Style investing and firm innovation

Style investing and firm innovation

Accepted Manuscript Title: Style Investing and Firm Innovation Authors: Koray Sayili, Gokhan Yilmaz, Douglas Dyer, A. Melih Kull ¨ u¨ PII: DOI: Refere...

647KB Sizes 2 Downloads 40 Views

Accepted Manuscript Title: Style Investing and Firm Innovation Authors: Koray Sayili, Gokhan Yilmaz, Douglas Dyer, A. Melih Kull ¨ u¨ PII: DOI: Reference:

S1572-3089(17)30266-8 http://dx.doi.org/10.1016/j.jfs.2017.08.005 JFS 569

To appear in:

Journal of Financial Stability

Received date: Revised date: Accepted date:

12-4-2017 15-8-2017 22-8-2017

Please cite this article as: Sayili, Koray, Yilmaz, Gokhan, Dyer, Douglas, Kull ¨ u, ¨ A.Melih, Style Investing and Firm Innovation.Journal of Financial Stability http://dx.doi.org/10.1016/j.jfs.2017.08.005 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Style Investing and Firm Innovation Koray Sayili a, Gokhan Yilmaz b,1, Douglas Dyer c and A. Melih Küllü d a

Department of Economics & Management, DePauw University, Greencastle IN 46135 [email protected] b Labovitz School of Business and Economics, University of Minnesota Duluth, Duluth MN 55812 [email protected] c College of Business Administration, Texas A&M University - Central Texas, Killeen, TX 76549 [email protected] d College of Business Administration, University of Central Florida, Orlando, FL 32816 [email protected]

Abstract We document that transient, dedicated and quasi-indexed institutional investors exhibit a high degree of within-group heterogeneity with respect to their investment styles (i.e., growth, value, and balanced). We find that growth institutional investors enhance firm innovation in terms of R&D expenditures, R&D intensity, quantity and quality of patents and patent radicalness while value institutional investors impede innovation. Balanced investors have no significant association with innovation. Findings are consistent with style investing literature that growth and value styles are substitutes. Using investment styles, we present evidence that reconcile literature’s mixed findings on how transient and dedicated investors affect R&D and innovation, and why quasi-indexed investors, the largest group among all investors, have an insignificant effect. We also show that the effect of institutional investors depends on the firm’s relative level of innovativeness.

Keywords: Innovation, institutional ownership, style investing, R&D, patents. JEL Classifications: O31, O32, G23, G32

1. Introduction

1

Corresponding author. We would like to thank the participants at 2013 Financial Management Association Annual Meeting at Chicago, research seminar participants at Rensselaer Polytechnic Institute and Queen’s University. We also express our gratitude to two anonymous referees and the managing editor Iftekhar Hasan for their invaluable comments and suggestions on the earlier draft. All remaining errors and omissions are ours.

Institutional investors are major powerhouses that greatly influence the course of corporate development. Their control over the equity markets has risen drastically in the last few decades. According to recent statistics presented in Heineman & Davis (2011), institutional investors control over 73% of equity in top 1,000 US firms in 2009. The significance of institutional investors is evident in international data as well. For instance, according to Gonard, Kim & Ynesta (2008), total asset size of institutional investors in 17 OECD countries reached $40.3 trillion in 2005. A more recent statistics announced by IPE (Investments & Pensions Europe) webpage2 shows that institutional investors have been growing steadily and the top global 400 institutional investors’ total asset size reached to almost $60 trillion in 2016.3 Institutional investors’ ability to invest large amounts of capital into firms as shareholders provides them the opportunity to exert control and have a strong say in many corporate decisions. One of the decisions that attracted the attention of scholars in the last two decades is the firms’ R&D related efforts. How do institutional investors affect firms’ R&D investments? The evidence presented in the current literature on this subject is mixed. Some scholars (e.g. Mitroff, 1987; Graves, 1988) find a negative effect, some (e.g. Francis & Smith, 1995; Wahal & McConnell, 2000; Brossard, Lavigne & Sakinc, 2013; Rong, Wu & Boeing, 2017) find a positive effect and some (e.g. Kochhar & David, 1996) find no statistical relationship between institutional stock holdings and R&D investments. These mixed findings are not surprising and they potentially stem from at least three reasons: Firstly, not all institutional investors share the same goals or want to participate in the decision-making process in similar fashion. Secondly, the effect of institutional investors may differ based on the level of innovativeness (or R&D investment level) of the invested firm. Lastly, it is hard to argue that the relationship is unidirectional. In other words, it is hard to distinguish whether the involvement of institutional investors incentivizes firms to innovate more or the institutional investors choose firms based on innovative ability and success. In order to address the first issue mentioned above, Bushee (1998) categorizes the institutional investors based on their investment horizon as transient, dedicated, and quasi-indexed 2

http://www.ipe.com/reports/special-reports/top-400-asset-managers/top-400-asset-managers-2016-global-assetsnow-563trn/10013542.article 3 Recent studies show that foreign institutional investors tend to invest more in equity markets where governance standards are similar to their home countries (e.g. Abdioglu, Khurshed & Stathopoulos, 2013) and countries with high level of social trust (e.g. Jin et al., 2016).

institutional investors. In his seminal paper, Bushee (1998) finds that while dedicated institutional investors positively affect R&D investment, transient institutional investors affect it negatively.4 Kochhar & David (1996) finds that while overall institutional ownership has no significant effect, the pressure-resistant institutional investors have positive effect on innovation. A recent study by Aghion, Van Reenen & Zingales (2013) show that while quasi-indexed institutions have no significant effect, both transient and dedicated institutions have positive effect on cite-weighted patents. Since R&D expenditures are the inputs to patent development, the mixed findings of transient investors affecting R&D expenditures negatively (i.e., Bushee, 1998) but cite-weighted patents positively (i.e., Aghion et al., 2013) is worth investigating. In addition, it is surprising that investors with opposite investment behavior affect cite-weighted patents in similar fashion. Another interesting finding of both papers is that quasi-indexed institutional investors have no significant effect on firm’s innovative activities. Since more than two-thirds of the institutional investors are classified as quasi-indexed, further investigation of how institutional investors impact innovation is needed. In this paper, we address all three issues listed above and provide possible explanations to the findings of Bushee (1998) and Aghion, Van Reenen & Zingales (2013). In doing this, we seek the answers for the following questions: Is there a significant relationship between the level of institutional ownership and innovation? Which type of institutional investors support or hinder innovation? When they invest into firms, do institutional investors influence innovation by taking into account the relative level of innovativeness? Do institutional investors care more about the quality and/or the quantity of innovations? In answering these questions, we benefit from an important characteristic of institutional investors: The way they allocate their funds among (growth and value) stocks, which defines how they reach their expected return (i.e., through capital gain or dividends) goals. This decision is known as style investing. (Capaul, Rowley & Sharpe, 1993; Bernstein, 1995; Swensen, 2000; Barberis & Shleifer, 2003; and Teo & Woo, 2003). For example, an institutional investor whose clientele prefer to defer taxes would invest more in growth firms since tax on capital gains is postponed until realized. On the other hand, in order to generate income for paying off periodic 4

The negative effect of transient institutional investors arises from their myopic behavior. They target higher shortterm returns and shift managerial focus from R&D to earnings and accruals at the expense of lower long-term stock returns. For evidence on accruals and stock returns, see Peng, Yan & Yan (2016).

retirement benefits, pension funds would invest more in value firms, which distribute dividends on a regular basis. Mullainathan (2000) argues that classifying assets into categories facilitate fairly easier processing of large amounts of information, making the investment decision-making a simpler task. Sharpe (1992) suggests that asset classes help evaluating the performance of money managers since they create comparable groups of managers, which invest in the same asset classes. Similarly, Barberis & Shleifer (2003) argue that institutional investors, which invest in the same style share many common characteristics, advertise themselves as following a particular style and restrict themselves to selecting stocks within that particular style. Froot & Teo (2008) find strong evidence that institutional investors reallocate their funds between style categories more intensively than random stock groupings. Moreover, style investing helps predict future stock returns after controlling for size, book-to-market and past stock returns as shown by Wahal & Yavuz (2013). An important note that we need to make is that the Bushee (1998) classification is concentrated around quasi-indexed institutions and, by design; it does not take into account the investment styles of institutional investors. A cross-tabulation of Bushee (1998) classification and Bushee & Goodman (2007) classification that takes into account the investment styles of institutional investors reveals that there is a high degree of within-group heterogeneity in each of the transient, dedicated and quasi-indexed groupings in terms of investment styles (Figure 1). Bushee & Goodman (2007) classifies institutional investors into three groups based on their investment styles: Institutions that invest in growth firms, institutions that invest in value (income) firms and institutions that invest in balanced firms. Such categorization may capture another dimension of how differently institutional investors function and their differing motivations.5 Therefore, we use style investing in explaining the mixed findings that existing literature finds with respect to institutional stock holdings, R&D and innovation. [Figure 1 Here] Figure 1 shows that while growth institutional investors are the largest subgroup within both the dedicated and transient investors, balanced (growth & income) investors are the largest subgroup 5

