio performance he constructed his own analytic tool which is better known as “Jensen’s alpha”.

𝛼𝑖𝑝 = 𝑅𝑖𝑝− 𝑅𝑟𝑓+ 𝛽𝑖𝑝∗ (𝑅𝑚− 𝑅𝑟𝑓) + 𝜀𝑖 𝒘𝒉𝒆𝒓𝒆: 𝛼𝑖𝑝= 𝑎𝑙𝑝ℎ𝑎 = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝐽𝑒𝑛𝑠𝑒𝑛′𝑠 𝑎𝑙𝑝ℎ𝑎 𝑜𝑓 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑣𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑅_𝑖𝑝 = 𝑡𝑜𝑡𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑣𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑅_𝑟𝑓 = 𝑟𝑖𝑠𝑘 𝑓𝑟𝑒𝑒 𝑟𝑎𝑡𝑒 (30𝑦 𝑈𝑆 𝑡𝑟𝑒𝑎𝑠𝑢𝑟𝑦 𝑏𝑜𝑛𝑑) 𝑅_𝑚 = 𝑡𝑜𝑡𝑎𝑙 𝑚𝑎𝑟𝑘𝑒𝑡 𝑟𝑒𝑡𝑢𝑟𝑛 (𝑆&𝑃500) 𝛽_𝑖𝑝= 𝑏𝑒𝑡𝑎 = 𝑓𝑎𝑐𝑡𝑜𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑣𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝜀𝑖 = 𝑒𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚

(16)

The Jensen’s alpha, or Jensen’s measure, or Jensen’s index, is a risk adjusted performance measure. If we mention alpha we will refer to Jensen’s alpha. Alpha represents the returns on average that an investment portfolio makes above or below the predicted return. If alpha is positive then the portfolio is earning abnormal excess returns. Excess returns are the returns of the portfolio minus the risk free rate, or the portfolio is “beating the market”. If alpha is

negative, then the portfolio is underperforming relative to the market. The expected portfolio returns are predicted by the capital asset pricing model (CAPM) that we will cover with the Fama-French 3 factor model below. Alpha is measured given the portfolio’s beta and the average returns of the market.

The beta of a portfolio is a measure of volatility relatively to the market. If Beta is 1 than the portfolio has the same volatility as the market. If beta is above 1, the portfolio is riskier than the market and if beta is between 0 and 1 than the portfolio is less risky than the market. Beta can also be negative, this will show that the chosen portfolio has an inverse relation to the market but this is highly unlikely. Beta is calculated with the covariance of the return of the innovative portfolio and return of the market and the variance of the market.

𝛽_𝑖𝑝= 𝐶𝑜𝑣𝑎𝑟𝑖𝑎𝑛𝑐𝑒(𝑅𝑖𝑝, 𝑅𝑚)

𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒(𝑅_𝑚)

Furthermore, the risk free rate or 𝑅_𝑟𝑓 is nonexistent in the real word. There is no

security that is entirely risk free. To come as close as possible the 30 year monthly treasury

bill rate is used in this research. 𝑅_𝑖𝑝 will be determined by the different innovative portfolios

that we have made based on different innovation measures. The market returns (𝑅_𝑚) will be

that of the S&P500, this is the chosen because the portfolio’s has been made based on the S&P500 and thus most similar in terms in size, growth and geographical measures. The error

term 𝜀𝑖 represents the residual term in this model. The expected mean of this error term is

zero.

In this research we will examine if the various portfolios that are constructed in this research have an alpha that is significant different from zero. By doing this it can be

determined if our innovative portfolios have abnormal returns compared to the market. And because alpha is the additional excess return on top of the markets excess return adjusted for

(17)

its volatility, this will be the measure that reassembles the real world the most. And thus giving us the most reliable results of the impact of R&D on stock performance.

4.2 Fama-French Three-factor model

The Fama and French Three-Factor model (Fama-French model) is an expansion of the CAPM model that is developed in 1992. CAPM describes the relationship between expected excess returns for securities and systematic risk. The CAPM model is displayed below in bold letters. If the CAPM model was rewritten as a function of alpha you would have the Jensen’s index. The Fama-French model expands the CAPM model by adding two variables, which take into account that size risk and value risk. Eugene Fama and Kenneth French constructed this new model because they found that CAPM was not complete in explaining the expected excess returns. And they found that the Small minus Big (SMB) and High minus Low (HML) do a good job explaining the cross-section of average returns on NYSE, Amex and NASDAQ stocks for the 1963-1990 period (Fama and French, 1993).

