Master thesis
MSc BA: Strategic innovation management
The relationship between internal funding and R&D intensity,
moderated by the level of investor protection
Authored by
Hidde Drees (
s2984792)
Supervisor: F. Noseleit
Co-assessor: H. van der Bij
Word count: 14.200
Abstract
1
1. Introduction
It is commonly recognized that innovation is crucial for economic growth. Hence, research and development (R&D) is a crucial driver for technological change (Solow, 1956; Brown, Fazzari & Petersen, 2009). A firm's expenditure behavior depends on many different determinants, and the financing issue of R&D is one of them (Bakker, 2013). Schumpeter (1939, 1942) was the first to argue that firms are not able to attract sufficient external funds to finance R&D undertakings, due to market imperfections. Therefore, firms depend on internal funds. The more recent literature agrees with Schumpeter’s argumentation and it is found that cash flow is used as a measure of a firm’s internal funds available to finance R&D (Fazzari, Hubbard & Petersen, 1988; Gilchrist & Himmelberg, 1995; Bloch, 2005). Consequently, the study of Hall (2002) indicates that a firm’s inclination to be dependent on internal finance differs. A possible explanation for this phenomenon includes the institutional differences per country (Hall, 2002).
By learning from the research gap that Hall (2002) identifies, it is this papers’ objective to study the indirect effect of the legal system on the investment decisions being made between outside investors and a firm’s managers and corporate shareholders (the insiders). In order to assess this objective, this study is considered in answering in what way investor protection moderates the use of cash flow as internal funding indicator for financing R&D. This is done by analyzing on both firm-level and country-level.
2
protection may be recognized as a promoter of R&D financing. And since innovation advances society, economic growth is also fostered (Baumol, 2002).
The main reason why the current research on the issue of how the legal system impacts the R&D financing decisions being made on firm-level is remarkably scarce, is due to its difficult observables interactions. For that reason, this study its purpose is to investigate a method in order to research these hardly observable interactions empirically. To my knowledge, only few studies considers the interaction effect between institutional factors on country level, and the finance decision making of R&D on firm-level (Hillier, Pindado, De Queiroz & De La Torre, 2011; Demirgüc et al. 2002). This study aims to provide additional evidence by contributing to the already existing literature in this field of interest. By conducting a longitudinal study there is investigated in what way the relationship between internal funds dependence (firm-level) and R&D intensity is moderated by the level of investor protection (country-level).
It is important to conduct research in this particular field, as the current literature has suggests that financial constraints influences the firms’ capacity to investing R&D. However, different results are found in the decision making of how firms finance R&D. For instance, stronger dependence on internal funding is found for small firms, for young firms and for firms that are located in different countries, such as in the United Kingdom and Germany (Hall, 2002). Obviously, many factors come into play that explain these differences and a firms (limited) access to external financing done by outside investors is one of them. It is important to study how outside investors are influenced by the legal system, since a study done by Beck & Levine (2002) shows that efficient legal systems notably fosters the growth of small companies. In addition, it is found that small firms tend to depend more on external financing (Beck & Levine, 2002). The study of Demirgüc et al. (2002) supports these findings by explaining that an effective legal system is important, since firm wishes to attract potential outside investors. However, a firm must be able to control opportunistic behavior that could be exploited by the firm’s insiders.
3
Heritage Foundation database for the indexes of institutional differences, in order to measure the level of investor protection.
The negative binomial regression results indicate that R&D and its internal funding, cash flow have a positive relationship. Thus, this study provides additional evidence that cash flow is an important main determinant of a firms R&D, and that using cash flow as indicator may help to explain a firm’s expenditure behavior. Consequently, there is found that the relationship between a firms’ internal funding of R&D is negatively moderated by the level of investor protection. Therefore, this study suggests that as the level of investor protection increases, the relationship between a firm’s inclination to be reliant on internal funding decreases. The negative interaction effect is explainable, since one can argue that a firm’s opportunity to benefit from external financing provided by outside investors increases whenever a firm is located within a country that facilitates a decent level of investor protection. Therefore, a firm may fund R&D less internally by using its cash flow, and so the relationship between R&D and cash flow decreases as investor protection increases. This effect shows evidence that an effective legal system should be able to mitigate informational asymmetry and/or moral hazard that may occur between firm’s insiders and outside investors.
4
countries. Whereas only in a legal environment that is able to protect its outside investors, the investors are more willing to take the risk and face uncertain prospects.
This study is structured as follows: Firstly, the literature review examines R&D as activity, and how its investments are characterized. Furthermore, underlying theories are discussed, such as how R&D investments are financed – internally or externally. Alongside the hypotheses are developed based on the previously discussed theories. Then, the conceptual model, the used control variables and econometric model are explained. Afterwards the methodology section is described. Subsequently, the results are interpreted and discussed. Finally, this study ends with a conclusion, together with its theoretical and managerial implementations.
2. Literature review
Research & Development as activity
The Frascati Manual of the Organization for Economic Co-operation and Development (OECD) defines R&D as “creative work undertaken on a systematic basis in order to increase the stock of knowledge, including knowledge of man, culture and society, and the use of this stock of knowledge to devise new applications” (OECD, 2002, see page 30).
In the broadest sense, R&D involves activities executed by firms in order to create radical or incremental products and/or processes innovations. The OECD (2002) lays out three basic processes of R&D: basic research, applied research, and development. Basic research is research that is attempted to acquire new knowledge for improved understanding or prediction of phenomena without a need for practical implementation. Applied research is research directed to develop more practical implementation, like technology or inventions, which are based on existing research results (OECD, 2002).