A recent study by Huang & Paul (2017) show that growth institutional investors and value institutional investors are quite different in selecting stocks for their portfolios. According to their results, the growth investors target fast growing firms while the value investors are interested in high dividend paying firms.

within the quasi-indexed investors. We believe that this is a potential explanation of why Aghion, Van Reenen & Zingales (2013) find a positive effect on innovation for transient and dedicated institutional owners but not for quasi-indexed investors. In that regard, our findings show that institutional investors reveal their preferences for R&D investment and innovation through their investment styles. Our empirical analysis in this paper yields the following important findings: First and foremost, we find that growth institutional investors and value institutional investors have opposite and significant effects on almost all firm innovativeness measures. Specifically, the effect of growth institutional investors is positive on R&D expenditures, R&D intensity, granted patent counts, cite-weighted patents, and patent quality measures while value institutional investors have a negative influence on all of these measures. Balanced investors do not have any statistically significant effect on these measures. Secondly, we find that the effect of institutional investors on future innovation changes based on the relative innovativeness of the firms that they invest. In more detail, while growth institutional investors have a positive effect on all firms, value institutional investors do not have a significant effect on firms with below median R&D investment and number of patents. However, their effect is negative and significant for the firms with above median R&D investment and number of patents. Our conjecture for this finding is that value institutional investors may be revealing their preference for dividend distributions when firms make relatively large investments in R&D and innovation, which requires earnings retention. Finally, we show that growth (value) investors motivate (impede) firms to innovate more radical patents. This paper relates to two strands of literature. The first strand investigates the factors affecting the R&D and firm innovativeness. Some of the highly-cited factors that are believed to affect firm innovativeness are firm size (e.g. Kochhar & David, 1996; Bhattacharya & Bloch, 2004; Barker & Mueller, 2002), liquidity (Kochhar & David, 1996; Brown & Petersen, 2011; Brown, Martinsson & Petersen, 2012), product market competition (Aghion et al., 2005; Matsumura, Matsushima & Cato 2013), innovative track record such as patent stock or R&D stock (e.g. Shefer & Frenkel, 2005; Czarnitzki & Toole, 2011; Hottenrott, Hall & Czarnitzki, 2016), executive compensation incentives (e.g. Cheng, 2004; Lerner & Wulf, 2007; Francis, Hasan &

Sharma, 2011; Cassell et al., 2012) and CEO overconfidence (e.g. Hirshleifer, Low & Teoh, 2012; Galasso & Simcoe, 2011; Herz, Schunk & Zhender, 2014). The second strand of literature investigates the effect of institutional owners on firm’s innovativeness. As we mentioned previously, there is mixed evidence on this issue and this may stem from the existence of two competing hypotheses: The first hypothesis argues that institutions lead to managerial myopia due to their short-term orientation or lack of ability to evaluate the long-term value creation capacity of firms and lead to R&D cut-backs (e.g. Mitroff, 1987; Graves, 1998; Bushee, 1998; Holden & Lundstrum, 2009). In this hypothesis, managerial myopia may be attributed to transient institutional investors due to their short-term investment horizon while the opposite may be true for dedicated institutional investors. The findings by the proponents of this hypothesis imply that this type of investment behavior may stem from the frequent performance evaluation of institutional fund managers. The second hypothesis argues that institutional investors provide additional monitoring of management, consequently alleviate the agency problems and channel more funds to R&D (e.g.; Hansen & Hill, 1991; Francis & Smith, 1995; Wahal & McConnell, 2000; Eng & Schakell, 2001; Brossard, Lavigne & Sakinc, 2013). Based on this hypothesis, institutional investors enhance monitoring on managerial performance and this can lead to better results on R&D investments and firm innovation. In fact, a recent study by McCahery, Sautner & Starks (2016) show that intense monitoring and intervention are used by many institutional investors, especially the long-term (i.e., dedicated) institutional investors. It is important to note that institutional investors’ impact on the firms that they invest is not limited to R&D and innovation. Switzer & Wang (2017) find that ownership by short-term institutional investors reduces firms’ credit spreads during normal economic periods while ownership by long-term institutional investors has a similar effect during the crisis period. They argue that this is consistent with their monitoring and information production roles. Davies et al. (2014), using an analytical framework, argue that output by firms would have been 20% higher if short-termism was eliminated which is consistent with the managerial myopia hypothesis. Finally, An & Zhang (2013) find that while long-term institutional investors significantly decrease stock price synchronicity and crash risk, the short-term institutional investors have the opposite effect on these variables.

The contribution of our paper to the literature is three-fold. Firstly, different from the existing literature on institutional investors and innovation, we use a style investing based classification of institutional investors, developed by Bushee & Goodman (2007). The use of this classification allows us to approach the subject from a different angle and helps us offer a potential explanation for the puzzling finding of Aghion, Van Reenen & Zingales (2013) as previously mentioned. Secondly, our empirical analysis is more comprehensive than other studies as it investigates the effect of institutional investors on inputs (e.g. R&D expenditure and R&D intensity), outputs (e.g. number of patents and cite-weighted patents) and radicalness (e.g. citations received to citations made ratio and zero citation made patents) of innovation. Especially the results on radicalness of innovations are not only interesting but also novel, as the current literature pays little attention to this aspect of innovation. Lastly, our results show that the effect of institutional investors (based on their investing style) change based on firms’ relative level of innovativeness. To the best of our knowledge, this is the first study that provides such empirical evidence. The rest of this paper is organized as follows: Section 2 explains the data and descriptive statistics. Section 3 presents the empirical models and results. Section 4 discusses the robustness of our empirical models and Section 5 concludes. 2. Data We use the Thompson Reuters Investment Company Data to obtain institutional holdings information. The primary source of the institutional holdings data is the 13F forms that investment companies are required to file with the SEC every calendar quarter.6 Only the companies with over $100 million under their management are required to file, but others may still choose to file. Our source of information on company financials is the Compustat database. Innovation data is the patent data, which is publicly available from NBER (National Bureau of Economic Research).7 NBER Patent Data includes firm level patent data that shows the patent grant year, patent counts, citations made to other patents and citations received from other patents. We also

6

An institutional investment company is an entity that either invests in, or buys and sells, securities for its own account; or a natural person or an entity that exercises investment discretion over the account of any other natural person or entity. 7 The details of the NBER patent data are explained by Hall, Jaffe & Trajtenberg (2001).

calculate citation-weighted patents because patent counts alone do not convey information regarding the quality of the patents as suggested by Trajtenberg (1990). We also use data from the CRSP (The Center for Research in Security Prices) and I/B/E/S (Institutional Broker’s Estimate System) to calculate useful company characteristics (i.e., stock turnover, CRSP age of firms, S&P500 membership dummy and number of analysts covering the firms). In addition to above datasets, we use the Bushee & Goodman (2007) classification that groups institutional investors with respect to their investment styles (i.e. growth, value and balanced). Our sample period covers years 1991 through 1999. Institutional holdings data is not reliable for periods before 1991. In order to be consistent with previous research that we closely build upon (i.e., Aghion, Van Reenen & Zingales, 2013), we end our sample on 1999. We start preparing our data by first calculating the ownership percentage of each institutional investor in each firm in each calendar quarter. We sum the ownership percentages across all institutional investors that hold a firm in each quarter to get the total ownership percentage of that firm held by institutions in that particular quarter. Then, we average the total ownership percentage across quarters to calculate the average ownership percentage for each firm in each calendar year. After merging the institutional holdings data with Bushee & Goodman (2007) classification data, we repeat the same procedure and calculate the total ownership of firms by institutions that adopt growth, value and balanced investment styles in each calendar year. We also calculate the total number of institutional owners and the number of institutions based on each investment style that hold a firm in a calendar year. We, then merge the ownership data with Compustat and patent data. Figure 1 represents the distribution of institutional investors based on both Bushee (1998) and Bushee & Goodman (2007) classifications. Figure 1-a shows that almost two-thirds of the institutional investors fall into the quasi-indexed category in the entire Bushee (1998) classification that spans from 1981-2011. The transient institutional investors constitute approximately 30% of the sample and only a small fraction of the institutional investors can be considered as dedicated institutional investors. In Figure 1-b, when we use the Bushee & Goodman (2007) classification for the same period, we see a more balanced distribution across investment styles: Growth institutional investors constitute slightly less than one-third of the