𝑹𝒊𝒑− 𝑹𝒓𝒇 = 𝜶𝒊𝒑+ 𝜷𝒊𝒑∗ (𝑹𝒎− 𝑹𝒓𝒇) + 𝛽𝑠∗ 𝑆𝑀𝐵𝑡+ 𝛽𝑣∗ 𝐻𝑀𝐿𝑡+ 𝜺𝒊 𝒘𝒉𝒆𝒓𝒆: 𝑅𝑖𝑝− 𝑅𝑟𝑓 = 𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑒𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛 (𝑜𝑓 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑣𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜) 𝑅_𝑚− 𝑅_𝑟𝑓= 𝑒𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑡ℎ𝑒 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑜𝑟𝑓𝑜𝑙𝑖𝑜 (𝑜𝑓 𝑆&𝑃500) 𝛽_{𝑖𝑝,𝑠,𝑣} = 𝑓𝑎𝑐𝑡𝑜𝑟 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 (𝑖𝑝 = 𝑖𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑣𝑒 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜, 𝑠 = 𝑠𝑖𝑧𝑒, 𝑣 = 𝑣𝑎𝑙𝑢𝑒) 𝑆𝑀𝐵_𝑡 = 𝑠𝑖𝑧𝑒 𝑝𝑟𝑒𝑚𝑖𝑢𝑚 (𝑠𝑚𝑎𝑙𝑙 𝑚𝑖𝑛𝑢𝑠 𝑏𝑖𝑔) 𝐻𝑀𝐿𝑡 = 𝑣𝑎𝑙𝑢𝑒 𝑝𝑟𝑒𝑚𝑖𝑢𝑚 (ℎ𝑖𝑔ℎ 𝑚𝑖𝑛𝑢𝑠 𝑙𝑜𝑤)

The Fama-French model uses 3 factors to explain expected excess returns. The first factor is already mentioned in the Jensen’s index above and that is the volatility of the portfolio relative to the market multiplied with the excess return of the market. CAPM and Jensen’s index were not the most reliable models because only use one factor. Fama and French figured out this had to do with the simplifying condition of the models. According to the model of Fama and French, the low beta stocks, value stocks or small stocks could provide a positive alpha. To adjust for this problem they added the SMB and the HML factors to the models.

(18)

The second factor that is used to explain the excess portfolio returns, is the Small minus Big factor. This factor examines the market equity of a company and is also known as the “size” factor. Companies with a relatively small market equity tend to outperform the market and thus have a positive alpha. The SMB is the difference between the returns of a diversified portfolio consisting of small market equity companies and the returns of a

diversified portfolio consisting big market equity companies. The companies that are labeled as “small” have a market equity that is smaller than the median market equity of the proxy. The companies that are labeled as “big” have a market equity that is bigger than the median market equity of the proxy. The SMB factor is calculated by subtracting the average returns of the small companies by the average returns of the big companies in the portfolio.

The third and last factor to explain excess portfolio returns, is the High minus low factor. This is the factor that examines the book-to-market ratio and is also known as the “value factor”. Research has shown that companies with a relatively high market-to-book ratio tend to outperform companies that have a relatively low market-to-book ratio. The first step in calculating the HML factor is ranking the “small” and “big” companies obtained from

the SMB factor based on their market-to-book ratio. All the companies above the 70th

percentile are rated “value” companies split into “small value” and “big value” companies.

All the companies between the 30th and the 70th percentile are rated “neutral” and all the

companies till the 30th percentile are rated “growth”. The neutral and growth companies are

also split into small and big. Subsequentially the average returns of both the small and big value companies is subtracted from the average returns of both the small and big growth companies to obtain the HML factor.

I will use the Fama-French model because it gives more reliable results than CAPM and Jensen’s index. When the SMB and HML factors are not included in the model, omitted variable bias could occur which would lead to our regression being invalid. All the previous research also points in the direction that the Fama-French model has reliable results.

4.3 Chow test

After a linear regression is used to estimate an economic relationship, the question often arises whether the found relationship remains the same in different periods of time (Chow, 1960). So after we have done the regressions of the various innovative portfolio’s we have to test

(19)

whether our data and thus results have changed over time. To examine if there is a significant difference in the impact of R&D on stock performance we use the Chow test. The chow test examines whether there is a significant difference between data in certain time periods. Gregory Chow constructed a model in 1960 to test whether the true coefficients in two regressions on different data sets are equal. The Chow test is used most often in time series analysis to test if there is a structural break.

Because we are using the Fama-French model for our regressions we used it below to show how the Chow test works. To test if there is a structural break in the data, the model is split into different parts. Our model is as follows:

𝑅_𝑖𝑝− 𝑅_𝑟𝑓= 𝛼_𝑖𝑝+ 𝛽_𝑖𝑝∗ (𝑅_𝑚− 𝑅_𝑟𝑓) + 𝛽_𝑠∗ 𝑆𝑀𝐵_𝑡+ 𝛽_𝑣∗ 𝐻𝑀𝐿_𝑡+ 𝜀_𝑖

Now we can split our model into two groups and it will give:

𝑅𝑖𝑝1− 𝑅𝑟𝑓1 = 𝛼𝑖𝑝1+ 𝛽𝑖𝑝1∗ (𝑅𝑚1− 𝑅𝑟𝑓1) + 𝛽𝑠1∗ 𝑆𝑀𝐵𝑡1+ 𝛽𝑣1∗ 𝐻𝑀𝐿𝑡1+ 𝜀𝑖1

And it will give:

𝑅_𝑖𝑝2− 𝑅_𝑟𝑓2 = 𝛼_𝑖𝑝2+ 𝛽_𝑖𝑝2∗ (𝑅_𝑚2− 𝑅_𝑟𝑓2) + 𝛽_𝑠2∗ 𝑆𝑀𝐵_𝑡2+ 𝛽_𝑣2∗ 𝐻𝑀𝐿_𝑡2+ 𝜀_𝑖2

When we have split our data into two different models we can perform regression for the two different split models. If the results of the regressions are known the chow test can be done using the following formula:

(𝑆𝑆𝑅_𝑐 − (𝑆𝑆𝑅₁+ 𝑆𝑆𝑅₂))/𝑘 (𝑆𝑆𝑅₁+ 𝑆𝑆𝑅₂)/(𝑁₁+ 𝑁₂− 2𝑘) 𝒘𝒉𝒆𝒓𝒆: 𝑆𝑆𝑅𝑐 = 𝑆𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 𝑟𝑒𝑠𝑖𝑑𝑢𝑎𝑙𝑠 𝑐𝑜𝑚𝑏𝑖𝑛𝑒𝑑 𝑆𝑆𝑅₁ = 𝑆𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 𝑟𝑒𝑠𝑖𝑑𝑢𝑎𝑙𝑠 𝑜𝑓 𝑚𝑜𝑑𝑒𝑙 1 𝑆𝑆𝑅₂ = 𝑆𝑢𝑚 𝑜𝑓 𝑠𝑞𝑢𝑎𝑟𝑒𝑑 𝑟𝑒𝑠𝑖𝑑𝑢𝑎𝑙𝑠 𝑜𝑓 𝑚𝑜𝑑𝑒𝑙 2 𝑁₁ = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠 𝑖𝑛 𝑓𝑖𝑟𝑠𝑡 𝑔𝑟𝑜𝑢𝑝 𝑁₂ = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠 𝑖𝑛 𝑠𝑒𝑐𝑜𝑛𝑑 𝑔𝑟𝑜𝑢𝑝 𝑘 = 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟𝑠

(20)

In hypothesis 5, H0 states that the abnormal excess returns of the innovative companies are

not changing over time. Which means that 𝛼_𝑖𝑝1 = 𝛼_𝑖𝑝2, 𝛽_𝑖𝑝1 = 𝛽_𝑖𝑝2, 𝛽_𝑠1= 𝛽_𝑠2 and 𝛽_𝑣1=

𝛽_𝑣2. The number of parameters in our model is 5, so k=5. The degrees of freedom are 𝑁₁+

𝑁₂− 2𝑘. The chow test has an F distribution.

5. Results

5.1 Innovative portfolios (hypothesis 1, 2, 3 & 4)

In this section the first 4 hypothesis will be answered. The 4 different portfolios are based on different measures of innovation but the way how we got the results are for all of them the same and they will be answered together in this section. The first 4 hypothesis all have H0 that they will not outperform the market and H1 that they will outperform the market.

To answer hypothesis 1 till 4 and thus if the innovative portfolio’s haver a higher excess return than the market premium, we first have examine whether the variance of the error terms in the different data are constant. This is done by performing the Breusch-Pagan test using STATA. The Breusch-Pagan test is a test against heteroscedasticity and thus tests the variance of the error terms. In this test H0 assumes that the error term is constant and thus assumes homoscedasticity. The alternative hypothesis assumes that the error term is not constant and thus heteroskedasticity occurs. The test is done with a significance level of 5%.

Table 3, Test for heteroskedasticity for different innovative portfolio’s Breusch-Pagan test

Innovative portfolio Chi2 value P-value

CURDX 15.66 0.0013

AVRDX 15.09 0.0017

AVRDIX 13.90 0.0031

OVERX 15.42 0.0015

CURDX = the excess returns of the CURD portfolio, AVRDX = the excess returns of the AVRD portfolio, AVRDIX = the excess returns of the AVRDI portfolio and OVERX = the excess returns of the OVER portfolio.

(21)

Comparing the P-values from Table 3 with the significance of 0.005, we have to reject all the null hypothesis for the 4 portfolios (CURD: 0.0013<0.005, AVRD: 0.0017<0.005, AVRDI: 0.0031<0.005 and OVER: 0.0015<0.005). This means that the null hypothesis is rejected and that heteroskedasticity occurs for all 4 portfolio’s. Heteroskedasticity has serious

consequences for OLS because the estimated standard errors not reliable anymore. To adjust for heteroskedasticity a robust regression is being done.

To answer the question if the different portfolios are outperforming the market we have to look at Jensen’s alpha. Alpha is in the regression the estimated constant. If alpha is significant different from 0, than we reject the null hypothesis and we will accept the alternative hypothesis.

The H0 of hypothesis (1) states that the innovative portfolio measured by cumulative R&D spending does not outperform the market. The found t-value for the constant in portfolio CURDX has a p-value of 0.0. The critical value for p is 0.05 and therefore we can reject the null hypothesis (0.0<0.05) and accept the alternative hypothesis. We have enough evidence to conclude that the portfolio measured by cumulative R&D spending does outperform the market.