Research & Development as investment theory
5
research. Basic research is one fifth of the total R&D investment done by firms in developed countries (OECD, 2003).
In practice, roughly fifty percent (or more) of the total R&D investments is spent on the wages and salaries of highly educated scientists and engineers (Hall, 2002). Thus, the intangible assets of R&D are individual dependent and are critical to the firm, since its effort of the knowledge base generates profits for the future years (Hall, 2002). Additionally, the degree to which the knowledge can be addressed as “tacit” or “codified” also has significant influence on R&D investments, since it can be detrimental to the firm when it loses or is urged to fire its key employees that hold valuable knowledge (tacit), which may be difficult to transfer (Hall, 2002). Therefore, firms tend to keep their key employees and tend to smooth its R&D spending once in a while, which leads to high adjustment costs (Lach & Schankerman, 1989).
Nelson (1959) and Arrow (1962) were the first to address that R&D investments are strongly associated with indivisibility, inappropriability and uncertainty prospects due to its output. Indivisibility refers to returns to scale, because information of new products and processes can be expanded very quickly over many units at increasingly lower cost per unit (Nelson, 1959; Arrow, 1962). This can result in monopolistic competition in industries that are R&D intensive, which can lead to both over- or underinvestment in R&D (Reinganum, 1989).
On the contrary, inappropriability leads to underinvestment in R&D, because most inventions can be imitated once they are made. Therefore, firms do not want to bear the risk of proactively investing in R&D without profiting from its benefits (Teece, 1986). Imitators can benefit of these positive externalities or “spillovers”, and overall results imply that underinvestment rules overinvestment in the majority of sectors (Griliches, 1992).
6
Internal and external funding of R&D
Schumpeter (1939, 1942) was the first to argue that internal finance is one of the most important determinants of R&D investments. The common argument that is proposed by the literature is that due to capital market imperfections, firms are not allowed to depend on external funds to invest R&D resulting in being financial constrained. Hubbard (1998) states that capital market imperfection refers to the consequences of informational asymmetry in insurance and credit markets. The problem of asymmetric information experienced by borrowers and lenders, leads to a gap between the cost of internal and external financing, forcing firms to rely on internal funds (Hubbard, 1998). The argument of firms being financially constrained is supported by the results of Cohen (1995), showing a positive relationship between R&D investments and internal finance.
Fazzari et al. (1988) explains that a firm’s investment behavior depends on finance factors, including internal finance and access to new debt and equity finance. The study’s main argument is that a firm’s internal cash flow influences R&D investments, since internal funds may have cost advantages over debt or equity finance. Thus, internal financing decisions and R&D investments are interdependent (Fazzari et al., 1988). Furthermore, their study explains that internal financing is in general less costly in comparison to external financing, due to transaction costs, tax advantages, agency problems, costs of financial distress and asymmetric information between managers and potential outside investors and creditors (Fazzari et al., 1988). These issues determine whether the firms should finance the task internally by a firm’s insiders, or externally by interested parties within the market e.g. outside investors (Coase, 1937). The main reason why external financing is so expensive are transaction costs. Transaction costs are caused by search and information costs, bargaining costs and policing or monitoring costs (Coase, 1937). In addition to that, Williamson (1976) explains that humans are inclined to behave opportunistically and do not fully possess all information. Additionally, their cognitive abilities are limited, and decisions have to be made within a certain period of time, a concept that is known as bounded rationality (Williamson, 1976).
7
3. Hypotheses development
The effect of cash flow
Several empirical research points out that cash flow is an important determinant of R&D expenditures (Hall, 1992; Fazzari et al., 1988). Cash flow is the movement of a certain net amount of cash and cash-equivalent in and out of a business. Positive cash flow indicates the increasing amount of liquid assets and thus the firm’s ability to settle debts. Negative cash flow indicates a decreasing amount of liquid assets (Almeida, Campello, & Weisbach, 2004).
8
Since the literature considers cash flow to be an important determinant of R&D investments and several empirical research suggests that it is has a positive relationship (e.g. (Ascioglu, Hegde, & McDermott, 2008; Hillier et al., 2011) the following hypothesis is set:
Hypothesis 1: Cash flow has a positive relationship with the firm’s R&D intensity.
The moderating role of investor protection
In this chapter the second hypothesis is developed. For this it is examined why the level of investor protection may be beneficial to both the firm’s insiders and outside investors. Furthermore, it is examined in what way investor protection interacts with a firms internal funding – the use of cash flow. The concept of “investor protection” is considered a broad sense, its building blocks are measured in terms of “strong property laws”, “juridical effectiveness” and “government integrity”. These building blocks are gathered from the Heritage database indexes. Further details of these indexes are addressed in the methodology section.
Theory of investor protection
Preexistent literature clarifies that macro institutional factors influence the investments being made by a firm’s insiders and by suppliers of external finance (Beck & Levine, 2002; La Porta et al., 1997, 1998). As Demirgüc et al. (1998) explains, an underdeveloped legal system prevents outside investors from investing in firms, because of the high level of risks and uncertain prospects. This leads to a loss of R&D investment opportunities for firms, as they can benefit from external funding. Furthermore, the study finds that a legal environment in which the firm operates, has a direct consequence on the informational asymmetry between the firms' insiders and potential investors (Demirgüc et al., 1998). Due to the informational asymmetry, the outside investor is less protected because opportunistic behavior could be exploited by the firm's insiders, namely managers and controlling shareholders (Demirgüc et al., 1998).
