The effect of research and development expenses on stock price and stock return

(1)

1

The effect of research and development expenses on stock

price and stock return.

Amsterdam Business School

Name Darryl Yorks

Number 10205063

BSc in Business Economics Specialization Economics & Finance Supervisor name Rafael Perez Ribas Completion date 29-6-2015

Abstract

This paper researches the effect of research and development expenses on stock price and stock return. To find this effect, two multiple regression models have been developed and tested. The sample focused on consists of all indexed firms in North-America reporting positive research and development expenses from the years 2000 – 2014. The findings of this research show a negative or no effect of research and development expenses on both stock price and stock return.

(2)

2 Statement of Originality

This document is written by Student Darryl Yorks 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 contents.

(3)

3 Table of contents

1. Introduction 4

2. Literature review 5

2.1 The positive effect on stock prices and returns 5

2.2 The negative effect on stock prices and returns 7

3. Empirical strategy 8

3.1. Data and descriptive statistics 8

3.2. Method 10 4. Results 10 4.1. Main Results 10 4.2 Robustness check 13 5. Conclusion 19 References 21 Appendix 22

(4)

4 Introduction

Maximizing shareholder value and staying ahead of competition are the main goals for companies. To do so, it is not only important to fully exploit the current resources, but also to develop new ones. Coming up with, for example new products, is perceived to be one of the most critical things to do to sustain competitive advantage and long-term well-being (Lee et al., 2009). Makadok (2010) mentions having competitive advantage will lead to greater profits when competitors cannot compete it away. Cohen and Levinthal (1989) say that to stay ahead of competition it is not only important to invest in new products, but also in obtaining new information or improving the ability to exploit existing information. Besides mergers and acquisitions, to develop new products or new information, money needs to be invested in research and development (R&D). Developing new products, patenting them and obtaining profits from them, is the up side of R&D. The down side of R&D is that it costs money and there is no certainty of a payoff in the future. For example, the competition might develop the same product and patents it, making the investment worthless. Because of the uncertainty R&D is hard to valuate for shareholders, also since according to Chan et al. (2001) they base their valuation mostly on net assets.

This thesis attempts to estimate the effect of R&D expenses on the stock price and stock returns. The hypothesis is that R&D expenses have a positive relationship with stock price and stock return. Since many studies by Chan et al. (1990) or Stein (1988) rely on data form more than 20 years ago, we try to offer a new insight with recent data. Past studies focused mostly on the effects of R&D announcements. In this study we will look at the reports.

In the existing literature researchers are divided about the effects of R&D expenses on future profits and shareholder value. For example, Chan et al. (1990) find a positive market effect when planned R&D expenses increase. They state that investors look at the long-term earnings of R&D investments. Lee and Chen (2009) find a positive effect on shareholder value for larger firms due to economies of scale and scope. Furthermore, Pandit et al (2011) conclude that the future performance is positively related to the quality of the patent a firm has, meaning patenting new information or products leads to improved performance. On the other hand, the paper by Stein (1988) also discusses a negative effect on short term earnings and stock price. Since they are divided in their conclusions, we try to give more insight in the effects of R&D expenses on the value of companies on the stock market. Our findings suggest that the relationship between R&D expenses and stock price and stock return is negative. This

(5)

5 indicates that an increase in R&D expenses will lead to a decrease in stock price and stock returns.

Data come from all indexed companies in North-America in 2000-2014 that report a positive amount of R&D expenses in this period. Two panel models are estimated, which control for firms’ systematic differences. Controlling for firm fixed effects is important, because there might be time-invariant firm characteristics that drive stock price and stock return. It is also important to control for year effects, since there might be changes in the economy that have an impact on all firms, such as the credit crunch in 2008. The first model has stock price as the dependent variable and the second model uses stock returns as the dependent variable. The model for stock price accounts for effects on the short term, since it is only for one period. The model for stock return shows effects on the long term, since they consider the changes in stock price over time. Further, the sample will be divided into a tech and a low-tech group. We expect to find a difference between the two groups, since high-tech firms tend to be more dependent on R&D.

The remainder of this research consists of the following sections. Section 2 gives an analysis of the existing literature. Section 3 consists of the data section, discussing the hypotheses, methodology, and data and shows the summary statistics. Section 4 presents the results found from the regressions and section 5 summarizes and concludes the findings of this research.

2. Literature review

In the existing literature the effect of R&D expenses is ambiguous. Some studies find that R&D expenses are not valuated correctly and thus have a negative effect on the stock price and stock returns, whereas other studies show that there is a significantly positive effect. In this part the findings from earlier research will be discussed.

2.1 The positive effect on stock prices and returns

Stock prices are mainly determined by expectations of investors based on future cash flows. Chen et al. (2013) research two different factors that drive stock price, namely cash flows or discount rates. In their research they look at the effect of news about cash flows and discount rates on stock price and stock returns. They find that positive news shows a significant

(6)

6 increase in the stock price and stock returns, for both cash flow and discount rate news. This means that obtaining higher cash flows, which include profits, increases stock price and stock returns.