The estimated Beta is approximately 0.85 which means that this portfolio is less volatile than the S&P500. This means that we have found evidence that a portfolio with more innovative companies included, ranked on CURDX has less risk than the market proxy. The found beta also has a p-value of 0.0 and with a significance level of again 5% this means that beta is significant (0.0<0.05).

The SMB factor is also significant with a p-value of 2*0.013=0.026 and this is smaller than 0.05. This means that we have found scientific evidence that the small market equity companies outperform the big market equity companies in this portfolio. In contrast the HML factor is not significant (0.958>0.05) and this does not support previous research that claims that value companies outperform growth companies in this portfolio. In hypothesis 2, 3 and 4 we only mention differences in results about the three factors if they are different from hypothesis 1.

Secondly the H0 of hypothesis (2) states that the innovative portfolio measured by average R&D spending does not outperform the market. The found t-value for the constant in portfolio AVRDX has a p-value of 0.0. Therefore we can also reject the null (0.0<0.05) and accept the alternative hypothesis. We have enough evidence to conclude that the portfolio

(22)

measured by average R&D spending does outperform the market. In contrast to the results on the results of the portfolio based on CURDX (hypothesis 1), we have SMB factor that requires a higher significance level than 0.05 to be significant. The p-value is 0.07 and is bigger than 0.05. This factor could be significant when the level of significance was higher.

Thirdly, the H0 of hypothesis (3) states that the innovative portfolio measured by average R&D intensity does not outperform the market. The found t-value for the constant in portfolio AVRDIX has a p-value of 0.0. Therefore we can also reject the null hypothesis (0.0<0.05) and accept the alternative hypothesis. We have enough evidence to conclude that the portfolio measured by average R&D intensity does outperform the market. In contrast to the results from hypothesis 1, every factor is significant with a significance level of 0.05 because the p-value of every factor is 0.0. This means that we have found evidence that in this portfolio the value companies outperform the growth companies.

Lastly, the H0 of hypothesis (4) states that the innovative portfolio measured by the overall ranking of innovation does not outperform the market. The found t-value for the constant in portfolio OVERX has a p-value of 0.0. Therefore we can reject the null hypothesis (0.0<0.05) and accept the alternative hypothesis. We have enough evidence to conclude that the portfolio measured by the overall ranking of innovation does outperform the market. The results of this hypothesis is the same as hypothesis 1.

(23)

Table 4, The regression of all the portfolios

Robust Regression

Portfolio Variable Coefficient std error t-value

CURDX RM-RF 0.85056* .02910 29.23 N=360 SMB 0.10143* .04055 2.50 F=289.13* HML -0.00240 .04579 -0.05 R2=0.7713 Constant 0.00404* .00105 3.85 AVRDX RM-RF 0.83501* 0.02971 28.11 N=360 SMB 0.07246*** 0.03991 1.82 F=266.13* HML 0.02023 0.04654 0.43 R2=0.7614 Constant 0.00419* 0.00106 3.97 AVRDIX RM-RF 0.99619* 0.03791 26.28 N=360 SMB 0.35074* 0.06288 5.58 F=267.43* HML -0.26840* 0.06720 -3.99 R2=0.7271 Constant 0.00583* 0.00147 3.96 OVERX RM-RF 0.89528* 0.03130 28.61 N=360 SMB 0.13351* 0.04901 2.72 F=286.64* HML -0.06230 0.05404 -1.15 R2=0.7591 Constant 0.00446* 0.00115 3.89

(*) = significant with a level of 0.01, (**) = significant with a level of 0.05, (***) = significant with a level of 0.1

5.2 The change in R&D impact over time (hypothesis 5)

To answer hypothesis 5 we had to find a structural break in the alpha that was estimated in the different portfolios we have constructed. We have done this by performing the Chow test as described in the methodology. Since we had data that is heteroskedastic there is some research that suggests that the Chow test will predict unreliable results. However in a study done by Toshihisa Toyoda in 1974 shows that this is only the case when the sample size is small. And since we have a sample size of 360 we have still used the Chow test to see whether there is a structural break in de Jensen’s alpha. We have changed the name of the Chow test to the Wald test because the Wald test is the Chow test when heteroskedasticity does occur.

(24)

Since the structural breaks are not entirely clear, we have performed the test without knowing the structural breaks. We started off with the CURD portfolio which had an insignificant factor which was the HML factor. We have performed a second robust regression without this factor. When the Wald test was performed, it showed a break in 31/12/1996 as can be seen in appendix 1. We have found enough evidence that there was a significant change in alpha (0.0006<0.05) at the end of 1996.

The AVRD portfolio also had an insignificant and that was also the HML factor. As you can see in appendix 2, performing a Wald test on this portfolio gives us a structural break on 31/03/1997 and this is also a significant structural break (0.0007<0.005).

The factors used in our model were all significant in the AVRDI portfolio. As you can see in appendix 3, performing a Wald test on this portfolio gives us a structural break on 31/07/1998 and this is also a significant structural break (0.0000<0.005).