sample, value institutional investors constitute around 30% of the sample and the balanced (hybrid) institutional investors constitute slightly more than 38% of the sample. Figure 1-c shows the cross-classifications of institutional investors: Institutions that adopt the growth investment style make up the largest subcategories of both transient (with 43.6%) and dedicated institutions (with 42.4%). Quasi-indexed classification is dominated by balanced (hybrid) institutions by 44.6%. Although not reported in Figure 1, for our sample period of 1991-1999, as also used by Aghion et al. (2013), the proportion of the quasi-indexed institutions within the Bushee (1998) classification is even larger with a 69.8%. Transient and dedicated institutions follow the quasiindexed institutions with 23.9% and 6.3%, respectively. 2.1. Descriptive Statistics The summary statistics of our variables are presented in Table 1. Our total sample has 3,115 firm-year observations from 665 publicly listed firms between 1991 and 1999. Not all variables are available for every firm-year observation and as a result, our regression estimations are based on a smaller number of observations than featured in Table 1. Because our innovation data (i.e., R&D expenditure, number of patents and cite-weighted patents) is highly skewed we winsorize our sample at 5th percentile on each tail. [Table 1 Here] Firms in our sample spend almost $66 million on R&D and are granted slightly more than 16 patents (38 cite-weighted patents) on average.8 However, the median firm spends only $20 million on R&D and it is granted five patents (six cite-weighted patents).9 The average net sales of our sample firms is around $2 billion and their current assets are more than twice their current liabilities on average. 32% of our sample firms are listed in S&P 500 and they are slightly older than 25 years on average while the youngest (oldest) firm is only 3 (67) years old. The average (median) institutional ownership rate is 52% (54%). The breakdown of average ownership based

8

We have eliminated the firms with missing R&D expenditure data from our sample. This is because we believe that arbitrarily replacing the missing R&D expenditure with zero can produce unreliable results.

on investment styles show that growth (value) institutional investors own 10% (14%) of firms’ stocks. Table 2 shows the pairwise correlations between our variables. While the average institutional ownership rate is positively correlated with R&D expenditure, it is very moderately and negatively correlated with R&D intensity. The average growth (value) institutional ownership rate is positively (negatively) correlated with the R&D intensity. This can be seen as an early indication of the direction of their relationship, despite the fact that the direction and/or significance can change drastically in the regression setting. Finally, there is a strong positive correlation between the R&D expenditure and number of (or cite-weighted) patents as expected. [Table 2 Here] 3. Empirical Analyses 3.1. R&D Expense & R&D Intensity We start our analysis by investigating whether there exists any relationship between institutional ownership and R&D spending (R&D intensity) of firms. In order to test this relationship, we run the following regressions: 𝑌𝑖,𝑡 = 𝛽0 + 𝜶𝑿𝒊,𝒕−𝟏 + 𝜸𝒁𝒊,𝒕−𝟏 + 𝜏𝑡 + 𝜑𝑠 + 𝜀𝑖,𝑡

(1)

Where; 𝑌𝑖,𝑡 represents the logarithm of R&D expenditure (or R&D intensity) of firm i at year t, 𝑋𝑖,𝑡−1 represents our variables of interest which are the average institutional ownership rate, average growth institutional ownership rate, average value institutional ownership rate, and average balanced institutional ownership rate of firm i at year t-1, 𝑍𝑖,𝑡−1 represents the control variables, which are ln(𝑆𝑎𝑙𝑒𝑠) , ln(𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑟𝑎𝑡𝑖𝑜) , 𝐿𝑒𝑟𝑛𝑒𝑟 𝑖𝑛𝑑𝑒𝑥, (𝐿𝑒𝑟𝑛𝑒𝑟 𝑖𝑛𝑑𝑒𝑥)2 and the average number of institutional investors for firm i at year t-1, 𝜏𝑡 represents the year fixed effects, and 𝜑𝑠 represents the industry fixed effects (based on two-digit SIC codes). We believe that there is a gain in using R&D intensity as an additional dependent variable since it scales the R&D expense with sales and makes it possible to compare firms with different sizes. In our regressions where R&D intensity is the dependent variable, we drop 𝐿𝑛(𝑆𝑎𝑙𝑒𝑠) from our

set of control variables. All logarithmic transformations are done with “1+variable” to assure that we do not have a negative value for these variables. The relationship between firm size and R&D spending (or R&D intensity) is known since Schumpeter (1950). The current literature argues that the larger firms are more inclined to and capable of conducting R&D. To control for this relationship, we use 𝐿𝑛(𝑆𝑎𝑙𝑒𝑠) as the proxy of firm size. Our second control variable, 𝐿𝑛(𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝑅𝑎𝑡𝑖𝑜) is a measure of liquidity. Kochhar & David (1996) find that better liquidity positively affects innovation. This is intuitive as less liquid firms may not be able to dedicate resources to long-term R&D projects. Aghion et al. (2005) show that there is a non-linear (i.e., inverted-U) relationship between product market competition and innovation. Thus, we control for this effect by including 𝐿𝑒𝑟𝑛𝑒𝑟 𝐼𝑛𝑑𝑒𝑥 and (𝐿𝑒𝑟𝑛𝑒𝑟 𝐼𝑛𝑑𝑒𝑥)2 in our regressions. We calculate the Lerner Index as the ratio of the cumulative EBITDA to the cumulative sales in each industry based on three-digit SIC in each year. A lower Lerner Index indicates higher product market competition. We control for industry and year fixed effects in our regressions. The industry fixed effects are based on two digit SIC codes. In some specifications, we also control for the average number of institutional investors (based on investment style) to account for ownership concentration. In columns 1 and 2 of Table 3, we find a positive and significant relationship between institutional ownership and R&D expenditure. According to column 2, a 10-percentage point increase in institutional ownership corresponds to approximately three percent increase in R&D expenditure. In columns 5 and 6, we do not find a significant relationship between institutional ownership and R&D intensity. On the other hand, we find interesting results in columns 3, 4, 7 and 8, when we break down the institutional ownership into growth, value and balanced investor categories. In columns 3 and 4, growth institutional ownership have substantial effect on R&D expenditure. According to column 3, a 10-percentage point increase in growth institutional ownership is associated with approximately 17.8-percent increase in R&D expenditure while an equivalent increase in value institutional ownership is associated with approximately 5.3-percent reduction. Thus, for our median firm which spends $20 million on R&D projects, a 10-percentage point increase in growth (value) institutional ownership raises (reduces) this expenditure to $23.56 ($18.94)

millions. In columns 7 and 8, we observe similar patterns for the effect of growth and value institutional ownership on R&D intensity. [Table 3 Here] In Table 4, we examine whether the relationships that we find in Table 3 depends on the relative levels of R&D expenditure and R&D intensity. In doing so, we utilize a quantile regressions approach. Table 4 Panel-A (columns 1-3) shows that institutional ownership has a positive and significant effect on R&D expenditure for firms at and below median level of R&D expenditure. For firms with above the median level of R&D expenditure, an increase in institutional ownership does not result in an additional increase in R&D expenditure. In columns 4-6, growth institutional ownership has a positive and significant effect on R&D expenditure in all quantiles while value institutional ownership has a negative and significant effect only for firms with above the median R&D expenditure. These results indicate that while growth institutional investors encourage their investees for becoming more innovative, value institutional investors are not entirely against their investees dedicating more funds for innovation as long as it is at a reasonable level. The coefficients we find for balanced institutional ownership are between those of growth and value institutional ownership. This is consistent with their hybrid (balanced) investment mandate. In Table 4 Panel-B, we run quantile regressions for R&D intensity and find qualitatively similar results to what we report in Panel A. [Table 4 Here] 3.2. Patents and Cite-weighted Patents While R&D is the input, patents (and citations) are the outputs of the innovation process. Therefore, it is important to investigate the effect of institutional ownership on number of granted (and citation-weighted) patents. We use the citation-weighted patents because it reveals the quality of innovation as well as the quantity of innovation. Both number of patents and cite-weighted patents are non-negative integers known as count data. Therefore, we use count data models in addition to OLS regressions for estimating the effect of institutional ownership on innovation. Our estimation with OLS regression follow the same functional form as in equation (1), while other count data model estimations are based on Poisson and negative binomial regressions. All independent variables are lagged one period.