9
literature that is addressed, this study suggests that the effectiveness of a legal environment should be able to suppress the problem of informational asymmetry and opportunistic behavior. Furthermore, the literature suggests that external financing of R&D is associated with moral hazards, because of limited accountability (Hubbard, 1998). For instance, if R&D projects are financed externally by an investor, the firm’s insiders may be more willing to take risks since its original funding is not from the firm’s own internal funding (Holmstrom & Tirole, 1997). Therefore, this study suggests that an effective legal system should be able to facilitate investor protection against the threat of moral hazard.
The literature further emphasizes that minority shareholder protection plays an important part in the protection of investors (La Porta et al., 1998). Since a minority shareholder does not have much control in the decision making of a firm, their interest might be ignored by its management (Jensen & Meckling, 1976). Therefore, law enforcement is a necessity, so that the voting, reorganization and creditor rights of shareholders are enforced by regulators or courts, so that external financing costs are reduced. This stimulates investors to finance firms, and thus external finance is supplied more easily.
Consequently, La Porta et al. (1999) explains that legal rules and regulations are needed in order to enforce contracts, since financial contracting is most often involved between the firms' insiders and outside investors once they decide to engage in business (Jensen & Meckling, 1976; Easterbrook & Fischel 1991). However, financial contracting must not be overly regulated, because this causes unnecessary obstacles between the firm’s insiders and outside investors. Thus, effective regulations are necessary in order to maintain the investment and financial freedom (La Porta, 1999). Consequently, Coase (1961) explains that the degree to which contracts are enforced differs per country, since it is possible that some courts are often unable or unwilling to solve difficult disagreements between the involved parties. Reasons why a court may be ineffective are slow court processes, corruption or political pressures (Coase, 1961).
10
legal control, it is most likely that outside investors are protected from informational asymmetry and moral hazard, and corrupt, inefficient juridical and court systems.
Taking all previous arguments into account, this study suggests that whenever a firm is located within a country with an effective legal system, the probability that investors are protected increases. Therefore, a high level of investor protections leads to a higher willingness of outside investors to supply external finance to firms. Consequently, a firm’s dependence on internal funding, which in this case would be cash flow, decreases. Therefore, the following hypothesis is set:
Hypothesis 2: Investor protection influences the positive relationship between cash flow and
R&D intensity in such a way that the R&D becomes less dependent on cash flow as the level of investor protection increases.
4. Conceptual model
In the developed conceptual model (figure 1) cash flow is set as the independent variable of R&D and is moderated by the level of investor protection. This study predicts that cash flow positively influences the R&D intensity. Furthermore, this study suggests that the level of investor protection negatively moderates the relationship between the firm’s inclination to be dependent on internal funding (cash flow) for financing R&D. Consequently, as the level of investor protection increases, the relationship between dependency on internal funds and R&D intensity decreases. As addressed in the hypothesis section, this is because higher investor protection stimulates outside investors to invest in firms, because informational asymmetry and moral hazards are limited. Meanwhile, firms benefit from this since their probability to get access to external financing increases.
Figure 1: Conceptual model Internal funding
(Cash flow) (R&D expenditure / Firm size) R&D intensity
Level of investor
protection variables Control
H1 +
11
5. Methodology
In order to contribute to the literature on in what way a legal system indirectly impacts a firms internal funding dependence of R&D financing, a sufficient sample is gathered of firms that operate within the pharmaceutical industry.
A longitudinal study approach is conducted to capture the proposed effects of cash flow and investor protection on R&D intensity over a limited period of time, namely from 2007 to 2016. This paper is structured as a theory-testing study and follows the deduction, testing and evaluation steps of the empirical cycle (Aken, Van Berends & Van der Bij, 2012). In order to go through the theory-testing process, a conceptual model is developed and two hypotheses that stem from findings of previous empirical studies are tested (Camisón-Zornoza, Lapiedra-Alcami, Segarra-Cipres & Boronat-Navarro, 2004). A total of 270 small, medium or large sized firms in the pharmaceutical industry that are located in 14 different countries in Europe, United States, China or Japan are included in the sample. The reason why many different countries are included is to keep the findings generalizable for this particular industry. This study considers the pharmaceutical industry to be an ideal candidate to conduct research in, since this particular industry is under growing pressure of increasingly cost-constrained healthcare systems and more demanding regulatory requirements (Paul et al., 2010). For this reason, the continuity of pharmaceutical firms are in jeopardy and in order to tackle these challenges, the pharmaceutical industry must focus on R&D productivity, without facing unsustainable R&D costs (Paul et al., 2010). As healthcare budgets become increasingly strained it is appropriate to conduct research in this particular is industry, since especially this industry can benefit highly from external financing suppliers.
Select studies as input for analysis
12
samples domain had to be limited. By using peer reviewed studies with a high amount of citations, it is found that cash flow is an important indicator of explaining R&D investments and that the firm’s investment decision making is reliant on institutional measures that differs per country. Additionally, it is found that empirical research in regards to how country institutional differences have impact on firm-level decisions being made in order to finance R&D is scarce. Therefore, this study is directed at conducting research in this particular field.
Data collection
Orbis database - In order to collect data of firms, Orbis is used. Orbis is a database that contains
information of 200 million firms worldwide, with an emphasis on private firm information. In order to ensure that the use of secondary data collection can be replicated, a description is given.