The AVRD portfolio also had an insignificant and that was also the HML factor. As you can see in appendix 4, performing a Wald test on this portfolio gives us a structural break on 31/12/1996 just like the CURD portfolio. The p-value is smaller than 5% so this portfolio also gives us enough evidence of a significant break (0.0000<0.005).

Concluding, we have found significant evidence that there is a structural break between the impact of R&D on stock performance between the period 12-31-1996 until 07-31-1998. A reason for this structural break could be the upcoming of the internet. The internet had a breakthrough at the mass population around the year 1996. This is in line with previous research.

(25)

6. Conlcusion

We have found that there was a lot of previous research about R&D and firm performance. The link between R&D and the productivity factor, patents, firm performance and

profitability is been studied many times. All evidence suggests that R&D has a positive

outcome on all of these different measures (Mansfield, 1980; Pakes and Griliches, 1984;Artz

Kendall et. Al, 2010;Ruberta and Kirca, 2012). The direct relation between R&D and stock

performance was less clear. And answering this question has been the main objective for this research.

This research aimed to identify the impact of R&D on stock performance in the U.S. in the period from January 1990 till December 2019. We have done this by constructing 4 different portfolios based on different measure of R&D. The market proxy we have used was the S&P500 in the same period, from January 1990 till December 2019. The hypothesis 1 till 4 are tested using the Fama-French model and the Jensen’s index to make an OLS regression with a significance of 5%. The H0 of the first 4 hypothesis were that the different portfolios were not outperforming the market. Based on a quantitative and qualitive analysis of the 4 different and self-constructed portfolios that were based on innovation, it can be concluded that we have found significant evidence that investing in portfolios based on R&D stock performance will have a positive and significant effect on the stock returns compared with the market. So H0 gets rejected in hypothesis 1 till 4 and H1 is being accepted. This was in line with what we have expected.

We also found in previous research that the competitive landscape has changed since the technological revolution (Betis and Hitt, 1995) and we wanted to test this to get a better understanding of the changes in the importance of innovation and thus in R&D. We have constructed and tested the H0 which states that the excess abnormal returns of the innovative companies are not changing over time. We tested this by performing a Chow test (in our case Wald because the error term was not constant) and we can conclude that there is enough evidence to accept our alternative hypothesis. This means that we can conclude that there has been a structural break in alpha somewhere in between the period 12-31-1996 until 07-31-1998.

(26)

7. Limitations

Every research has its limitations due to the limited data, limited scope and limited time. This paper also has its limitations and this implies that certain assumptions are made to conduct this research.

The first limitation is found when constructing the different innovative portfolios the data concerning R&D expenditures was limited. Of the 500 companies that are in the Standard and Poor’s index only 243 companies had R&D expenditure data available. This means that 257 companies were not included in any portfolio because of the lack of data.

The second limitation was found when comparing the different innovative portfolios to the S&P500, the weighing of the companies if different. In the S&P500 the weight of the companies is being calculated by market capitalization. The “bigger” (companies with a higher market cap) companies are being represent more in the S&P500 because the weight is calculated dividing the market cap of the company by the market cap of the entire index. In our research the companies are equally weight and this might also explain why the HML factor in several of our portfolios was insignificant.

The third limitation is the geographical factor. Because we have only considered companies and markets that were in the United States we can not make any assumptions about the impact of R&D on stock performance in general. Also the time changes that have been found are only significant for companies in the United States within the S&P500.

The fourth and last limitation is that there is a lot of debate whether the outperformance tendencies are due to the market efficiencies or in contrast the market inefficiency. For this reason it is also not entirely clear if the HML factor and the SMP factor capture actual risk or just persistent mistakes of investors and speculators.

Due to the limitations that occur in this research we have tried to minimize deviations caused by these limitations.

(27)

References

Ahuja, G. and Katila, R. (2001). Technological Acquisition and the Innovative Performance of Acquiring Firms: A Longitudinal Study. Strategic Management Journal 22(3):197-220

Artz, Kendall W. et al. (2010). “A Longitudinal Study of the Impact of R&D, Patents, and Product Innovation on Firm Performance.” Journal of Product Innovation

management 27.5: 725–740. Web.

Bettis, Richard A., and Michael A (1995). Hitt. “The New Competitive Landscape.” Strategic

Management Journal 16.SPEISS: n. pag. Print.

Cardinal, Laura B, and Donald E Hatfield (2000). “Internal Knowledge Generation: The Research Laboratory and Innovative Productivity in the Pharmaceutical Industry.”

Journal of Engineering and Technology Management 17.3: 247–271. Web.

Chandy, Rajesh K. and Gerard J. Tellis (1998), “Organizing for Radical Product Innovation: The Overlooked Role of Willingness to Cannibalize,” Journal of Marketing Research, 35 (November), 474–87.

Chakravarty, Anindita and Rajdeep Grewal (2011). “The Stock Market in the Driver’s Seat! Implications for R&D and Marketing,” Management Science, 57 (9), 1594–1609 Chiesa, V. (1996). Managing the Internationalization of R&D activities. IEEE Transactions

on Engineering Management 43(1):7–23.