Different from our regressions for R&D expenditure and R&D intensity, we use patent stock and cite-weighted patent stock as additional explanatory variables to account for the existing innovative knowledge and experience that firms possess. These measures are calculated as the cumulative sum of all patents (and citation-weighted patents) that a firm has been granted up to a given year. Including these variables in our regressions as controls allows us to understand whether institutional investors self-select into innovative firms. In Table 5 Panel-A (columns 4-6), we find that growth institutional ownership has positive and significant effect while value institutional ownership has negative and significant effect on the number of granted patents. If we look at the economic significance of these factors, we can say that a 10-percentage point increase in growth (value) institutional ownership rate increases (reduces) the number of patents by 4.96-percent (5.38-percent). Thus, a firm with 40 granted patents per year (on average) could get additional two patents with the 10-percentage point increase in growth institutional ownership rate. Moreover, the signs and magnitudes of the coefficients for growth and value institutional ownership likely offset each other, which may explain the insignificant coefficients that we find for institutional ownership in columns 1-3. Our findings are consistent through all three regression approaches. Trajtenberg (1990) suggests that the raw number of patents do not take into account the quality of patents. The most-widely used method to overcome this drawback is to weigh the patents with the citations that they receive. We present our results for cite-weighted patents in Table 5 PanelB. Similar to our results in Panel-A, we observe that the effect of growth and value institutional ownership are significant yet opposite. The coefficients on patent stock and citation-weighted patent stock are positive and significant in Panel-A and Panel-B, respectively. This indicates that future innovation is a function of past innovative knowledge and success of firms. [Table 5 Here] In Table 6, we run quantile regressions based on the relative level of the number of patents and cite-weighted patents to investigate whether institutional investors with different investment styles have differential effects. Our results (in columns 4-6) indicate that growth institutional investors have a positive impact on patent quality in all quartiles while their effect on the number

of patents is positive and significant only at the 75th percentile. This finding suggests that growth institutional investors place more emphasis on the quality of patents than the quantity of patents. The effect of value institutional investors is negative and significant on the number of granted patents (i.e., 50th and 75th percentiles) and cite-weighted patents (i.e., 75th percentile) indicating that increases in value institutional ownership may impede innovation for highly innovative firms in terms of quality and quantity of patents. Finally, balanced investors do not have a significant effect on either innovation measures. [Table 6 Here] 3.3.Patent Radicalness Cite-weighted patents alone are not informative enough about the importance of patents. Trajtenberg (1990) suggests that a patent, which is heavily cited by other patents, can be described as important (or radical) innovation since they inspire a wide range of future innovations. Trajtenberg et al. (1992) and Ahuja & Lampert (2001) argue that a patent that does not cite any other patents can be assumed to be a radical innovation. In the spirit of Trajtenberg (1990), Trajtenberg et al. (1992) and Ahuja & Lampert (2001), we construct a measure of patent radicalness, “average citations received-to-made” variable, by dividing the total number of citations a firm receives on its all patents in a given year with the total number of citations they make in that specific year. In addition to this measure, we construct an additional patent radicalness measure following Quintana-Garcia & Benavides-Velasco (2008). This measure, zero-citations-made indicator, is a binary variable which takes the value of 1 if a firm is granted at least one patent that does not cite any other patents up to a given year or 0 otherwise. Next, we estimate the effect of institutional investors on both measures of patent radicalness. We use OLS regression in our estimation when the dependent variable is the average citations received-to-made ratio and a logistic regression when the dependent variable is the zerocitations-made indicator. In the logistic regression, along with other variables, we also control for a binary variable, “PastZeroCitations”, which takes the value of 1 if the firm had at least one patent with zero citations-made in previous years.

We present our findings in Table 7. We find that while growth institutional investors have a positive and significant effect on both measures of patent radicalness, value institutional investors have a negative and significant effect. If we look at column 2, we see that a 10percentage point increase in growth (value) institutional ownership rate corresponds to 1.35percent increase (0.94-percent decrease) in citations received to citations made ratio. Balanced investors have no significant effect on patent radicalness, which is consistent with their hybrid (balanced) style. In columns 1 and 2, the results indicate that firms with a larger cite-weighted patent stock also tend to have more radical patents. Similarly, in columns 3 and 4, we find that firms, which previously had patents that made zero citations to others, are more likely to innovate radical patents in the future. [Table 7 Here] 4. Robustness The relationship between institutional ownership and innovation can be endogenous in such a way that institutional investors may choose to invest in more innovative firms rather than firms becoming more innovative because institutions invest in them. In order to address this concern, we use 2SLS regression method with four potential instrumental variables for institutional ownership. Our first instrumental variable, as in Aghion et al. (2013) and Clay (2000), is a dummy variable that indicates S&P500 index membership of firms. It takes a value of one if a firm is a member of the index in at least one calendar quarter in a given year and a value of zero otherwise. Our second instrumental variable, as used by Cornett et al. (2007), is the number of analysts that cover the firm.10 Using the I/B/E/S database, we calculate the analyst coverage as the number of analysts who issued EPS estimates for a firm in a given year. The rationale behind choosing these two instruments is that while they make firms more visible to institutional investors, it is not economically grounded to believe that firms in S&P500 index or firms followed by more analysts need to innovate more. We see that this economic intuition is also supported by the correlation between R&D intensity and S&P500 index membership dummy (i.e., -0.094) and the correlation between R&D intensity and number of analysts (i.e., In addition, O’Brien & Bhushan (1990) suggest that institutional investors’ decision to buy the stocks of a firm and analysts’ decision to cover the firm are interrelated and they must be simultaneously analyzed. The authors also show that there is no causal link between analyst coverage and firm (size) growth, which may stem from innovation. These results also support the notion that analyst coverage can serve as a good instrumental variable in our model. 10

0.07). Moreover, the correlation between institutional ownership and S&P500 membership dummy (number of analysts) is 0.4 (0.0502) which indicates that they are likely good instruments. After further scanning the literature for alternative instrumental variables, we employ two additional IVs in our analysis. One of these IVs is the share turnover rate as used by Bennett, Sias & Starks (2003) and Hartzell & Starks (2003). We calculate the share turnover as the natural logarithm of one plus the ratio of average monthly volume to average monthly number of shares outstanding in a given year. The other IV is the firms’ CRSP age as used by Gompers & Metrick (2001) and Yan & Zhang (2009) in estimating the level of institutional ownership. Following these papers, we calculate the CRSP age as the time elapsed in months between a firm's inclusion in the CRSP database and up to the end of the year that the regressions are estimated. Based on the correlations they have with the number of patents and cite-weighted patents, these two additional variables are more suitable to serve as instruments for those 2SLS regressions where dependent variables are number of patents and cite-weighted patents. We run the following 2SLS regressions after determining the best possible instrumental variables: First Stage:

𝑋𝑖,𝑡 = 𝛽0 + 𝝍𝑽𝒊,𝒕−𝟏 + 𝜸𝒁𝒊,𝒕−𝟏 + 𝜏𝑡 + 𝜑𝑠 + 𝜀𝑖,𝑡

(2)

Second Stage:

̂ 𝒊,𝒕−𝟏 + 𝜸𝒁𝒊,𝒕−𝟏 + 𝜏𝑡 + 𝜑𝑠 + 𝜀𝑖,𝑡 𝑌𝑖,𝑡 = 𝛽0 + 𝜶𝑿

(3)

Where; 𝑌𝑖,𝑡 represents the ln(1+innovation measure) of firm i at year t, 𝑉𝑖,𝑡−1 represents our instrumental variables for firm i at year t-1, 𝑋𝑖,𝑡−1 represents our variable of interest which is the average institutional ownership rate of firm i at year t-1, 𝑍𝑖,𝑡−1 represents the control variables for firm i at year t-1, 𝜏𝑡 represents the year fixed effects, and 𝜑𝑠 represents the industry fixed effects (based on two-digit SIC codes). We present our second stage regression results in Table 8. 11 In columns 1, 3 and 5, we use S&P500 membership dummy and the number of analysts as our instrumental variables in the

11

Despite not being reported, all instrumental variables are individually and jointly significant in our first stage regressions. These results are available upon request.

first stage. In columns 2, 4 and 6, all four instrumental variables are included in the first stage regressions. In our regressions with R&D intensity (columns 1 and 2), we find that the insignificant effect of institutional ownership reported in Table 3 becomes positive and significant. A similar change in sign and significance also occurs in our regressions where the dependent variables are number of patents (columns 3 and 4) and cite-weighted patents (columns 5 and 6). Thus, our 2SLS results confirm that institutional investors have a positive effect on firm innovation, consistent with the findings of the existing literature. Although we wish to implement our 2SLS regressions with respect to each investment style, we could not identify proper instrumental variables. Therefore, our attempt to understand endogeneity is limited to overall institutional ownership. [Table 8 Here] 5. Conclusion We examine the relationship between institutional ownership with respect to the investing styles of institutional investors and firm innovation. In our empirical analysis, we also investigate whether the effect of institutional investors differ based on firms’ relative level of innovation. To the best of our knowledge, this is the first paper that examines this relationship from an investing style perspective. In regards to investment styles, we use Bushee & Goodman (2007) classification, which segregates institutional investors into three categories: Value institutional investors that prefer a steady dividend income, growth institutional investors that prefer earnings retention and reinvestment, and balanced (hybrid) institutional investors that prefer the midst way of value and growth institutional investors. This classification offers an additional benefit compared to Bushee (1998) classification: Bushee & Goodman (2007) classification is less concentrated than and more heterogeneous within the Bushee (1998) classification, which characterizes institutions based on their investment horizon as transient, dedicated, and quasi-indexed. We find growth and value institutional investors have opposing effects on firm R&D and innovation. While growth institutional investors have a positive effect on pretty much all innovation measures that we used, value institutional investors have a negative effect. Balanced investors generally have no significant effect on firm innovativeness. With these findings, we