Firstly, data is found with a focus on “Manufactures of basic pharmaceutical products and pharmaceutical preparations“. Using the industry classification option, a list consisting of 98.036 results is generated. Secondly “country”, “last available year” “(from year -1 to year -9)”, “R&D expenses”, “cash flow”, "long term debts", “number of employees”, “date of incorporation” and “net sales” are selected as the necessity columns needed to obtain the variables. After selecting these columns, the Orbis data is exported to excel by taking an initial sample of 500 random pharmaceutical companies. In the initial sample, firms are excluded that do not invest, or barely invest in R&D (less then 25.000 USD), since those are firms in which there is no interest. Furthermore, firms are excluded by limiting the sample by selecting specific countries, which will be addressed detailed later on. Taking these steps, the initial sample of 500 is brought down to a sample of 270 firms.
13
10.748.809.135 USD. Cash flow may vary from a negative number of minus 11.237.188.000 USD to a positive number of 21.594.518.710 USD. Finally, long term debt may vary from 0 to 37.010.561.000 USD.
A total of 14 countries are used for this research and are shown in table 1. The table shows that a total of sixty-five (23,5%) firms are located in Europa and a total of sixty-three (22,7%) firms are located in Asia (22,7%). Additionally, the table shows that the sample is dominated by firms located in the United States (53,79%)..
Country Nr. of companies Perc. of companies
Belgium 2 0.72% China 33 11.91% Denmark 7 2.53% France 10 3.61% Germany 6 2.17% Ireland 5 1.81% Italy 1 0.36% Netherlands 3 1.08% Spain 4 1.44% Sweden 5 1.81% Switzerland 6 2.17% United Kingdom 16 5.78% Japan 30 10.83%
United States of America 149 53.79%
Table 1: Sample of firms categorized per country
“Number of employees” is used as measurement for firm size. The firm size sample includes small, medium and large sized firms and the number of employees may vary from 1 (micro firms) to 135,696 employees. The taken sample includes approximately 72 small-sized firms (<50 employees), approximately 77 medium-sized firms (<250) and approximately 127 large-sized firms.
The Heritage Foundation - In order to gather data for the institutional differences, The
14
protection could be measured against countries that provide a high level of investor protection. And finally, it is important to point out that for every year (2007-2016) the same index score is used, thus there is no variation between the years regarding the average score of investor protection.
Measurements
Dependent variable
R&D intensity – R&D expenditures divided by firm size is used as the dependent variable
in this study (R&D intensity). In general, the literature uses input measures such as R&D expenditures as indicator for R&D, due to its homogenous (economic value) character. An alternative measure of R&D that could be used is the output of patent counts. However, the number of patents is heterogeneous as its counts may differ variously across industries, and many of them never get implemented. Additionally, the use of patents may have diverse objectives such as blocking competitors and thus a patent count is not necessarily “innovative” by definition. On the contrary, a high amount of R&D expenditures does not necessarily lead to a high innovative output either (Becker, 2013). Nevertheless, this study preferred using R&D expenditures as measurement, due to its homogenous character.
As the data of R&D expenditures is greatly spread, the dependent variable R&D intensity is normalized by dividing R&D expenditures with firm size. The relationship between R&D expenditures and firm size is mixed, as several studies suggests that the relationship is linear and positive, however, other studies also show different results such as an independent relationship (Lee & Sung, 2005). The R&D and firm size relationship is addressed in more detail later on in this section.
Additionally, the dependent variable R&D intensity is also calculated by dividing R&D expenditures with “Net sales” as the literature suggests that growth R&D expenditures is highly associated with past sales (Coad & Rao, 2010; Domadenik et al., 2008). By calculating R&D intensity in two different ways, the main regression results can be confirmed.
Independent variable
Cash flow – In order to measure a firm’s internal funding, cash flow is used as the
15
agreement that cash flow is an important determinant for explaining R&D expenditures (Fazzari et al., 1988; Bloch, 2005; Himmelberg & Petersen, 1994). This study predicts that cash flow is positively related with R&D expenditures. For instance, a study by Ascioglu et al. (2008) shows evidence that firms that have high amounts of cash flow are more inclined to invest in R&D activity. Some alternative ways of measures of financial constraints are identified as well, such as external equity or net profit (Brown et al., 2009; Bougheas, Görg & Strobl, 2003). However, since the literature on cash flow influencing R&D expenses is regarded to be very mature, this study uses cash flow as the main determinant of R&D expenditures.
Moderating variable
Investor protection – This study suggests that investor protection negatively moderates
the relationship between the firm’s inclination to be dependent on internal funding (cash flow) for financing R&D. The level of investor protection may help in reducing informational asymmetry and moral hazards between the firm’s insiders and outside investors. In order to measure the average level of investor protection, this is calculated by using the index scores of property rights, government integrity and judicial effectiveness as building blocks. This study considers the use of these index scores as measurement for the level of investor protection to be logical, since they are in line with the reasoning learned from several academic papers, so that the hypotheses could be developed. As The Heritage Foundation (2017) database shortly prescribes, property rights are an important subcomponent of investor protection, since this is the extent to which a country’s legal systems allows individuals to freely possess property, secured by laws that are enforced. Juridical effectiveness is an important subcomponent of investor protection, since a well-functioning legal system is necessary for protecting the rights of all citizens against unlawful acts by others, including by governments and powerful private parties. Government integrity is used since corruption crumbles the economic freedom and thus harming outside investors due to introducing insecurity and uncertainty (Heritage, 2017).