Cho, Hee‐Jae, and Vladimir Pucik. (2005) “Relationship Between Innovativeness, Quality, Growth, Profitability, and Market Value.” Strategic Management Journal 26.6: 555– 575. Web.

Chow, G. (1960). Tests of Equality between Sets of Coefficients in Two Linear Regressions.

Econometrica, 28.

Doukas, A., Guo, J., Lam, H., & Xiao, S. (2016). Media Endorsements of New Product Announcements: A New Marketing Strategy. European Financial Management, 22(3), 394–426.

Fama, E., & French, K. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33(1), 3–56.

(28)

Hagedoorn, J. and Duysters, G. (2002). The Effect of Mergers and Acquisitions on the Technological Performance of Companies in a High-Tech Environment. Technology

Analysis and Strategic Management 14(1):67–85

Hauser, John, Gerard Tellis, and Abbie Griffin (2006). “Researchon Innovation and New Products: A Review and Agenda for Marketing Science,” Marketing Science, 25 (6), 687–717.

Hua, S., & Wemmerlöv, U. (2006). Product Change Intensity, Product Advantage, and Market Performance: An Empirical Investigation of the PC Industry. Journal of

Product Innovation Management, 23(4), 316–329.

Jensen, M. (1968). THE PERFORMANCE OF MUTUAL FUNDS IN THE PERIOD 1945– 1964. Journal of Finance, 23(2), 389–416.

Kirca, Rubera (2012), “Firm Innovativeness and Its Performance Outcomes: A Meta-Analytic Review and Theoretical Integration.” Journal of Marketing, 76 (3), p130-147.

Leković Miljan (2017), “Mutual Funds Portfolio Performance Evaluation Models: Sharpe, Treynor and Jensen Index.” Bankarstvo 46.4: 108–133. Web.

Mansfield, E. (1962). Entry, Gibrat’s Law, Innovation, and the Growth of Firms. American

Economic Review 55(5):1023–1051.

Mansfield, E. (1980). Basic Research and Productivity Increase in Manufacturing. American

Economic Review 70(5):863–873.

Pakes, A. and Griliches, Z. (1984). Patents and R&D at the Firm Level: A First Look. In: Patents and Productivity, ed. Z. Griliches. Chicago: University of Chicago Press, 55– 72.

Phuoc, L.T. (2018), “Jensen’s Alpha Estimation Models in Capital Asset Pricing Model.”

Journal of Asian Finance, Economics and Business 5.3: 19–29. Web.

Roberts, P. (1999). Product innovation, product–market competition and persistent profitability in the U.S. pharmaceutical industry. Strategic Management Journal, 20(7), 655–670.

Schumpeter, J.A. (1934). The Theory of Economic Development. Cambridge, MA: Harvard

(29)

Song, Michael et al. (2011). “Does Strategic Planning Enhance or Impede Innovation and Firm Performance?(Report).” Journal of Product Innovation Management 28.4: 503– 520.

Sood, Ashish ; Tellis, Gerard (2009). Do Innovations Really Pay Off? Total Stock Market Returns to Innovation. Marketing Science, 28(3), 442–456.

Sorescu and Jelena Spanjol (2008). “Innovation’s Effect on FirmValue and Risk: Insights from Consumer Packaged Goods,” Journal of Marketing, 72 (March), 114–32. Toyoda, T. (1974). Use of the Chow Test under Heteroscedasticity. Econometrica, 42(3),

601–608.

Wetter, John J. (2011). The Impacts of Research and Development Expenditures The Relationship Between Total Factor Productivity and U.S. Gross Domestic Product Performance. 1st ed. 2011. New York, NY: Springer New York.

Yanadori, Y., and V. Cui. (2013). “Creating Incentives for Innovation? The Relationship Between Pay Dispersion in R&D Groups and Firm Innovation Performance.” Strategic Management Journal 34.12: 1502–1511.

(30)

Appendix

Appendix 1, Wald test of cumulative R&D spending (CURD)

Number of obs 360

Full sample 1 - 360

Trimmed sample 55 - 307

Estimated break date 84

Test Statistic 23.7949

p-value 0.0006

Coefficients included in test constant

Appendix 2, Wald test of average annual R&D spending (AVRD)

Number of obs 360

Full sample 1 - 360

p-value 0.0007

Appendix 3, Wald test of average R&D intensity (AVRDI)

Number of obs 360

Full sample 1 - 360

p-value 0.0000

Appendix 4, Wald test of ranking overall (OVER)

Number of obs 360

Full sample 1 - 360

p-value 0.0000

(31)

Appendix 5, Companies portfolio cumulative R&D spending (CURD) *1000 GENERAL MOTORS CO 211.204,40 FORD MOTOR CO 193.310,00 MICROSOFT CORP 191.532,15 INTEL CORP 177.061,71 PFIZER INC 174.655,30 JOHNSON & JOHNSON 169.504,00 INTL BUSINESS MACHINES CORP 154.929,00 AMAZON.COM INC 146.790,60 MERCK & CO 142.739,70 ALPHABET INC 134.042,57 CISCO SYSTEMS INC 107.229,19 LILLY (ELI) & CO 103.241,90 BRISTOL-MYERS SQUIBB CO 92.555,00 APPLE INC 90.546,07 GENERAL ELECTRIC CO 82.908,00 HP INC 81.096,00 BOEING CO 78.851,00 ORACLE CORP 72.737,46 MOTOROLA SOLUTIONS INC 69.548,00 AMGEN INC 68.657,27