offer a potential explanation to why Aghion, Van Reenen & Zingales (2013) find a positive effect for transient and dedicated institutional owners but not for quasi-indexed ones: Growth institutional investors are the largest subcategory within both dedicated and transient investors while balanced (hybrid) investors are the largest subcategory in quasi-indexed investors. Lastly, we show that the effect of institutional investors change based on the innovativeness level of the firms they invest in. More specifically, value institutional investors have no significant effect when firms have below median level of innovation and have negative (and significant) effect when the innovativeness is above the median level. Our interpretation of this finding is that value institutional investors may be revealing their preference for dividend distributions when firms make relatively large investments in R&D and innovation that requires retention. One potential problem, which may affect our findings, is the endogeneity bias stemming from reverse causality. Put differently, institutional investors may choose to invest into firms based on their innovative success rather than influencing them to innovate more. To alleviate this concern, we use multiple instrumental variables in a 2SLS setting and we show that our results withstand. Our sample entails data from 1991 to 1999 for comparability purposes to the existing studies that we build upon. Therefore, one should be careful about generalizing our findings to other periods or countries, as there may be significant differences with respect to patenting rights, regulations, investment environment and corporate governance. We encourage future work to investigate the mechanisms and channels through which investment styles impact innovation. In addition, examining the impact of investors’ preferences towards large and small cap stocks along with their investment styles (i.e., large-cap growth, small-cap value, etc.) on R&D and innovation may yield interesting findings.

References Abdioglu, N., Khurshed, A., Stathopoulos, K. (2013). Foreign Institutional Investment: Is Governance Quality at Home Important?. Journal of International Money and Finance, 32, 916-940. Aghion, P., Bloom, N., Blundell, R., Griffith, R., Howitt, P. (2005). Competition and Innovation: An Inverted-U Relationship. The Quarterly Journal of Economics, 120(2), 701-728. Aghion, P., Van Reenen, J. M., Zingales, L. (2013). Innovation and Institutional Ownership. American Economic Review, 103(1), 277-304. Ahuja, G., Lampert, C. M. (2001). Entrepreneurship in the Large Corporation: A Longitudinal Study of How Established Firms Create Breakthrough Inventions. Strategic Management Journal, 22, 521543. An, H., Zhang, T. (2013). Stock Price Synchronicity, Crash Risk, and Institutional Investors. Journal of Corporate Finance, 21, 1-15. Barberis, N., Shleifer, A. (2003). Style Investing. Journal of Financial Economics, 68, 161-199. Barker, V. L., Mueller, G. C. (2002). CEO Characteristics and Firm R&D Spending. Management Science, 48(6), 782-801. Bennett, J. A., Sias, R. W., Starks, L. T. (2003). Greener Pastures and the Impact of Dynamic Institutional Preferences. The Review of Financial Studies, 16(4), 1203-1238. Bernstein, R., (1995). Style Investing. Wiley: New York. Bhattacharya, M., Bloch, H. (2004). Determinants of Innovation. Small Business Economics, 22, 155. Brossard, O., Lavigne, S., Sakinc, M. E. (2013). Ownership Structures and R&D in Europe: The Good Institutional Investors, The Bad and Ugly Impatient Shareholders. Industrial and Corporate Change, 22(4), 1031-1068. Brown, J. R., Martinsson, G., Petersen, B. C. (2012). Do financing constraints matter for R&D?. European Economic Review, 56(8), 1512-1529. Brown, J. R., Petersen, B. C. (2011). Cash holdings and R&D smoothing. The Journal of Corporate Finance, 17(3), 694-709.

Bushee, B. J. (1998). The Influence of Institutional Investors on Myopic R&D Investment Behavior. Accounting Review, 73(3), 305-333. Bushee, B. J., Goodman, T. H. (2007). Which Institutional Investors Trade Based on Private Information About Earnings and Returns?. Journal of Accounting Research, 45(2), 289-321. Capaul, C., Rowley, I., Sharpe, W. F. (1993). International value and growth stock returns. Financial Analysts Journal, 27-36. Cassell, C. A., Huang, S. X., Sanchez, J. M., Stuart, M. D. (2012). Seeking Safety: The Relation between CEO Inside Debt Holdings and The Riskiness of Firm Investment and Financial Policies, Journal of Financial Economics, 103(3), 588-610. Cheng, S. (2004). R&D Expenditures and CEO Compensation. The Accounting Review, 79(2), 305328. Clay, D. G. (2000). The Effects of Institutional Investment on CEO Compensation. University of Southern California Working Paper. Cornett, M. M., Marcus, A. J., Saunders, A., Tehranian, H. (2007). The Impact of Institutional Ownership on Corporate Operating Performance. Journal of Banking and Finance, 31, 1771-1794. Czarnitzki, D., Toole, A. A. (2011). Patent Protection, Market Uncertainty, and R&D Investment. The Review of Economics and Statistics, 93(1), 147-159. Davies, R., Haldane, A. G., Nielsen, M., Pezzini, S. (2014). Measuring the costs of short-termism. Journal of Financial Stability, 12, 16-25. Eng, L. L., Schakell, M. (2001). The Implications of Long-Term Performance Plans and Institutional Ownership for Firms’ Research and Development (R&D) Investments. Journal of Accounting, Auditing & Finance, 16(2), 117-139. Francis, B. B., Hasan, I., Sharma, Z. (2011). Incentives and Innovation: Evidence from CEO Compensation Contracts, Bank of Finland Research Discussion Paper No. 17/2011. Francis, J., Smith, A. (1995). Agency costs and innovation: Some Empiricial Evidence. Journal of Accounting and Economics, 19, 383-409.

Froot, K., Teo, M. (2008). Style investing and institutional investors. Journal of Financial and Quantitative Analysis, 43(04), 883-906. Galasso, A., Simcoe, T. S. (2011). CEO Overconfidence and Innovation. Management Science, 57(8), 1469-1484. Gompers, P. A., Metrick, A. (2001). Institutional Investors and Equity Prices. The Quarterly Journal of Economics, 116(1), 229-259. Gonard, E., Kim, E. J., Ynesta, I. (2008) Recent Trends in Institutional Investors Statistics. Financial Market Trends. OECD Report. Graves, S. B. (1988). Institutional Ownership and Corporate R&D in the Computer Industry. The Academy of Management Journal, 31(2), 417-428. Hall, B. H., Jaffe, A. B., Trajtenberg, M. (2001). The NBER patent citation data file: Lessons, insights and methodological tools (No. w8498). National Bureau of Economic Research. Hansen, G. S., Hill, L. C. (1991). Are Institutional Investors Myopic? A Time-Series Study of FourTechnology-Driven Industries. Strategic Management Journal, 12(1), 1-16. Hartzell, J. C., Starks, L. T. (2003). Institutional Investors and Executive Compensation. The Journal of Finance, 58(6), 2351-2374. Heineman, B. W., Davis, S. (2011). Are Institutional Investors Part of the Problem or Part of the Solution? Key Descriptive and Prescriptive Questions About Shareholders’ Role in U.S. Public Equity Markets. Committee for Economic Development and Yale School of Management, Millstein Center for Corporate Governance and Performance. Herz, H., Schunk, D., Zhender, C. (2014). How Do Judgmental Overconfidence and Overoptimism Shape Innovative Activity?. Games and Economic Behavior, 83, 1-23. Hirshleifer, D., Low, A., Teoh, S. H. (2012). Are Overconfident CEOs Better Innovators?. The Journal of Finance, 67(4), 1457-1498. Holden, C. W., Lundstrum, L. L. (2009). Costly Trade, Managerial Myopia, and Long-term Investment. Journal of Empirical Finance, 16(1), 126-135. Hottenrott, H., Hall, B. H., Czarnitzki, D. (2016). Patents as Quality Signals? The Implications for Financing Constraints on R&D. Economics of Innovation and New Technology, 25(3), 197-217.

Huang, W., Paul, D. L. (2017). Institutional Holdings, Investment Opportunities and Dividend Policy. The Quarterly Review of Economics and Finance, 64, 152-161.