Control variables
Long term debt – A study of Islam & Mozumdar (2007) investigates whether the impact
16
development, making external financing more expensive due to transaction costs. As the cost of external financing increases, firms are urged to be more dependent on their internal funding. This results in firms cutting back their investments, due to cash flow shortfalls (Islam & Mozumdar, 2007). Firms that experience problems in paying off loans and/or financial obligations due to cash flow problems tend to fund most of their ventures by starting new debts, resulting in an increasing amount long term debts. Long-term debts consist of loans and financial obligations lasting over one year (Levine & Zervos, 1998). Therefore, the firm bears the risk of bankruptcy (Blundell, Griffith & Reenen, 1999; Rajan & Zingales, 1998). As a firm becomes more dependent on its internal funding and bears probability of building higher long term debts, the degree to which the firm is financially constrained increases. Financially constrained firms are limited in their ability to invest in R&D and taking previous arguments into account, the literature finds that long term debts are an important determinant on the financing of R&D and thus it is used as control variable. Furthermore, it is expected that it has a negative relationship with R&D intensity (Hillier et al., 2008).
Firm size - The argument that R&D expenditures increases more than proportionally with
17
expenditures are internally financially constrained (Hyytinen & Toivanen, 2005). Several literature studies suggest that firm’s capacity to invest in R&D is positively related with its size (Fisher & Temin, 1973; Dosi, 1988; Acs & Audretsch, 1988). However, some studies tend to contradict this notion and provide a different perspective on how firm size is related to a firm’s R&D performance (Lee & Sung, 2005). For instance, some studies suggested a non-linear or insignificant relationships (Klette & Griliches, 2000). To see how firm size and R&D are related in the used sample, a scatterplot is generated by using Stata, the plot can be found in appendix 1. The scatterplot shows an uphill pattern from left to right. This indicates a positive relationship between firm size and R&D expenditures. Despite that the plot shows a possible relationship between the two variables, this does not necessarily imply that a cause-effect relationship exists. Nevertheless, in this study it is expected that the relationship between R&D investment and firm size is linear and positive, and it is used as measure for the dependent variable, R&D intensity. Finally, firm size will be used as control variable so that the dis/economies of scale can be captured. This is necessary, because if the dependent variable is not normalized, a positive relationship between R&D expenditures and firm size will always occur.
Firm age – a study by Brown, Fazzari and Petersen (2009) shows that cash flow has an
18
(Quevedo, Pellegrino & Vivarelli, 2014). Therefore, firm age is used as a control variable and it is expected to be positively related with R&D.
Negative binominal regression
This study conducts a negative binomial regression and uses the statistical software package StataSE 14. The website of UCLA: Statistical Consulting Group is used as an important source as support, in order to perform the procedure of the negative binominal correctly.
In this study count data had to be dealt with. Count data reflects the number of occurrences of a behavior in a fixed period of time (Coxe, West & Aiken, 2009). For instance, the dependent variable "R&D intensity" can only take non-negative integer variables, since the value of an investment or firm size can never be below zero. In the result section, the count variable is found to be over dispersed, meaning that standard deviation exceeds the conditional mean. Since the count data is over dispersed, a negative binominal regression is preferred over a Poisson regression (Hilbe, 2007; Gardner, Mulvey, & Shaw, 1995).
As the standard deviation exceeds the mean of the dependent, it is inappropriate to consider the use of a Poisson regression over the use of a negative binomial regression, since a Poisson regression may generate inconsistent outcomes (Hausman, 1984). Reasons include that the negative binominal regression has a similar mean structure as Poisson regression, the negative binominal regression provides an extra parameter to model the over dispersed count data. Therefore, the confidence intervals for the Negative binomial regression are more likely to be narrower, compared to the confidence intervals of the Poisson regression (UCLA, 2017). Thus, this study assumes that the dependent variable (R&D intensity) is over-dispersed and does not have an excessive number of zeros. Finally, it is sensible to apply the negative binominal regression, since the panel data sample is quite large (270 firms). Furthermore, it is important to mention that the negative binominal regression does not have an equal R squared measure in comparison with OLS regression. Thus, the Pseudo R-square is interpreted with caution when necessary (Long & Freese, 2006).
19
Inflator Factor test is conducted to quantify the strictness of multicollinearity and to see whether these values are below 10 (Kutner, Nachtsheim & Neter, 2004). In addition, boxplots are used to check for outliers that might influence the regression results.
For the statistical analysis, which can be found in the results section, histogram plots are carried out in order to find whether the data is positively or negatively skewed. Whenever a variable is skewed, the logarithm is most often to be used in order to use a distribution with no skew. However, it is important to mention that the logarithm cannot be taken from the independent variable cash flow, since the variable frequently contains negative values. Therefore, there is decided to limit the use of logarithms.
Furthermore, a robust regression is conducted. A robust regression can be useful as it aims to mitigate the effects of outliers, autocorrelation and/or heteroscedasticity.