According to Chan et al. (1990) markets tend to respond positively on a planned increase in R&D expenses and negatively on a decrease. Increasing planned R&D expenses is picked up by the market as a positive signal, whereas a decrease will be seen as a negative one. Their findings suggest that investors look beyond the short-term earnings when valuing stock price. Furthermore, they have found a difference between high-tech and low-tech industries. In their paper they stated, “stock-price response clearly tends to be positive for the high-tech group (for example, electronics, pharmaceuticals, or scientific instruments), but negative for the low-tech group (for example, steel, oil refining, or nonferrous metals) “(Chan, Martin and Kensinger, 1990, p. 269). Their evidence suggests that the stock market gives a higher value to increased R&D expenses for high-tech firms than for low-tech firms. Their research does suffer from potential bias. They only use situations where management announces a change in R&D expenses and their findings also indicate that managers only announce an increase of the money spend on R&D when a positive effect on stock-price is anticipated.

Branch (1974) shows three possible ways in which profits, which also influence stock price and returns, and R&D expenses are related. Firstly, there is a positive effect of R&D expenses on both profits and growth. Secondly, he discovered profits and growth also positively influences R&D spending. Finally, he found a third factor determining profits and R&D spending simultaneously, namely government support. So next to the effect of R&D on profits there is also a positive effect the other way. According to Branch (1974), the

stimulation of R&D, caused by an increase in profits, can be highly dependent on managerial behavior. For the relation to hold, management needs to increase R&D once profits rise. If the management is content with the current profits, there is no need for them to increase the R&D expenses and the relation no longer holds. Furthermore, because R&D has a relatively long time horizon, borrowing funds is an unlikely way to finance R&D expenses. This causes changes in R&D to be associated with changes in profits. The positive effect of R&D on profits and growth, which also influences stock prices and stock returns, is explained by firms taking advantage of high-yield, high-risk opportunities.

Another positive effect is found by Lee and Chen in 2009. They look at the introduction of new products, which is influenced by R&D expenses. For example, they account for technological sophistication, environmental turbulence and competitive intensity,

(7)

7 but the most important variable was firm size. Large firms tend to have economies of scale and scope that aid in launching and developing new products, so they can take a first mover advantage. Moreover, large companies usually sense customer needs and have greater market knowledge and can therefore convert their R&D programs into firm value. Just like Chan et al. (1990), the firms have been divided based on their industries. Companies that research sophisticated new products, such as computer software, are more likely to benefit from higher profits. This suggests that high-tech companies get a higher valuation of their R&D expenses than low-tech firms.

Furthermore, Chan et al. (2001) find that companies have large amounts of intangible assets, such as R&D, and are generally not recorded on financial statements in the United States. Since they are not reported, investors have to estimate the intangibles and thus mispricing arises. In general investors tend to overestimate the benefits from R&D and therefore overvalue companies that are R&D-intensive, such as the high-tech sector. This causes an increase in the stock price and stock returns and thus shows a positive effect of R&D.

2.2 The negative effect on stock prices and returns

Other than positive effects, previous research has also found negative effects of R&D expenses on stock prices and returns. Doukas and Switzer (1992) find that both an increase and a decrease in R&D expenses can have a significant effect on the company’s stock price. It could signal that the current R&D technology is improving, so a lower budget is needed to spend. On the other hand, management chooses a R&D program to maximize the expected discounted value of the net cash flows, which would mean an increase in stock price. In the same study where Lee and Chen (2009) find a positive effect, they also find a negative effect on certain spending levels. According to them, “there is a U-shape nonlinear effect of R&D resource intensity and the relationship between new product introductions and shareholder value.” (p. 100). Stock holders already have expected a basic amount of money spend on R&D that is needed to facilitate new product development. Further, R&D involves costs, but these costs may not be larger than the benefits the shareholders are expecting from the R&D activities. Based on this, if companies are only investing a minimum in R&D, this will have a negative effect on the stock price.

In a published paper by Hall et al. (1993), there is suggested that investors only have a short time horizon. Shareholders are holding shares in numerous companies and expect to

(8)

8 hold them only for a short period of time. They fail to get access to the inside information and base their decisions only on measurable company attributes, such as profits. Due to this, they fail to take the long-term benefits of R&D expenses into account and undervalue them or, just see them as expenses rather than investments.

An investor not being able to see beyond current earnings is also discussed by Stein (1988). Because of this, long-term payoffs, like R&D projects, are being avoided. Managers are afraid of declining short-term earnings, which will lower stock prices. Therefore they choose for short-term investments with a lower payoff. This poor decision making might work in the short-term, but will harm firms in the long-term.

3. Empirical strategy

3.1. Data and descriptive statistics

The data used are from Compustat, which includes all firms in North-America in 2000 – 2014. However, only firms that report positive R&D expenses are used. Firms are divided into two groups based on their sector: “high-tech” and “low-tech”. A difference is expected

between high-tech and low-tech, because high-tech firms are more dependent and focused on coming up with, for example new software or technology to stay ahead of competition and generate higher profits. This is not the main focus for low-tech. Sectors firms operate in, are based on the global industry classification standard (GICS), a GICS starting with “45” is marked as “high-tech”.

To estimate the variables, they are defined as following: The variable Ln(PRICE) is measured as the logarithm of close price in the end of the fiscal year. Ln(RETURN) is the logarithm of close price in the end of the fiscal year minus the logarithm of close price in the previous year. Ln(RD) is the logarithm of R&D expenses in the current year over total assets in the previous year. Ln(CAPX) is the logarithm of capital expenditures in the current year divided by total assets from last year. Ln(SIZE) is the logarithm of total assets in the current year. Ln(CF) is the logarithm of total cash flows in the current year over total assets from the previous year. RD is the R&D expenses in the current year over total assets in the previous year. Ln(RD)² is the square of Ln(RD). Ln(FRD) is the logarithm of R&D expenses in the coming year over total assets in the current year. The descriptive statistics are presented in Table 1. Further explanation of the variables can be found in Appendix A.