Investment & Pensions Europe (2016). Top 400 Asset Managers 2016: Global Assets Now €56.3trn. retrieved from https://www.ipe.com/reports/special-reports/top-400-asset-managers/top-400-assetmanagers-2016-global-assets-now-563trn/10013542.article on March 25, 2017. Jin, D., Wang, H., Wang, P., Yin, D. (2016). Social Trust and Foreign Ownership: Evidence from Qualified Foreign Institutional Investors in China. Journal of Financial Stability, 23, 1-14. Kochhar, R., David, P. (1996). Institutional Investors and Firm Innovation: A Test of Competing Hypotheses. Strategic Management Journal, 17(1), 73-84. Lerner, J., Wulf, J. (2007). Innovation and Incentives: Evidence from Corporate R&D. The Review of Economics and Statistics, 89(4), 634-644. Matsumura, T., Matsushima, N., Cato, S. (2013). Competitiveness and R&D Competition Revisited. Economic Modelling, 31, 541-547. McCahery, J. A., Sautner, Z., Starks, L. T. (2016). Behind the Scenes: The Corporate Governance Preferences of Institutional Investors. The Journal of Finance, 71(6), 2905-2932. Mitroff, I. I. (1987). Business NOT as Usual: Rethinking Our Individual, Corporate, and Industrial Strategies for Global Competition. Jossey Bass: San Francisco. Mullainathan, S. (2000). Thinking Through Categories. Unpublished Working Paper: MIT. O’Brien, P. C., Bhushan, R. (1990). Analyst Following and Institutional Ownership. Journal of Accounting Research, 28, 55-76. Peng, E. Y., Yan, A., Yan, M. (2016). Accounting Accruals, Heterogeneous Investor Beliefs, and Stock Returns. Journal of Financial Stability, 24, 88-103. Quintana-Garcia, C., Benavides-Velasco, C. A. (2008). Innovative Competence, Exploration and Exploitation: The Influence of Technological Diversification. Research Policy, 37, 492-507. Rong, Z., Wu, X., Boeing, P. (2017). The Effect of Institutional Ownership on Firm Innovation: Evidence from Chinese Listed Firms. ZEW Discussion Papers, No. 17-005, retrieved from https://www.econstor.eu/bitstream/10419/149749/1/878841741.pdf on June 19, 2017.

Schumpeter, J. A. (1950). Capitalism, Socialism, and Democracy. 3rd edition. Harper & Row: New York. Sharpe, W. F. (1992). Asset Allocation: Management Style and Performance Measurement. Journal of Portfolio Management, 18(2), 7-19. Shefer, D., Frenkel, A., (2005). R&D, Firm Size and Innovation: An Empirical Analysis. Technovation, 25(1), 25-32. Swensen, D. F. (2000). Pioneering Portfolio Management. The Free Press: New York. Switzer, L. N., Wang, J. (2017). Institutional Investment Horizon, the Information Environment, and Firm Credit Risk. Journal of Financial Stability, 29, 57-71. Teo, M., Woo, S-J. (2004). Style effects in the cross-section of stock returns. Journal of Financial Economics, 74, 367-398. Trajtenberg, M. (1990). A Penny for Your Quotes: patent Citations and the Value of Innovations. The RAND Journal of Economics, 21(1), 172-187. Trajtenberg, M., Henderson, R., Jaffe, A. (1992). Ivory Tower versus Corporate Lab: AN Empirical Study of Basic Research and Appropriability. NBER Working paper No. 4146. Wahal, S., McConnell, J. J. (2000). Do institutional investors exacerbate managerial myopia? Journal of Corporate Finance, 6, 307-329. Wahal, S., Yavuz, M. D. (2013). Style Investing, Comovement and Return Predictability. Journal of Financial Economics, 107, 136-154. Yan, X., Zhang, Z. (2009). Institutional Investors and Equity Returns: Are Short-term Institutions Better Informed?. The Review of Financial Studies, 22(2), 893-924.

Figures Figure 1 – Institutional Investor Categorization Figure 1-a

Figure 1-c

Figure 1-b

Tables

Table 1: Summary Statistics Sample Period: 1991-1999 All variables are winsorized at 5th and 95th percentile levels Variable R&D Expenditure (mln USD) Avg. Ownership % Avg. Own. % (Growth) Avg. Own. % (Value) Avg. Own. % (Balanced) Avg. No. of Owners Avg. No. of Owners (Growth) Avg. No. of Owners (Value) Avg. No. of Owners (Balanced) Sales (mln USD) Current Ratio Lerner Index No. of Analysts S&P 500 Membership Stock Turnover CRSP Age (Months) No. of Patents Cite Weighted Patents Avg. No. of Citations Received per Patent Avg. No. of Citations Made No. of Patents that Made Zero Citations

N 2,660 3,115 3,115 3,115 3,115 3,115 3,115 3,115 3,115 3,114 3,019 3,115 2,757 3,115 3,099 3,099 3,115 3,115 3,115 3,115 3,115

Mean 65.91 0.52 0.10 0.14 0.28 111.86 20.05 26.50 64.15 1,912.31 2.37 0.14 10.82 0.32 0.92 306.72 16.27 37.87 2.27 13.41 0.16

26

Median 20.22 0.54 0.07 0.12 0.28 74.25 12.25 16.75 43.00 653.03 2.04 0.13 9.00 0.00 0.63 265.00 5.00 6.00 1.20 11.33 0.00

S.D. 105.63 0.17 0.08 0.08 0.12 101.36 21.22 25.26 56.59 2,812.81 1.17 0.04 8.23 0.47 0.82 218.03 26.21 74.90 2.61 7.69 0.49

Min. 1.22 0.19 0.00 0.03 0.06 11.25 0.25 2.50 6.75 38.23 1.02 0.08 1.00 0.00 0.15 37.00 1.00 0.00 0.00 4.00 0.00

Max. 418.87 0.79 0.31 0.31 0.48 367.00 77.67 92.50 201.25 10,647.59 5.55 0.24 30.00 1.00 3.26 805.00 100.00 291.00 9.00 33.27 2.00

Table 2: Pairwise Correlations Variables

(1)

(2)

(3)

(4)

(5)

(1) R&D Expenditure

1.00

(2) R&D Intensity

0.22

1.00

(3) Avg. Ownership %

0.32

-0.06

1.00

(4) Avg. Own. % (GRO)

0.08

0.26

0.44

1.00

(5) Avg. Own. % (VAL)

0.08

-0.23

0.40

-0.32

1.00

(6) Avg. Own. % (BAL)

0.36

-0.14

0.80

0.11

0.15

1.00

(7) No. of Analysts

0.64

0.07

0.42

0.19

-0.02

0.47

1.00

(8) S&P 500 Membership

0.59

-0.11

0.40

-0.01

0.16

0.50

0.65

1.00

(9) Stock Turnover

0.16

0.51

0.17

0.50

-0.17

-0.00

0.22

-0.03

1.00

(10) CRSP Age (months)

0.38

-0.32

0.23

-0.17

0.23

0.32

0.37

0.54

-0.25

1.00

(11) No. of Patents

0.67

0.11

0.30

0.11

0.02

0.33

0.51

0.46

0.15

0.28

1.00

(12) Cite Weight. Patents

0.43

0.15

0.19

0.14

-0.03

0.18

0.40

0.31

0.18

0.11

0.75

27

(6)

(7)

(8)

(9)

(10)

(11)

(12)

1.00

Table 3: R&D Expenditure (R&D Intensity) and Institutional Ownership All independent variables are in one-lagged form. Lerner index is the ratio of the cumulative EBITDA to the cumulative sales in each industry based on 3-digit SIC in each year.

Ln(1+R&D Expenditure) Avg. Ownership %

(1)

(2)

0.494*** (0.125)

(3)

Ln(1+R&D Intensity) (4)

(5)

(6)

0.289**

0.003

-0.008

(0.119)

(0.006)

(0.006)

(7)

(8)

Avg. Own. % (GRO)

1.641***

1.361***

0.051***

0.035***

(0.223)

(0.251)

(0.011)

(0.013)

Avg. Own. % (VAL)

-0.541**

-0.235

-0.048***

-0.046***

(0.232)

(0.251)

(0.011)

(0.013)

Avg. Own. % (BAL)

0.331*

0.129

-0.005

-0.015*

(0.184)

(0.182)

(0.008)

(0.009)

Ln(Sales)

Ln(Cur. Ratio)

Lerner Ind.

(Lerner Ind.)2

0.847***

0.709***

0.845***

0.693***

(0.016)

(0.025)

(0.016)

(0.025)

0.166***

0.114***

0.134***

0.103***

0.011***

0.010***

0.010***

0.010***

(0.018)

(0.018)

(0.018)

(0.018)

(0.001)

(0.001)

(0.001)

(0.001)

-8.803***

-9.903***

-10.224***

-10.362***

-0.825***

-0.838***

-0.883***

-0.856***

(3.023)

(2.936)

(2.976)

(2.923)

(0.151)

(0.151)

(0.148)

(0.149)

49.022***

47.156***

50.738***

48.201***

3.848***

3.748***

3.912***

3.792***

(9.269)

(9.029)

(9.098)

(8.972)

(0.480)

(0.481)

(0.469)

(0.475)

Avg. No. Owners (GRO)

0.012***

0.004

0.000

(0.003)

0.001*** (0.000)

(0.002)

Avg. No. Owners (VAL)

-0.003

0.001

0.000

0.000

(0.002)

(0.003)

(0.000)

(0.000)

Avg. No. Owners (BAL)

0.002

0.004**

0.000

0.000

(0.001)

(0.001)

(0.000)

(0.000)

(0.000)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Year FE N

Yes 2060

Yes 2060

Yes 2060

Yes 2060

Yes 2059

Yes 2059

Yes 2059

Yes 2059

Adj. R2

0.742

0.758

0.751

0.761

0.421

0.434

0.438

0.440

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

28

Table 4 (Panel A): Quantile Regressions of R&D Expenditure and Institutional Ownership All independent variables are in one-lagged form. Lerner index is the ratio of the cumulative EBITDA to the cumulative sales in each industry based on 3-digit SIC in each year.