Econometric model
This study contains panel data and the R&D investments of firms are observed within a period of time (one year time lag): 2007-2016. The following econometric model is used in this study:
(𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦)𝑖𝑡 = 𝛽0+ 𝛽1𝐶𝐹𝑖𝑡 + 𝛽2𝐼𝑃𝑖𝑡+ 𝛽3𝐿𝑇𝐷𝑖𝑡+ 𝛽4𝑆𝐼𝑍𝐸𝑖𝑡+ 𝛽5𝐴𝐺𝐸𝑖𝑡+ 𝛽6(𝐶𝐹 ∗ 𝐼𝑃)𝑖𝑡+ 𝜀𝑖𝑡
Where R&D intensity, SIZE, CF, IP, LTD, AGE denote R&D expenditures divided by firm size, Cash flow, Investor protection, Long term debt, Firm age, followed by the interaction term: 𝛽(𝑥1 ∗ 𝑥2) and finally, 𝜀𝑖 is the error term. The subscript of 𝑖 refers to the firm, and the subscript
20
6. Results
A descriptive table (table 2) is generated in order to quantitatively describe the features of the collected information.
Variable Observ. Mean Std dev. Min Max
(1) RD intensity 2,148 206303.2 301621.6 0 4933572 (2) CF 2,362 5.45e+08 2.21e+09 -12e+10 2.16e+10 (3) IP (log) 2,381 4.316297 .1487394 3.97 4.5
(4) LTD 2,379 8.54e+08 3.40e+09 0 3.70e+10
(5) SIZE 2,316 6613.481 20514 1 135696
(6) AGE 2,381 24.0042 24.01563 0 153
Table 2: Summary of statistics
Table 2 shows that a sufficient number of observations is generated, most observations are determined at approximately 2300. Furthermore, the standard deviation is significantly higher than the mean, with the exception of the investor protection, firm age and number of employees variables, and thus the data is greatly spread. Additionally, the mean of the dependent variable: R&D intensity and other variables are exceeded by the standard deviation. These differences suggest that the data is over-dispersed, and therefore the use of a negative binominal regression is appropriate. Finally, the standard deviations and min – max values of the variables cash flow and long term debt are determined in numbers of billions USD. Additionally, a Pearl correlation table (table 3) is generated in order to measure a linear correlation between the variables.
Variable 1 2 3 4 5 6 (1) RD intensity 1.0000 (2) CF -0.1143 1.0000 (3) IP (log) 0.1768 0.1136 1.0000 (4) LTD -0.1012 0.8284 0.1179 1.0000 (5) SIZE -0.1638 0.8496 0.0501 0.7391 1.0000 (6) AGE -0.1917 0.3837 0.0362 0.3360 0.4549 1.0000
21
By learning from the Pearson’s correlation matrix (table 3), it can be seen that long term debt and firm size highly correlate with cash flow. However, since this study uses panel data, these correlation results do not necessarily imply that there is multicollinearity “within” the firm-level, but instead simply “between” the firms. Therefore, the differences of the variables cash flow and long term debt throughout the years are calculated, and used in the correlation model which can be found in appendix 2. From this correlation model it is assumed that there is most likely not a very strong correlation “within” the panel data overtime. In order to be completely sure, a Variance Inflection Test (VIF) is conducted in order assess the potential multicollinearity issue. By looking at the VIF test results in appendix 3, none of the values seem to exceed the critical value of 10, thus the issue of multicollinearity is not considered (Chatterjee et al., 2000).
In order to determine whether the data is skewed, histograms are carried out which can be found in appendix 4. The histogram shows that every variable used is positively skewed. However, the logarithm cannot be taken from cash flow, since it has negative values. Furthermore, it is found that the indexes of investor protection fluctuates a lot, thus the logarithm of this variable has been taken.
Additionally, boxplots are used to identify outliers, which can be found in appendix 5. From the boxplots it can be learned that the variables cash flow and long term debt have values that are far away from other data values, meaning that these variables may strongly affect the results of the regression.
Regression results
22
coefficients could be interpreted more accurately. The results of the negative binominal regression are represented in table 5.
Table 5: Negative binomial regression
Variables Model 1 Model 2 Model 3 Model 4
LTD -0.0275616* -0.0041898 -.0056458 -.0244012 (0.0150063) (.0167029) (.0148206) (.0151177) SIZE -0.0000103*** -.000018*** -.000015*** -.000058***
(2.45e-06) (3.56e-06) (3.11e-06) (6.39e-06) AGE -.0142035*** -0.0137524*** -.0123234*** -.0087795*** (.0015701) (.0015725) (.0013814) (.0014529) CF .1110929*** 3.328424*** 2.362042** (.037006) (.9633187) (.93926) IP (log) 9.967434*** 9.64782*** (.2664228) (.2675066) CF*IP (log) -.7519862 -.5255151** (.2238137) (.218219) SIZE*SIZE 3.57e-10*** (4.88e-11) Observations 1930 1922 1922 1922 Wald's Chi-square Wald’s p-value 323.49 (0.00) 350.95 (0.00) 1828.24 (0.00) 1800.24 (0.00) Log-likelihood -11838.287 -11778.318 -11452.581 -11426.051 Pseudo R-squared 0.0097 0.0101 0.0375 0.0397
Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
23
In the second model, the negative binominal regression results for testing hypothesis 1 are generated. A total number of 1,922 observations are generated. As we can learn from the results, cash flow positively significant on a 5% level. This result is in line with what is expected, and therefore we find support for hypothesis 1: Cash flow has a positive relationship with the firm’s R&D intensity. By learning from the coefficient, the result can be interpreted as follows: each one-unit increase on cash flow, the R&D intensity increases by 0.111, while keeping all other variables in the model constant. Furthermore, long term debt is found to be insignificant and firm size and firm age are found to be negatively significant. These results are not in line with what is expected.