(9)

9 Table 1: Descriptive statistics

N. of obs. Mean Std. Dev. Min Max

All Firms Ln(PRICE) 46199 1,371 2,132 -9,210 7,910 Ln(RETURN) 46118 -0,067 0,951 -8,716 11,191 Ln(RD) 37076 -2,772 1,725 -12,334 9,085 Ln(CAPX) 40388 -3,662 1,340 -13,037 8,513 Ln(SIZE) 42933 4,736 2,973 -6,908 13,590 Ln(CF) 24207 -2,321 1,001 -10,063 7,481 RD 38302 0,979 56,119 -14,000 8825,000 Ln(RD)² 37076 10,659 11,653 0,000 152,147 Ln(FRD) 42894 -2,744 1,738 -12,335 9,085 High-Tech Ln(PRICE) 16144 1,302 1,961 -9,210 7,022 Ln(RETURN) 16116 -0,096 0,950 -8,716 8,083 Ln(RD) 13542 -2,328 1,157 -10,762 7,153 Ln(CAPX) 14180 -3,769 1,211 -10,902 5,159 Ln(SIZE) 14800 4,561 2,552 -6,908 12,079 Ln(CF) 8313 -2,332 1,073 -10,063 6,938 RD 13886 0,401 13,374 -14,000 1278,000 Ln(RD)² 13542 6,757 6,752 0,000 115,814 Ln(FRD) 15725 -2,307 1,172 -10,762 7,153 Low-Tech Ln(PRICE) 30055 1,407 2,218 -9,210 7,910 Ln(RETURN) 30002 -0,051 0,951 -8,517 11,191 Ln(RD) 23534 -3,028 1,933 -12,335 9,085 Ln(CAPX) 26208 -3,603 1,401 -13,037 8,513 Ln(SIZE) 28133 4,827 3,168 -6,908 13,590 Ln(CF) 15894 -2,315 0,962 -9,608 7,481 RD 24416 1,308 69,560 -5,500 8825,000 Ln(RD)² 23534 12,905 13,187 0,000 152,147 Ln(FRD) 27169 -2,996 1,950 -12,335 9,085

In Table 1 the difference between the high-tech and low-tech are shown. On average low-tech firms have a higher amount of total assets but still have a higher value of RD. This would mean, contrary to the expectation, that low-tech firms spend more money on R&D expenses, as a ratio over total assets from last year, than high-tech firms. Once the logarithmic transformation is applied, the high-tech sector shows a higher average. Stock price, which represents future cash flows, are higher on average for low-tech firms. This is in line with the

(10)

10 cash flow findings, since they show a higher value for low-tech firms.

3.2 Method

To estimate the effect of R&D expenses on the stock price and stock return consider the following models:

𝐿𝑛(𝑃𝑅𝐼𝐶𝐸)_𝑖,𝑡 = 𝛼_𝑖 + 𝛽₁𝐿𝑛(𝑅𝐷)_{𝑖,𝑡−1}+ 𝛽₂𝐿𝑛(𝐶𝐴𝑃𝑋)_{𝑖,𝑡−1}+ 𝛽₃𝐿𝑛(𝑆𝐼𝑍𝐸)_{𝑖,𝑡−1}+ 𝛽₄𝐿𝑛(𝐶𝐹)_{𝑖,𝑡−1}+ 𝜇_𝑡+ 𝜀_𝑖,𝑡(1)

𝐿𝑛(𝑅𝐸𝑇𝑈𝑅𝑁)𝑖,𝑡 = 𝛼𝑖 + 𝛽1𝐿𝑛(𝑅𝐷)𝑖,𝑡−1+ 𝛽2𝐿𝑛(𝐶𝐴𝑃𝑋)𝑖,𝑡−1+ 𝛽3𝐿𝑛(𝑆𝐼𝑍𝐸)𝑖,𝑡−1+

𝛽₄𝐿𝑛(𝐶𝐹)_{𝑖,𝑡−1}+ 𝜇_𝑡+ 𝜀_𝑖,𝑡(2)

There is controlled for capital expenses, firm size, cash flow, yearly effects and firm fixed effects. The parameter αi represents firm fixed effects, βi is the regression coefficient, μt is the

year-specific effect and ε_i is the error term.

For both models the null hypotheses β1 = 0 will be tested. If β1 > 0 in model 1,

increasing RD by 1% will have a positive effect on PRICE with β1%, otherwise it will

decrease PRICE by β1%. When β1 > 0 in model 2, increasing RD by 1% will increase

RETURN by β1%, otherwise it will lower RETURN by β1%.

Both models are being regressed six times. Once for all the firms with and without a firm fixed effect, once for the hi-tech group with and without firm fixed effect and once for the low-tech group with and without firm fixed effects.

4. Results

4.1. Main Results

The goal of this paper is to discover what the effect of R&D expenses are on both the stock price and stock return, which is done by applying multiple regressions on the sample

containing every listed company in North-America reporting positive R&D expenses. There is being controlled for capital expenses, firm size, cash flow and a yearly dummy has been created to check for yearly market growth or big events, such as the credit crunch.