QUALCOMM INC 60.002,44 ABBOTT LABORATORIES 59.945,47 PROCTER & GAMBLE CO 50.115,00 RAYTHEON TECHNOLOGIES

CORP

49.126,40

FACEBOOK INC 48.374,00 ABBVIE INC 47.860,80 TEXAS INSTRUMENTS INC 42.852,00 GILEAD SCIENCES INC 42.055,20 DUPONT DE NEMOURS INC 39.822,00 3M CO 39.288,00 CATERPILLAR INC 35.027,00 APTIV PLC 33.433,00 HONEYWELL INTERNATIONAL INC 33.005,00 MEDTRONIC PLC 31.707,60 APPLIED MATERIALS INC 28.989,69 ADVANCED MICRO DEVICES 28.314,90 BIOGEN INC 26.579,85 LOCKHEED MARTIN CORP 26.166,00 DEERE & CO 24.642,40 EXXON MOBIL CORP 23.637,00

BOSTON SCIENTIFIC CORP 23.125,59 XEROX HOLDINGS CORP 22.435,00 MICRON TECHNOLOGY INC 21.862,17 SCHLUMBERGER LTD 20.783,05 ELECTRONIC ARTS INC 19.902,78 WESTERN DIGITAL CORP 19.838,09 BAXTER INTERNATIONAL INC 19.336,00 BROADCOM INC 18.431,00 SEAGATE TECHNOLOGY PLC 18.122,00 AT&T INC 17.509,00 NVIDIA CORP 17.218,29 EBAY INC 16.926,79 REGENERON PHARMACEUTICALS 16.734,06

AGILENT TECHNOLOGIES INC 16.254,00 ADOBE INC 15.608,08 DANAHER CORP 15.415,74 NORTONLIFELOCK INC 14.733,57 CORNING INC 14.605,20 JUNIPER NETWORKS INC 14.086,70 HEWLETT PACKARD

ENTERPRISE

13.780,00

Appendix 6, Companies portfolio average annual R&D spending (AVRD) *1000

ALPHABET INC 7.446,81 GENERAL MOTORS CO 7.040,15 FORD MOTOR CO 6.443,67 MICROSOFT CORP 6.384,41 INTEL CORP 5.902,06 AMAZON.COM INC 5.871,62 PFIZER INC 5.821,84 JOHNSON & JOHNSON 5.650,13 INTL BUSINESS MACHINES CORP 5.164,30 FACEBOOK INC 4.837,40 ABBVIE INC 4.786,08 MERCK & CO 4.757,99 CISCO SYSTEMS INC 3.574,31 LILLY (ELI) & CO 3.441,40

APPLE INC 3.233,79 BRISTOL-MYERS SQUIBB CO 3.085,17 GENERAL ELECTRIC CO 2.763,60 HP INC 2.703,20 BOEING CO 2.628,37 ORACLE CORP 2.424,58

MOTOROLA SOLUTIONS INC 2.318,27

AMGEN INC 2.288,58

QUALCOMM INC 2.069,05 ABBOTT LABORATORIES 1.998,18 HEWLETT PACKARD ENTERPRISE 1.968,57 PROCTER & GAMBLE CO 1.670,50 RAYTHEON TECHNOLOGIES CORP 1.637,55 BROADCOM INC 1.535,92

APTIV PLC 1.519,68

CORTEVA INC 1.479,00 TEXAS INSTRUMENTS INC 1.428,40 GILEAD SCIENCES INC 1.401,84

DOW INC 1.380,50

DUPONT DE NEMOURS INC 1.327,40

3M CO 1.309,60

CATERPILLAR INC 1.167,57 HONEYWELL INTERNATIONAL INC 1.100,17 MEDTRONIC PLC 1.056,92 APPLIED MATERIALS INC 966,32 SEAGATE TECHNOLOGY PLC 953,79

ADVANCED MICRO DEVICES 943,83 PAYPAL HOLDINGS INC 931,71 BIOGEN INC 885,99 LOCKHEED MARTIN CORP 872,20 DEERE & CO 821,41 BOSTON SCIENTIFIC CORP 797,43 EXXON MOBIL CORP 787,90 NVIDIA CORP 748,62 XEROX HOLDINGS CORP 747,83 AGILENT TECHNOLOGIES INC 738,82 MICRON TECHNOLOGY INC 728,74 EBAY INC 705,28 SCHLUMBERGER LTD 692,77 ACTIVISION BLIZZARD INC 690,67 ELECTRONIC ARTS INC 663,43 WESTERN DIGITAL CORP 661,27 BAXTER INTERNATIONAL INC 644,53 JUNIPER NETWORKS INC 612,47 AT&T INC 583,63 REGENERON PHARMACEUTICALS 557,80