In the third model, the negative binominal regression results for testing hypothesis 2 are generated. As we can learn, cash flow and investor protection are positively significant with R&D intensity on a 1% level. The positive relationship between investor protection and R&D intensity can be interpreted as follows: as each one-unit increases on the investor protection, the R&D intensity increases by 9.97, while keeping all other variables in the model constant. Additionally, we can see that the interaction effect of investor protection with cash flow is negatively significant with R&D intensity on a 1% level. This result is in line with what is expected, and thus support is found for hypothesis 2. By learning from the coefficient, the result can be interpreted as follows: each one-unit increase on the interaction effect between cash flow and investor protection, the R&D intensity decreases by 0.752, while keeping all other variables in the model constant.
In the fourth model, the interaction effect of firm size with firm size is included in order to find out in what way firm size is related with firm intensity. The result finds that the interaction effect is positively related with R&D intensity. Finding this result is interesting, as this result indicate that the relationship between R&D and firm size may be less than proportional or u-shaped.
Additionally, log-likelihood is included. In model 1, the log-likelihood value of minus 11838.287 has no meaning in and of itself. However, this number can be used to help compare the other nested models. The more maximized the log value is to zero, the better fitted the model.
24
variables create a statistically significant improvement in the fit of the overall model. Finally, the pseudo R-squared values are found to be relatively low. The value of the pseudo R-squared increases as model 1 and 2 are compared to model 3 and 4, indicating a more superior fit.
Robustness
As mentioned in the methodology section, the taken sample may contain the issue of heteroscedasticity and autocorrelation. Furthermore, outliers have been detected regarding the variables of cash flow and long term debt. Therefore a robustness check is performed as the test may be helpful in limiting the issue of outliers, heteroscedasticity and autocorrelation (see table 6). The robustness test is conducted similarly to the main regression test, starting off with the testing of control variables.
Table 6: Robustness negative binomial regression
Variables Model 1 Model 2 Model 3 Model 4
LTD -.0275616** -.0041898 -.0056458 -.0244012 (.0096705) (.0115303) .0103327 (.0108294) SIZE -.0000103*** -.000018*** -.000015*** -.000058***
(1.85e-06) (3.07e-06) (2.38e-06) (5.41e-06) AGE -.0142035*** -.0137524*** -.0123234*** -.0087795*** (.0012799) (.0012808) (.0012242) (.0013347) CF .1110929*** 3.328424** 2.362042* (.0312579) (1.48705) (1.34916) IP (log) 9.967434*** 9.64782*** (.2643034) (.2553074) CF*IP (log) -.7519862** -.5255151* (.3403144) (.3086663) SIZE*SIZE 3.57e-10*** (3.79e-11) Observations 1930 1922 1922 1922 Log-likelihood -11838.287 -11778.318 -11452.581 -11426.051 Wald's Chi-square Wald’s P-value 562.54 (0.00) 547.30 (0.00) 2040.77 (0.00) 2152.95 (0.00) Pseudo R-squared 0.0097 0.0101 0.0375 0.0397
25
As can be learned from model 1, all control variables are negatively significant, these results are similar to the main model. As model 2 is introduced, all results are significant with the exception of long term debt, these results are in line with the main model. By looking at the results of model 3, in which the interaction effect is included, we can see that the independent variable cash flow has become less significant in comparison to the main model test (on a 5% level). Furthermore, we can see that the interaction effect of cash flow and investor protection has become less significant (on a 5% level). And finally, the results of model 4 show that the firm size interaction is positively significant (on a 1% level) and may indicate less than proportional or an u-shaped relationship with R&D. Considering the results of the performed robustness check, no contradicting results were found and thereby making the main model robust.
As final confirmation for the main regression results, the dependent variable “R&D intensity” is changed by dividing R&D expenditures with Net Sales (see appendix 6). With the exception of testing hypothesis 1 in model 2, the generated robust results are similar to the main regression results. While testing hypothesis 1 in model 2, cash flow is found to be insignificantly related with R&D intensity, however in model 3 and 4 cash flow is found to be positively associated with R&D intensity. Therefore, partial support for hypothesis 1 still can be considered. Nevertheless, based on regression results in appendix 6, confirmation is found for the main regression results and so this study considers the overall regression results to be valid.
7. Discussion
As stated in the results section, the negative binominal regression results are in line with two expected hypotheses.
26
cash flow having a positive effect on R&D expenditure may strengthen the argument that this is due to a firm’s limited access to external financing, caused by informational asymmetries between firms and suppliers of external finance (Gilchrist & Himmelberg, 1995). However, another way of interpreting the effect of cash flow its positive relationship with R&D investment is that it is not caused by capital market imperfections, but as a predicting role of opportunities in R&D investments. For instance, a positive fluctuation in cash flow may indicate an expectation of greater R&D investments in the future (Bloch, 2005). However, using cash flow as measure for indicating credit market imperfections has been criticized. For instance, a study by Kaplan and Zingales (1997) shows that financial constraints may not be binding in all periods, and thus subgroups should not be fixed over the entire period. In addition, their study argues that there is no theoretical ground to assume that there is a monotonic relationship between cash flow and R&D investments, as they explain that financially troubled firms have the probability to experience lower cash flow sensitivities (Kaplan & Zingales, 1997). Fazzari et al. (2000) have responded to this by providing arguments that the model used by Kaplan & Zingales (1997) fails to capture the approach of most previous researches, due to their flawed integrating technique. Additionally, the study notes that the provided empirical findings are too difficult to interpret and are too inaccurate to determine whether firms are financially constrained or not (Fazzari et al., 2000). Nevertheless, this study finds support for hypothesis 1, implying that cash flow is an important determinant of R&D expenditure. Therefore, it is important to conduct further exploration in cash flow, since it may be an important indicator of explaining a firm’s R&D investment behavior.