Estimates of equation (1) are in Table 2, with stock price as a dependent variable. The relationship between stock price and R&D is positive significant at 1%. This means that firms with high levels of R&D tend to have higher stock prices. Once controlled for firm fixed

(11)

11 effects, the relationship becomes insignificant, meaning that once a given firm changes its R&D expenses, the stock price will not change. Only the share price of high-tech firms is significantly affected, at a 10% significance level. Yet this effect is negative. A 1% increase in R&D reduces the stock price by 0.055%. This means that the market has as much

information as the firm, when it comes to investment opportunities. The changes in R&D have already been taken into account to determine the stock prices. If companies want to invest more in R&D, it will be seen as overinvestment and decreases the stock price. The findings are in line with the findings of Stein(1988) and Hall et al. (1993). According to them it shows that shareholders tend to only have a short time horizon and do not valuate R&D expenses correctly.

Further, control variable Ln(CAPX) is shown to be significant in five out of six regressions (table 2) only showing a positive relationship between capital expenditure and stock price. This means that investments in, for example new machines, is seen as a positive influence on future cash flows and thus is positively reflected on stock price. Firm size has shown to have a significantly positive relationship with stock price in all regressions.

Increasing the size of the firms’ total assets will also increase the stock price, meaning that the resources available to generate future income are shown of importance to determine stock price. Findings indicate the same as findings by Lee and Chen in 2009. They claim that larger companies are able to enjoy economies of scale and scope, causing higher profits. So,

according to them, increasing firm size leads to high stock prices due to the higher profits that will be reaped. Cash flows show a significantly positive relationship between stock prices as well. As shown by Chen et al (2013), cash flows are an important factor for investors and their expectations. Increasing total cash flows will also increase the firms’ stock price.

In equation (2), where stock return is used as the dependent variable (Table 3), there is no significant relationship found between R&D expenses and stock return. This means that having high amounts of R&D expenses has no effect on stock returns. After controlling for firm fixed effects, the relationship becomes negatively significant for all firms and high-tech firms. The values indicate an increase of 1% in R&D expenses will decrease stock returns by 0.017% and by 0.028% for high-tech firms. This shows the market expects that if firms increase R&D investments, they are overinvesting. That could mean they are wasting money and get involved in projects that decrease stock prices over time. These results go against the findings of Chan et al. (1990) stating a positive effect on future returns and especially a bigger effect on high-tech firms.

(12)

12

Table 2: Results equation (1)

Dependent variable: Ln(PRICE)

(1) (2) (3) (4) (5) (6) Ln(RD) 0,056*** 0,086*** 0,082*** -0,016 -0,055* -0,002 (9,96) (6,59) (11,43) (-1,07) (-1,95) (-0,09) Ln(CAPX) 0,047*** 0,071*** 0,019 0,054*** 0,076*** 0,040*** (4,94) (4,67) (1,55) (5,49) (5,15) (3,02) Ln(SIZE) 0,376*** 0,388*** 0,369*** 0,086*** 0,065** 0,103*** (95,37) (51,52) (79,40) (4,60) (2,32) (4,09) Ln(CF) 0,257*** 0,196*** 0,297*** 0,163*** 0,117*** 0,209*** (24,17) (13,45) (19,50) (18,82) (10,77) (16,25) Intercept 1,039*** 0,985*** 1,178*** 2,436*** 2,209*** 2,580*** (13,03) (9,09) (9,99) (22,93) (15,43) (17,21)

Year effects Yes Yes Yes Yes Yes Yes

Firm fixed effects No No No Yes Yes Yes

R² 0,454 0,411 0,465 0,315 0,233 0,348

Observations 20540 7699 12841 20540 7699 12841

*, ** and ***respectively indicate a 10%, 5% and 1% significance level. Regressions (1) and (4) represent all firms, (2) and (5) represent the high-tech firms and (3) and (6) represent low-tech firms.

Further, the control variables tend to be negative. In all six regressions Ln(CAPX) has a significantly negative value which means a negative relationship with stock returns. This shows the opposite relationship as seen in Table 2, meaning an increase in stock price once capital expenses increase, but a declining effect on growth. When not controlling for firm fixed effects, firm size has a positive relation with all firms and low-tech firms. Once

controlling for them, the relationship becomes negative for all 3 groups. If a firm increases its total assets, stock returns will decrease and the growth in stock price will decrease over time. This is an interesting finding, since in Table 2 increasing firm size shows a significantly positive relationship. A possible explanation for these findings can be that once a firm grows, its stock price increases. But once that has happened, it requires more growth effectively to keep the stock price increasing by the same percentage, so the growth decreases over time. Cash flows show a significantly positive relation with stock return in the regressions without firm fixed effects, but no effect for high-tech firms. Once controlled for firm fixed effects, the relation becomes negative at 1% significance. This again shows an opposite relationship to the findings from equation (2), showing that an increase in total cash flow will increase stock price but the growth will decline over time. The values of Ln(CF) in Table 2 and Table 3 are

(13)

13 respectively higher and lower than the values of the other control variables, showing a greater increase in stock price and a slighter decline in stock returns.

Dependent variable: Ln(RETURN)

(1) (2) (3) (4) (5) (6) Ln(RD) -0,001 0,010 0,005 -0,017* -0,028* -0,155 (-0,27) (1,45) (1,50) (-1,68) (-1,65) (-1,24) Ln(CAPX) -0,029*** -0,027*** -0,031*** -0,088*** -0,075*** -0,088*** (-5,93) (-3,70) (-4,65) (-10,37) (-6,53) (-7,36) Ln(SIZE) 0,004** 0,003 0,004** -0,220*** -0,234*** -0,208*** (2,31) (0,88) (2,03) -(17,25) -(13,17) (-11,71) Ln(CF) 0,009* -0,005 0,019** -0,055*** -0,066*** -0,042*** (1,67) (-0,69) (2,49) (-7,84) (-7,75) (-3,83) Intercept -0,190*** -0,377*** 0,047 0,772*** 0,500*** 1,006*** (-3,66) (-5,34) (0,65) (9,47) (4,92) (8,26)

R² 0,2605 0,323 0,2328 0,055 0,102 0,046

Observations 20057 7462 12595 20057 7462 12595

*, ** and *** respectively indicate a 10%, 5% and 1% significance level. Regressions (1) and (4) represent all firms, (2) and (5) represent the high-tech firms and (3) and (6) represent low-tech firms.