Ln(1+R&D Expenditure)

Percentile Avg. Ownership %

(1) 25th 0.692***

(2) 50th 0.528***

(3) 75th 0.184

(0.165)

(0.155)

(0.182)

Avg. Own. % (GRO)

Avg. Own. % (VAL)

Avg. Own. % (BAL)

(4) 25th

(5) 50th

(6) 75th

1.728***

1.363***

1.702***

(0.284)

(0.299)

(0.293)

-0.208

-0.471

-0.545*

(0.338)

(0.316)

(0.307)

0.512**

0.440**

-0.194

(0.245)

(0.224)

(0.238)

0.836***

0.862***

0.879***

0.840***

0.861***

0.878***

(0.025)

(0.022)

(0.020)

(0.024)

(0.021)

(0.019)

0.102***

0.177***

0.203***

0.075***

0.145***

0.161***

(0.026)

(0.024)

(0.022)

(0.025)

(0.026)

(0.021)

-7.335

-8.359**

-10.103**

-9.045**

-9.432**

-15.912***

(4.467)

(3.768)

(4.490)

(4.084)

(4.015)

(4.888)

(Lerner Ind.)2

44.157*** (13.687)

44.850*** (12.101)

53.580*** (13.065)

47.736*** (12.163)

45.759*** (12.821)

69.118*** (14.243)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N R2

2060 0.729

2060 0.740

2060 0.737

2060 0.741

2060 0.750

2060 0.746

Ln(Sales)

Ln(Cur. Ratio) Lerner Ind.

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

29

Table 4 (Panel B): Quantile Regressions of R&D Intensity and Institutional Ownership All independent variables are in one-lagged form. Lerner index is the ratio of the cumulative EBITDA to the cumulative sales in each industry based on 3-digit SIC in each year.

Percentile

Ln(1+R&D Intensity) (1) (2) th 25 50th

(3) 75th

Avg. Ownership %

0.014***

0.008

-0.012

(0.004)

(0.005)

(0.010)

Avg. Own. % (GRO)

Avg. Own. % (VAL)

Avg. Own. % (BAL)

(4) 25th

(5) 50th

(6) 75th

0.036***

0.048***

0.062***

(0.009)

(0.013)

(0.019)

-0.007

-0.015

-0.056***

(0.008)

(0.010)

(0.015)

0.015***

0.01

-0.032***

(0.005)

(0.007)

(0.011)

0.005***

0.011***

0.014***

0.005***

0.011***

0.011***

(0.001)

(0.001)

(0.001)

(0.001)

(0.001)

(0.001)

-0.509***

-0.964***

-1.216***

-0.499***

-0.968***

-1.222***

(0.112)

(0.144)

(0.244)

(0.110)

(0.140)

(0.238)

2.260***

4.212***

5.570***

2.183***

4.129***

5.441***

(0.394)

(0.499)

(0.720)

(0.393)

(0.486)

(0.719)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N

2059

2059

2059

2059

2059

2059

R2

0.376

0.418

0.421

0.392

0.432

0.440

Ln(Cur. Ratio)

Lerner Ind.

(Lerner Ind.)2

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

30

Table 5 (Panel A): Number of Granted Patents and Institutional Ownership All independent variables are in one-lagged form. Patent Stock is the cumulative sum of all patents up to a given year. (1) OLS Ln(1+No. Patents) 0.09 Avg. Ownership %

(0.120)

of

(2) Poisson

(3) Neg. Binomial

(4) OLS

(5) Poisson

(6) Neg. Binomial

No. of Patents

No. of Patents

Ln(1+No. of Patents)

No. of Patents

No. of Patents

-0.178

0.079

(0.261)

(0.166) 0.484**

0.689*

1.098***

(0.213)

(0.401)

(0.319)

-0.553**

-1.066***

-1.089***

(0.222)

(0.403)

(0.279)

0.021

-0.512

-0.233

(0.182)

(0.344)

(0.235)

Avg. Own. % (GRO)

Avg. Own. % (VAL)

Avg. Own. % (BAL) 0.487***

0.556***

0.471***

0.485***

0.551***

0.473***

(0.020)

(0.041)

(0.034)

(0.020)

(0.042)

(0.031)

0.295***

0.516***

0.492***

0.298***

0.527***

0.500***

(0.023)

(0.054)

(0.037)

(0.024)

(0.053)

(0.035)

0.071***

0.161***

0.125***

0.059***

0.116***

0.089***

(0.017)

(0.032)

(0.023)

(0.017)

(0.030)

(0.022)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N

2371

2371

2371

2371

2371

2371

Patent Stock

Ln(Sales)

Ln(Cur. Ratio)

Adj. R

2

0.651

Pseudo R2

0.653 0.788

0.792

McFadden's Adj R2

0.179

0.182

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

31

Table 5 (Panel B): Cite Weighted Number of Granted Patents and Institutional Ownership All independent variables are in one-lagged form. Cite Weight. Patent Stock is the cumulative sum of all cite-weighted patents up to a given year.

Avg. Ownership %

(1) OLS Ln(1+C.W. Patents) 0.350**

(2) Poisson

(3) Neg. Binomial

(4) OLS

(5) Poisson

(6) Neg. Binomial

C.W. Patents

C.W. Patents

Ln(1+C.W. Patents)

C.W. Patents

C.W. Patents

0.205

0.585

(0.174)

(0.496)

(0.521) 0.849***

1.097*

3.534***

(0.304)

(0.648)

(0.600)

-0.294

-1.909**

-2.949***

(0.330)

(0.881)

(0.968)

0.162

0.461

0.534

(0.254)

(0.515)

(0.677)

Avg. Own. % (GRO) Avg. Own. % (VAL) Avg. Own. % (BAL) 0.386***

0.531***

0.034

0.382***

0.500***

0.011

(0.019)

(0.052)

(0.025)

(0.019)

(0.053)

(0.026)

0.282***

0.562***

0.538***

0.291***

0.581***

0.557***

(0.029)

(0.075)

(0.055)

(0.030)

(0.074)

(0.058)

0.077***

0.175***

0.141**

0.063***

0.116**

-0.021

(0.024)

(0.056)

(0.059)

(0.024)

(0.050)

(0.055)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N

2371

2371

2371

2371

2371

2371

C.W. Patent Stock

Ln(Sales)

Ln(Cur. Ratio)

0.656

Adj. R2 Pseudo R

McFadden's Adj R

0.657 0.776

2 2

Robust standard errors in parentheses

0.783 0.019

0.023

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

32

Table 6: Quantile Regressions for Patents and Cite-Weighted Patents All independent variables are in one-lagged form. Patent Stock (Cite Weight. Patent Stock) is the cumulative sum of all (cite-weighted) patents up to a given year. Ln(1+No. of Patents) (1) Percentile Avg. Own. % (GRO)

Avg. Own. % (VAL)

Avg. Own. % (BAL)

Patent Stock

Ln(1+Cite W. Patents)

(2)

(3)

(4)

(5)

(6)

25 0.478

50th 0.381

75th 0.947***

25th 0.783**

50th 0.649*

75th 1.047**

(0.309)

(0.254)

(0.279)

(0.392)

(0.388)

(0.423)

-0.298

-0.759***

-0.803***

0.332

-0.206

-0.754*

(0.287)

(0.271)

(0.294)

(0.404)

(0.409)

(0.429)

0.174

-0.106

-0.019

0.483

-0.157

-0.483

(0.230)

(0.224)

(0.256)

(0.331)

(0.322)

(0.344)

0.554***

0.573***

0.503***

(0.025)

(0.020)

(0.027) 0.391***

0.436***

0.410***

(0.029)

(0.023)

(0.022)

th

C.W. Patent Stock

0.162***

0.247***

0.323***

0.141***

0.246***

0.351***

(0.028)

(0.026)

(0.033)

(0.040)

(0.035)

(0.038)

0.027

0.057***

0.081***

0.014

0.054*

0.115***

(0.023)

(0.021)

(0.023)

(0.032)

(0.031)

(0.036)

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N

2371

2371

2371

2371

2371

2371

0.634

0.651

0.649

0.646

0.660

0.653

Ln(Sales)

Ln(Cur. Ratio)

2

R

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

33

Table 7: Explorative Patents and Institutional Ownership All independent variables are in one-lagged form. Zero Backward-Citation Made Indicator (“Past Zero Cit” in the table) takes a value of 1 if an investee firm is granted at least 1 patent that does not cite any other patents up to the given year and a value of 0 otherwise. C. W. Patent Stock is the cumulative sum of all cite-weighted patents up to a given year.