(32)

Appendix 7, Companies portfolio average R&D intensity (AVRDI)

INCYTE CORP 76,47%

VERTEX PHARMACEUTICALS INC 64,96% REGENERON PHARMACEUTICALS 44,02% CADENCE DESIGN SYSTEMS INC 32,27%

SYNOPSYS INC 32,07%

DEXCOM INC 30,55%

TWITTER INC 29,55%

ALEXION PHARMACEUTICALS INC 28,44% ELECTRONIC ARTS INC 26,75%

AUTODESK INC 25,40%

LILLY (ELI) & CO 23,16%

BIOGEN INC 23,10%

ADVANCED MICRO DEVICES 22,78% MAXIM INTEGRATED PRODUCTS 22,38%

NVIDIA CORP 21,89%

JUNIPER NETWORKS INC 21,83% ARISTA NETWORKS INC 21,77%

ILLUMINA INC 21,52% AMGEN INC 21,31% QUALCOMM INC 21,20% ABBVIE INC 20,55% SERVICENOW INC 20,52% XILINX INC 20,26% FACEBOOK INC 19,84% ABIOMED INC 19,56%

GILEAD SCIENCES INC 19,04% ANALOG DEVICES 19,01% BROADCOM INC 18,82% ADOBE INC 18,24% ANSYS INC 17,99% INTUIT INC 17,66% BRISTOL-MYERS SQUIBB CO 17,57% QORVO INC 16,86% INTEL CORP 16,56% NORTONLIFELOCK INC 16,48% KLA CORP 16,28% PFIZER INC 15,98%

CITRIX SYSTEMS INC 15,89% MERCK & CO 15,63% F5 NETWORKS INC 15,54%

BOSTON SCIENTIFIC CORP 15,46% ACTIVISION BLIZZARD INC 14,87%

ALPHABET INC 14,86%

EDWARDS LIFESCIENCES CORP 14,52%

FORTINET INC 14,50%

LAM RESEARCH CORP 14,29% KEYSIGHT TECHNOLOGIES INC 14,20% MICROSOFT CORP 14,19% MICROCHIP TECHNOLOGY INC 14,00% SALESFORCE.COM INC 13,96% CISCO SYSTEMS INC 13,87%

NETAPP INC 13,79%

APPLIED MATERIALS INC 13,76%

ORACLE CORP 13,39%

SKYWORKS SOLUTIONS INC 12,69% TEXAS INSTRUMENTS INC 12,50% JOHNSON & JOHNSON 12,27% AGILENT TECHNOLOGIES INC 11,95%

GARMIN LTD 11,67%

MOTOROLA SOLUTIONS INC 11,12%

Appendix 8, Companies portfolio ranking overall (OVER) LILLY (ELI) & CO 1

INTEL CORP 2 PFIZER INC 3 ALPHABET INC 4 MICROSOFT CORP 5 ABBVIE INC 6 FACEBOOK INC 7 MERCK & CO 8 AMGEN INC 9 BRISTOL-MYERS SQUIBB CO 10 QUALCOMM INC 11

JOHNSON & JOHNSON 12 CISCO SYSTEMS INC 13

AMAZON.COM INC 14

GILEAD SCIENCES INC 15 ADVANCED MICRO DEVICES 16

BIOGEN INC 17

ORACLE CORP 18

MOTOROLA SOLUTIONS INC 19

BROADCOM INC 20

ELECTRONIC ARTS INC 21 INTL BUSINESS MACHINES CORP 22 ABBOTT LABORATORIES 23 GENERAL MOTORS CO 24

NVIDIA CORP 25

TEXAS INSTRUMENTS INC 26

FORD MOTOR CO 27

REGENERON PHARMACEUTICALS 28 APPLIED MATERIALS INC 29 BOSTON SCIENTIFIC CORP 30 JUNIPER NETWORKS INC 31

MEDTRONIC PLC 32

APPLE INC 33

APTIV PLC 34

BOEING CO 35

VERTEX PHARMACEUTICALS INC 36

ADOBE INC 37

HP INC 38

SYNOPSYS INC 39

NORTONLIFELOCK INC 40

MICRON TECHNOLOGY INC 41

ANALOG DEVICES 42

SEAGATE TECHNOLOGY PLC 43 AGILENT TECHNOLOGIES INC 44

3M CO 45

CADENCE DESIGN SYSTEMS INC 46

EBAY INC 47

INTUIT INC 48

WESTERN DIGITAL CORP 49 RAYTHEON TECHNOLOGIES CORP 50 ACTIVISION BLIZZARD INC 51 PROCTER & GAMBLE CO 52 GENERAL ELECTRIC CO 53

AUTODESK INC 54

HONEYWELL INTERNATIONAL INC 55

TWITTER INC 56

DUPONT DE NEMOURS INC 57 CATERPILLAR INC 58 HEWLETT PACKARD ENTERPRISE 59 BAXTER INTERNATIONAL INC 60

(33)

The impact of R&amp;D spending on stock performance in the U.S.