4.2 Robustness check

In this section, several alternations to the original equations (1) and (2) will be made, to see how the coefficients alter. Firstly, RD will be used without logarithmic transformation, namely:

𝐿𝑛(𝑃𝑅𝐼𝐶𝐸)𝑖,𝑡 = 𝛼𝑖 + 𝛽1𝑅𝐷𝑖,𝑡−1+ 𝛽2𝐿𝑛(𝐶𝐴𝑃𝑋)𝑖,𝑡−1+ 𝛽3𝐿𝑛(𝑆𝐼𝑍𝐸)𝑖,𝑡−1+ 𝛽4𝐿𝑛(𝐶𝐹)𝑖,𝑡−1+

𝜇_𝑡+ 𝜀_𝑖,𝑡(3)

𝐿𝑛(𝑅𝐸𝑇𝑈𝑅𝑁)_𝑖,𝑡 = 𝛼_𝑖 + 𝛽₁𝑅𝐷_{𝑖,𝑡−1}+ 𝛽₂𝐿𝑛(𝐶𝐴𝑃𝑋)_{𝑖,𝑡−1}+ 𝛽₃𝐿𝑛(𝑆𝐼𝑍𝐸)_{𝑖,𝑡−1}+ 𝛽₄𝐿𝑛(𝐶𝐹)_{𝑖,𝑡−1}+ 𝜇_𝑡+ 𝜀_𝑖,𝑡(4)

(14)

14 Secondly, Ln(RD)² will be added to the original equations, consider the models:

𝐿𝑛(𝑃𝑅𝐼𝐶𝐸)𝑖,𝑡 = 𝛼𝑖 + 𝛽1𝐿𝑛(𝑅𝐷)𝑖,𝑡−1+ 𝛽2𝐿𝑛(𝑅𝐷)²𝑖,𝑡−1+𝛽3𝐿𝑛(𝐶𝐴𝑃𝑋)𝑖,𝑡−1

+ 𝛽4𝐿𝑛(𝑆𝐼𝑍𝐸)𝑖,𝑡−1+ 𝛽5𝐿𝑛(𝐶𝐹)𝑖,𝑡−1+ 𝜇𝑡+ 𝜀𝑖,𝑡 (5)

𝐿𝑛(𝑅𝐸𝑇𝑈𝑅𝑁)𝑖,𝑡 = 𝛼𝑖+ 𝛽1𝐿𝑛(𝑅𝐷)𝑖,𝑡−1+ 𝛽2𝐿𝑛(𝑅𝐷)²𝑖,𝑡−1+𝛽3𝐿𝑛(𝐶𝐴𝑃𝑋)𝑖,𝑡−1+

𝛽4𝐿𝑛(𝑆𝐼𝑍𝐸)𝑖,𝑡−1+ 𝛽5𝐿𝑛(𝐶𝐹)𝑖,𝑡−1+ 𝜇𝑡+ 𝜀𝑖,𝑡(6)

Finally, Ln(FRD) is added to the original equations:

𝐿𝑛(𝑃𝑅𝐼𝐶𝐸)𝑖,𝑡 = 𝛼𝑖 + 𝛽1𝐿𝑛(𝑅𝐷)𝑖,𝑡−1+ 𝛽2𝐿𝑛(𝐹𝑅𝐷)𝑖,𝑡+𝛽3𝐿𝑛(𝐶𝐴𝑃𝑋)𝑖,𝑡−1

+ 𝛽4𝐿𝑛(𝑆𝐼𝑍𝐸)𝑖,𝑡−1+ 𝛽5𝐿𝑛(𝐶𝐹)𝑖,𝑡−1+ 𝜇𝑡+ 𝜀𝑖,𝑡 (7)

𝐿𝑛(𝑅𝐸𝑇𝑈𝑅𝑁)𝑖,𝑡 =

𝛼_𝑖 + 𝛽₁𝐿𝑛(𝑅𝐷)_{𝑖,𝑡−1}+ 𝛽₂𝐿𝑛(𝐹𝑅𝐷)_𝑖,𝑡+𝛽₃𝐿𝑛(𝐶𝐴𝑃𝑋)_{𝑖,𝑡−1}+ 𝛽₄𝐿𝑛(𝑆𝐼𝑍𝐸)_{𝑖,𝑡−1}+ 𝛽₅𝐿𝑛(𝐶𝐹)_{𝑖,𝑡−1}+ 𝜇_𝑡+ 𝜀_𝑖,𝑡(8)

Equations (3) and (4) are constructed to see what difference it makes to control for large outliers and non-linearity by applying a logarithmic transformation. The estimates are in Table 4 for equation (3) and in Table 5 for equation (4). Table 4 only shows a significant relationship between R&D expenses for low-tech firms after controlling for firm fixed effects and it is negative. This means if a firm increases R&D expenses, its stock price will decrease. This is in line with findings by Hall et al. (1993) who argue that shareholders tend to see R&D expenses as expenses rather than investments. In Table 2, four of the six regressions show a significant effect, so applying the logarithmic transformation increases significance. The control variables still show the same significant relationship as for equation (1).