Avg. Ownership %

Ln(1+Avg.(Cit. Rec./Cit. Made))

Zero Citation Made Indicator

OLS

Logistic Regression

(1) 0.007

(2)

(0.030)

Avg. Own. % (GRO)

Avg. Own. % (VAL)

Avg. Own. % (BAL)

C. W. Patent Stock

(3) 0.912

(4)

(0.561) 0.134**

1.854*

(0.058)

(0.970)

-0.094*

-1.863*

(0.049)

(1.098)

-0.049

1.043

(0.037)

(0.769)

0.011***

0.010***

(0.002)

(0.002)

Past Zero Cit.

2.324***

2.308***

(0.165)

(0.165)

-0.004

-0.002

0.518***

0.521***

(0.004)

(0.004)

(0.082)

(0.083)

0.015***

0.012***

0.225**

0.165*

(0.005)

(0.004)

(0.094)

(0.096)

Indus. FE

Yes

Yes

No

No

Year FE

Yes

Yes

Yes

Yes

N

2368

2368

2371

2371

0.414

0.418

Pseudo R

0.251

0.257

Robust standard errors in parentheses

* p<.10, ** p<.05, *** p<.01

Ln(Sales)

Ln(Cur. Ratio)

Adj. R2 2

Please see the Appendix for the explanation of the variables

34

Table 8: 2SLS Regressions – Second Stage Results All independent variables are in one-lagged form. Instrumental variables are S&P500 index membership dummy, number of analysts covering the firms, share turnover and CRSP age of firms in months. S&P500 index membership dummy takes a value of 1 if an investee firm is a member of the index in at least one calendar quarter in a given year. Number of analysts is the number of analysts that issued EPS estimates for a firm in a given year and is calculated from I/B/E/S. CRSP age is the time elapsed in months between an investee firm's inclusion in the CRSP Database and any given year-end. Share turnover is the natural logarithm of one plus the ratio of average monthly volume to average monthly number of shares outstanding in a given year. Only the S&P500 Index membership dummy and the number of analysts covering the firms are used as IVs in columns 1, 3, and 5. In columns 2, 4, and 6 all four IVs are included in the first stage. Cite Weight. Patent Stock is the cumulative sum of all cite-weighted patents up to a given year. Lerner index is the ratio of the cumulative EBITDA to the cumulative sales in each industry based on 3-digit SIC in each year.

Avg. Ownership %

Ln(1+R&D Intensity)

Ln(1+No. of Patents)

Ln(1+C.W. Patents)

(1) 0.089***

(2) 0.091***

(3) 4.341***

(4) 4.507***

(5) 6.988***

(6) 6.989***

(0.012)

(0.012)

(0.733)

(0.705)

(1.112)

(1.043)

0.062

0.053

-0.097

-0.096

(0.052)

(0.050)

(0.076)

(0.071)

Sales

Lerner Index

(Lerner Index)2

Current Ratio

-0.899***

-0.878***

(0.170)

(0.173)

4.003***

3.926***

(0.539)

(0.547)

0.013***

0.013***

-0.042

-0.041

-0.102**

-0.093**

(0.001)

(0.001)

(0.031)

(0.030)

(0.047)

(0.045)

0.450***

0.445***

(0.026)

(0.026) 0.344***

0.342***

(0.026)

(0.026)

Patent Stock

C.W. Patent Stock

Indus. FE

Yes

Yes

Yes

Yes

Yes

Yes

Year FE

Yes

Yes

Yes

Yes

Yes

Yes

N

1849

1841

2119

2111

2119

2111

Adj. R2

0.360

0.345

0.484

0.472

0.466

0.468

1st Stage Prob>F Stat

p<.01

p<.01

p<.01

p<.01

p<.01

p<.01

Robust standard errors in parentheses.

* p<.10, ** p<.05, *** p<.01

Please see the Appendix for the explanation of the variables

35

Appendix Variable Ln(1+R&D Expenditure) Ln(1+R&D

Variable Description

Explanation

R&D Expenditure

Natural logarithm of (1+) R&D expenditure by the sample firms. Natural logarithm of (1+) R&D intensity by the sample firms. R&D intensity

R&D Intensity

Intensity) No. of Patents

is the ratio of R&D expenditure to Sales Revenue of the sample firms. Number of patents granted to the sample firms during the sample period. This

Number of Patents

variable recognizes only the quantity of patents granted to the sample firms. Number of patents weighted with the number of citations they received. This

C.W. Patents

Cite-Weighted Patents

variable recognizes both the quantity and the quality of patents of the sample firms. Natural logarithm of (1+) citations received to citations made ratio of the

Ln(1+Avg.(Cit.

Citations

Received

Rec./Cit. Made))

Citations Made Ratio

to

sample firms. The ratio is calculated by dividing the sum of all citations received by a sample firm’s patents to the sum of all citations made by the same sample firm’s patents. The higher the ratio, the higher the impact of the patents.

Zero

Citation

Made Indicator

Own.

%

(GRO)

Avg.

Own.

%

(VAL)

Avg.

Citation

Made

Indicator

Avg. Ownership %

Avg.

Zero

Average

(BAL)

Ln(Sales)

%

a patent that makes no citations. Approval of such a patent shows the originality of the idea/innovation.

Institutional

Ownership Rate

The percentage of a sample firm’s shares held by institutional investors. Because this data is available quarterly, we use the mean of percentages in four quarters.

Average

Growth-

The percentage of a sample firm’s shares held by institutional investors with

Mandate

Institutional

growth-mandate. The variable represents the mean value of four quarters as in

Ownership Rate

average institutional ownership rate.

Average Value-Mandate

The percentage of a sample firm’s shares held by institutional investors with

Institutional

value-mandate. The variable represents the mean value of four quarters as in

Ownership

Rate Own.

A dummy variable, which takes the value of one if the sample firm is granted

average institutional ownership rate.

Average

Balanced-

The percentage of a sample firm’s shares held by institutional investors with

Mandate

Institutional

balanced (hybrid)-mandate. The variable represents the mean value of four

Ownership Rate Total Sales

quarters as in average institutional ownership rate. Natural logarithm of sales revenue by the sample firms. Sales revenue is the proxy for the size of the firm. Natural logarithm of (1+) current ratio. Current ratio is calculated with current

Ln(Cur. Ratio)

Current Ratio

assets divided by current liabilities of the sample firms. This ratio represents the liquidity (i.e., short-term payment ability) of the sample firms.

36

(Industry) Lerner index is the ratio of profit margin to its sales revenue. This ratio is calculated with cumulative EBITDA divided by cumulative sales Lerner Ind.

Lerner Index

revenue in each industry (based on 3-digit SIC). Lerner index measures the effect of product market competition (i.e., the market power of firms/industry). The squared term of the Lerner index defined above. This variable controls for

(Lerner

Ind.)2

Lerner Index Squared

the possible nonlinear effect of product market competition (i.e., the market power of firms/industry).

Avg. No. Owners (GRO)

Avg. No. Owners (VAL)

Avg. No. Owners (BAL)

Patent Stock

C.W. Patent Stock

Average

Number

Institutional

of

The number of institutional investors with growth mandate invested in the

Investors

sample firm in a given year. The annual value for this variable is calculated as

with Growth-Mandate Average

Number

Institutional

the mean value of four quarters. of

The number of institutional investors with value mandate invested in the

Investors

sample firm in a given year. The annual value for this variable is calculated as

with Value-Mandate Average

Number

Institutional

the mean value of four quarters. of

The number of institutional investors with balanced mandate invested in the

Investors

sample firm in a given year. The annual value for this variable is calculated as

with Balanced-Mandate Firm’s Patent Stock Firm’s

Cite-Weighted

Patent Stock

the mean value of four quarters. Cumulative sum of all patents the sample firm is granted up to the given year. This variable is the proxy for the innovative knowledge and ability of the firm. Cumulative sum of all cite-weighted patents the sample firm (up to the given year). This variable takes the usefulness (i.e., quality) of the patents into account by using the number of citations they received. This is a dummy variable that takes the value of one if the sample firm was

Past Zero Cit.

Zero Backward-Citation

granted a patent, which did not cite any other patents up to the given year.

Made Indicator

This variable is a proxy for the firm’s interest and ability to produce original (i.e., explorative) innovations.

37