(1) (2) (3) (4) (5) (6)

RD -0,000 0,000 -0,019 -0,001 0,002 -0,138*

(-1,61) (1,61) (-1,53) (-0,08) (0,35) (-1,81) Ln(CAPX) 0,042*** 0,082*** 0,009 0,054*** 0,067*** 0,047***

(15)

15 Ln(SIZE) 0,370*** 0,385*** 0,361*** 0,091*** 0,078*** 0,101*** (99,36) (52,57) (80,42) (5,18) (2,93) (4,07) Ln(CF) 0,262*** 0,199*** 0,313*** 0,159*** 0,115*** 0,204*** (24,60) (13,83) (20,55) (18,44) (10,58) (15,82) Intercept 0,890*** 0,839*** 0,930*** 2,425*** 2,223*** 2,589*** (11,39) (7,71) (8,14) (23,07) (15,31) (16,66)

R² 0,454 0,412 0,461 0,325 0,257 0,330

Observations 20918 7812 13106 20918 7812 13106

Looking at Table 5, a significantly negative relationship between R&D expenses and stock returns is shown before controlling for firm fixed effects. Afterwards, only low-tech firms show a significantly negative relationship, while in Table 3 all firms and high-tech firms show a significant value. The results are completely different form equation (2). Here an increase in R&D expenses, for low-tech firms, will decrease stock price and decrease growth in stock price over time. This would mean that the market is indifferent about R&D expenses in case of all firms or high-tech firms, but believes that low-tech firms will generate less future cash flows than they currently do. For the control variables the significant relationship remains the same as in Table 3.

(1) (2) (3) (4) (5) (6) RD -0,001*** -0,001*** -0,011*** 0,002 0,002 -0,157*** (-10,49) (-11,15) (-6,33) (0,34) (0,98) (-3,46) Ln(CAPX) -0,026*** -0,024*** -0,28*** -0,087*** -0,079*** -0,082*** (-5,49) (-3,36) (-4,35) (-10,58) (-6,93) (-7,28) Ln(SIZE) 0,004** -0,000 0,003 -0,212*** -0,232*** -0,205*** (2,14) (-0,05) (1,54) (-17,50) (-13,36) (-11,91) Ln(CF) 0,010* -0,002 0,022*** -0,057*** -0,067*** -0,043*** (1,87) (-0,33) (2,86) (-8,15) (-7,84) (-4,06) Intercept -0,169*** -0,372*** 0,061 0,788*** 0,537*** 1,076*** (-3,31) (-5,52) (0,85) (9,56) (5,09) (8,59)

(16)

16

Observations 20431 7574 12857 20431 7574 12595

Equations (5) and (6) are constructed to see if the effect of R&D expenses are either increasing or decreasing. The estimates presented in Table 6 show the estimates for the effect on stock price. Ln(RD) shows a significantly negative relationship between stock price before controlling for firm fixed effects in cases of all firms and low-tech firms. This is the opposite of what Table 2 reports. Once controlled for firm fixed effects the relationship between R&D expenses and stock prices becomes significant for all three regressions, meaning if a firm increases its R&D expenses it will decrease its stock price. The relationship between Ln(RD)² and stock price is significantly negative for all six regressions. This means that the negative effect an increase in R&D has on stock price will be increased. In this case, the bigger the overinvestment, the greater the stock price drop will be. Control variables show significance in the same direction in Table 6 as in Table 2.

(1) (2) (3) (4) (5) (6) Ln(RD) -0,148*** -0,053 -0,158*** -0,117*** -0,166** -0,085* (-6,28) (-1,11) (-4,90) (-2,87) (-2,13) (-1,67) Ln(RD)² -0,025*** -0,023*** -0,027*** -0,011*** -0,016* -0,008* (-9,09) (-3,17) (-7,73) (-2,84) (-1,74) (-1,83) Ln(CAPX) 0,053*** 0,080*** 0,023* 0,058*** 0,080*** 0,043*** (5,63) (5,13) (1,86) (5,83) (5,42) (3,18) Ln(SIZE) 0,370*** 0,382*** 0,364*** 0,070*** 0,050 0,091*** (93,43) (49,15) (78,84) (3,53) (1,64) (3,44) Ln(CF) 0,267*** 0,201*** 0,312*** 0,164*** 0,118*** 0,209*** (25,06) (13,65) (20,64) (18,84) (10,82) (16,23) Intercept 0,771*** 0,882*** 0,787*** 2,351*** 2,149*** 2,476*** (8,96) (7,63) (6,04) (21,56) (14,53) (16,06)

R² 0,458 0,412 0,470 0,291 0,208 0,348

Observations 20540 7699 12841 20540 7699 12841

(17)

17 The estimates for equation (6) are given in Table 7. After controlling for firm fixed effects, a negative relationship between Ln(RD) and stock returns is shown for all firms and low-tech firms. This indicates that changing R&D expenses of a high-tech firm does not influence growth, while for low-tech firms it means a decrease in stock returns if R&D

expenses increase. Ln(RD)² shows the same relationships as Ln(RD), meaning an increasingly negative growth in case of all firms and low-tech firms. The markets have the same amount of information as the firms do and believe that increasing R&D expenses will be an

overinvestment, leading to increasing negative effect on future cash flows, which is reflected in the stock price, and a decrease in stock returns in case of all firms and low-tech firms that is also increasing. The control variables show the same significant relationship as in Table 3.

(1) (2) (3) (4) (5) (6) Ln(RD) -0,019* -0,018 -0,004 -0,077*** -0,033 -0,099*** (-1,74) (-0,90) (-0,25) (-2,81) (-0,80) (-2,73) Ln(RD)² -0,002* -0,005 -0,001 -0,007*** -0,001 -0,008** (-1,72) (-1,53) (-0,64) (-2,43) (-0,13) (-2,49) Ln(CAPX) -0,028*** -0,025*** -0,030*** -0,086*** -0,074*** -0,086*** (-5,80) (-3,45) (-4,62) (-10,08) (-6,47) (-7,15) Ln(SIZE) 0,003** 0,002 0,004* -0,229*** -0,235*** -0,220*** (2,00) (0,49) (1,93) (-16,90) (-12,28) (-11,74) Ln(CF) 0,010* -0,004 0,020* -0,055*** -0,066*** -0,042*** (1,81) (-0,56) (2,54) (-7,80) (-7,75) (-3,81) Intercept -0,214*** -0,398*** 0,032 0,721*** 0,499*** 0,902*** (-4,00) (-5,52) (0,42) (8,70) (4,89) (7,31)

R² 0,261 0,323 0,233 0,053 0,101 0,043

Observations 20057 7462 12595 20057 7462 12595

Equations (7) and (8) have been constructed to see if there exists an anticipated effect of R&D expenses on stock price and stock returns. The estimates for equation (7) are in Table 8. Without controlling for firm fixed effects, the relationship between Ln(RD) and stock price is positively significant, but become insignificant when firm fixed effects are taken into account. Ln(FRD) shows a significantly negative relationship with stock price in case of all

(18)

18 firms and high-tech firms, but after controlling for firm fixed effects only high-tech firms’ stays significant. This could mean that R&D expenses from the previous years are less important than for the coming year when it comes to stock prices of high-tech firms. Relationships of the control variables stay the same as in Table 2.

(1) (2) (3) (4) (5) (6) Ln(RD) 0,080*** 0,218*** 0,047** -0,017 -0,037 -0,010 (4,67) (6,63) (2,26) (-1,08) (-1,18) (-0,55) Ln(FRD) -0,045*** -0,162*** 0,021 -0,024 -0,076** 0,001 (-2,62) (-4,74) (1,01) (-1,41) (-2,00) (0,04) Ln(CAPX) 0,029*** 0,039** 0,007 0,041*** 0,058*** 0,031** (2,92) (2,35) (0,53) (4,08) (3,74) (2,25) Ln(SIZE) 0,359*** 0,378*** 0,350*** 0,065*** 0,041 0,086*** (88,40) (47,45) (74,24) (3,18) (1,32) (3,23) Ln(CF) 0,258*** 0,191*** 0,302*** 0,153*** 0,114*** 0,192*** (23,31) (12,37) (19,38) (17,02) (9,72) (14,61) Intercept 1,059*** 0,889*** 1,257*** 2,459*** 2,178*** 2,608*** (13,03) (7,77) (10,42) (7,92) (12,89) (15,46)

R² 0,457 0,407 0,472 0,274 0,166 0,332

Observations 17491 6596 10895 17491 6596 10895

Table 8 shows the estimates for equation (8). Ln(RD) shows a negative relationship with stock return and it is significant for all groups after controlling for firm fixed effects. Ln(FRD) shows a significantly positive relationship with stock returns for all firms and low-tech firms and no significant relationship for high-low-tech firms. This could mean that the market finds future R&D expenses more important to estimate future cash flows than past R&D expenses. Other than that, it believes that past R&D expenses decreases firm growth, while future R&D expenses stimulate growth, except for high-tech firms. The control variables show the same significant relationships as in Table 3.

(19)

19

(1) (2) (3) (4) (5) (6) Ln(RD) -0,021** -0,007 -0,018* -0,035*** -0,052*** -0,030** (-2,46) (-0,45) (-1,69) (-2,85) (-2,66) (-2,01) Ln(FRD) 0,017** 0,011 0,023** 0,033*** 0,027 0,033** (1,98) (0,72) (2,15) (2,56) (1,27) (2,14) Ln(CAPX) -0,031*** -0,031*** -0,314*** -0,096*** -0,075*** -0,101*** (-5,88) (-3,90) (-4,39) (-10,73) (-6,25) (-7,87) Ln(SIZE) 0,001 0,000 0,001 -0,220*** -0,240*** -0,206*** (0,75) (0,14) (0,68) (-15,64) (-11,70) (-10,81) Ln(CF) 0,002 -0,010 0,010 -0,064*** -0,071*** -0,055*** (0,28) (-1,22) (1,22) (-8,90) (-7,61) (-5,06) Intercept -0,192*** -0,376*** 0,047 0,786*** 0,560*** 0,972*** (-3,56) (-5,12) (0,62) (8,55) (4,82) (7,20)

R² 0,283 0,346 0,255 0,064 0,115 0,055

Observations 17086 6389 10697 17086 6389 10697

5. Conclusion

The aim of this research was to find out what the effect of research and development expenses are on stock price and stock return. This has been done by regressing two different models. One model is being regressed with the stock price as the dependent variable and one with stock return as the dependent variable. Stock price is measured as the logarithm of close price in the end of the fiscal year and stock return is the logarithm of close price in the end of the fiscal year minus the logarithm of close price in the previous year.

Results are showing a positive effect of R&D expenses on stock price, but once controlled for firm fixed effects only the results for the high-tech firms are significant and indicate a negative relationship. This means, when it comes to high-tech firms, that markets and the firms have an equal amount of information regarding investment opportunities. Changes in R&D expenses are already taken into account in the stock prices in previous periods. If a company tries to invest more, it is not seen as a way to increase future cash flows and stock prices decrease.

In the case of stock returns, while controlling for firm fixed effects, all firms and high-tech firms show a significantly negative relationship, whereas the low-tech group did not

(20)

20 show a significant effect. This means markets expect if firms make wrong investment

decisions now, they are wasting money and get behind on competition, which could lead to a decrease in stock price over time.

Relating the results back to existing literature, they are not completely in line with what has been found before. Studies by Chan et al. (1990) and Lee and Chen (2009), state that there is a positive effect of R&D expenses on stock price and stock return. Before correcting for firm fixed effects, the same results were found in case of stock price. But once corrected, the coefficients become either negative or insignificant. On the other hand, Hall et al. (1993) and Stein (1988) argue that it will have a negative effect, because of the short time horizon that investors have. The results tend to be in line with these findings.

Regarding this research, there are a couple of limitations. Here the expenses are taken as a ratio over the lagged value of total assets. Even though firm size has shown to be an important factor within stock price and stock return, it is unclear if it is the best factor to use for the ratio. Furthermore, only the companies that reported positive R&D expenses were used in the sample. Therefore this study might suffer from a potential bias. Adding the companies without R&D expenses might give a different insight. Also, in this study the sample has been divided into high-tech and low-tech firms. Different sectors, such as pharmaceutical sectors, also tend to be dependent on their R&D facilities, so using only the two groups might not have been enough.

For further research, the different sectors could be taken into account instead of only high-tech and low-tech. Instead of taking R&D as a ratio over total assets from the last period, there could be experimented with R&D over sales or profits. Moreover, adding more

variables, such as firm age, may improve the models along with their outcomes. Furthermore, the robustness check showed some interesting results when it comes to adding Ln(RD)² and Ln(FRD), they require further research along with adding for example Ln(FRD)². Finally controlling for stock splits, or similar factors heavily influencing stock price, might improve the estimates of the models.

(21)

21 References

Branch, B. (1974). Research and Development Activity and Profitability: A Distributed Lag Analysis. Journal of Political Economy, 82(5), pp.999-1011.

Chan, L., Lakonishok, J. and Sougiannis, T. (2001). The Stock Market Valuation of Research and Development Expenditures. J Finance, 56(6), pp.2431-2456.

Chan, S., Martin, J. and Kensinger, J. (1990). Corporate research and development

expenditures and share value. Journal of Financial Economics, 26(2), pp.255-276. Chen, L., Da, Z. and Zhao, X. (2013). What Drives Stock Price Movements?. Review of

Financial Studies, 26(4), pp.841-876.

Cohen, W. and Levinthal, D. (1989). Innovation and Learning: The Two Faces of R & D. The

Economic Journal, 99(397), p.569.

Doukas, J. and Switzer, L. (1992). The stock market's valuation of R&D spending and market concentration. Journal of Economics and Business, 44(2), pp.95-114.

Hall, Bronwyn H., and Robert E. Hall, 1993, The value and performance of U.S. corporations, Brookings Papers on Economic Activity 1, 1–34

Lee, R. and Chen, Q. (2009). The Immediate Impact of New Product Introductions on Stock Price: The Role of Firm Resources and Size *. Journal of Product Innovation

Management, 26(1), pp.97-107.

Makadok, R. (2010). The Interaction Effect of Rivalry Restraint and Competitive Advantage on Profit: Why the Whole Is Less Than the Sum of the Parts. Management Science, 56(2), pp.356-372.

Pandit, S., Wasley, C. and Zach, T. (2011). The Effect of Research and Development (R&D) Inputs and Outputs on the Relation between the Uncertainty of Future Operating Performance and R&D Expenditures. Journal of Accounting, Auditing & Finance, 26(1), pp.121-144.

Stein, J. (1988). Takeover Threats and Managerial Myopia. Journal of Political Economy, 96(1), p.61.

(22)

22 Appendix A: Definitions and formulas

Name Definition Formula

Ln(PRICE) Logarithm of close price in the end of the fiscal year.

Ln(Price closet)

Ln(RETURN) _{Logarithm of close price in the end of} the fiscal year minus the logarithm of close price in the previous year.

Ln(Price closet

)-Ln(Price closet-1)

Ln(RD) _{Logarithm of R&D expenses in the} current year over total assets in the previous year.

Ln(R&D expensest/Total

assetst-1)

Ln(CAPX) _{Logarithm of capital expenditures in} the current year divided by total assets from last year.

Ln(Capital expensest/Total

assetst-1)

Ln(SIZE) _{Logarithm of total assets in the current} year.

Ln(Total assetst)

Ln(CF) _{Logarithm of total cash flows in the} current year over total assets from the previous year.

Ln(Total cash

flowst/Total assetst-1)

RD

R&D expenses in the current year over total assets in the previous year.

R&D

expensest/Total

assetst-1

Ln(RD)² Square of Ln(RD). Ln(RD)²

Ln(FRD) _{Logarithm of R&D expenses in the} coming year over total assets in the current year.

Ln(R&D

expensest+1/Total