27
development of financial systems, and that its development may differ across countries due to institutional differences (Porta et al., 1997, 2000; Kwok & Taddesse, 2000). Thus, enough empirical evidence is provided to assume that when firms are located in countries where the level of investor protection is low, the finance of R&D is more vulnerable to cash flow fluctuations (Durnev, Morck & Yeung, 2004). This is logical, since the degree of cash flow fluctuation could be clarified in a way that the firm’s R&D expenditures are influenced through the use of cash flow in order to finance R&D. However, firms that are located in a country that facilitates a decent amount of investor protection, a firm's R&D is less prone to be influenced by fluctuations of cash flow. This reasoning is in line with the study by Hillier et al. (2011) that also measured investor protection, and has used it as moderating variable for cash flow sensitivity and R&D. However, this study has a different research approach and used different data and therefore it is captivating to confirm that this study provides additional evidence. Thus, it can be stated that investor protection indeed is an important legal mechanism in decreasing a firms reliance of internal funding for financing R&D. Nevertheless, empirical evidence in this field of interest remains scarce, and deserves further study.
Regarding the control mechanisms, the expectations are generally not met. For instance, long term debt is found to be significant in the model 1 where the control variables are included, however in all other models the relationship is found to be insignificant.
28
pharmaceutical industry. These results are similar to the study of Acs and Audretsch (1991) in which a non-linear, u-shaped relationship is found between R&D and firm size. Possible reasons for this outcome are that a large firm size may suppress creativity or its ability to adapt to external opportunities due to a more bureaucratic organizational structure (Cohen & Klepper). Therefore, larger firms tend to be more effective in incremental innovation instead of radical innovation, like small firms are (Cohen & Klepper, 1996). Furthermore, Lee & Sung (2005) provide findings that are also somewhat similar to what is found in this study. Evidence is provided that show results of diminishing R&D productivity with respect to firm size, a finding that is consistent with previous studies (Scherer, 1965, 1984; Scott, 1984). The study of Lee & Sung (2005) suggest that the less than proportional relationship which is especially related to small firms. However, a “more than proportional relationship” is especially related to large sized firms due to learning economies of scale or scope and competence enhancing innovations in R&D undertakings (Lee & Sung, 2005). And interestingly, the study of Holbrook and Squires (1996) shows that the economies of scale effect may exhaust as the firm may grow, and that smaller firms may spend a higher percentage of their total investment R&D in comparison to larger firms. Nevertheless, as the literature remains to be inconclusive about the relationship between R&D expenditures and firm size, this field of interest deserves further exploration.
29
Therefore, some firms that are old of its origin may look young after the merger was succeeded by only learning from the data. Acknowledging this issue, it may lead to sample selection bias and so unreliable regression results may have been generated. Therefore, the negative relationship between R&D intensity and firm age must be interpreted with caution.
8. Conclusion
The main objective of this study is to gain insight in the indirect effect of the legal system on the investment decisions being made between outside investors and managers and corporate shareholders (the insiders). This study considers answering the question in what way investor protection moderates the use of cash flow as internal funding indicator for financing R&D by analyzing on both firm-level and country-level. Empirical evidence is obtained by using a sample of 270firms within the pharmaceutical industry and so a longitudinal study is conducted between the years 2007-2016. By using a negative binominal regression analysis, both predicted hypotheses are supported. Cash flow indeed has a positive significant relationship with R&D. Consequently, this study finds that investor protection negatively moderates the relationship between cash flow and R&D intensity. Thus, as investor protection increases, the relationship between internal funding and R&D decreases. This is an important finding because these results show evidence that investor protection may work as an important facilitator of R&D expenditure, and indeed decreases the firm’s dependency on internal funding, because then a firm may be able to gain access to external funding done by outside investors. Evidently, effective investor protection seems to weaken informational asymmetry and moral hazards between outside investors and a firm’s insiders and external funding therefore becomes a more accessible. Furthermore, this study brings theoretical and practical implications and contains limitations. These are discussed in the next sections.
Theoretical implications
30
able to profit from external funds, as the level of investor protection stimulate outside investors to bear the risk and face uncertain prospects.
Since this study shows empirical proof that an effective legal system increases the likelihood of firms attaining more external funds as the cost of external finance is reduced, one can assume that the effectiveness of a legal system is a very important component and influences how firms make financial decisions. Therefore, it is recommended that future researchers seek further exploration into the effect of the legal system on a firm’s finance decisions. Additionally, researchers may provide additional empirical evidence of how investor protection impacts a firm’s financial decisions, since research on this is still remarkably scarce.
Finally, as Hall (2002) has explained, a firm’s tendency to depend on internal funding may also be caused by a different structure of financial and capital markets, and due to different attitudes towards risk and uncertainty by employees. It is recommended to seek further exploration into this phenomenon, by analyzing on both firm-level (e.g. firm culture) and country-level (e.g. society culture).
Practical implications
In addition to theoretical implications, this study provides meaningful managerial and practical implications. Firstly and most importantly, these results might stimulate awareness of the crucial role that policymakers have in developing the law. As this study shows, the implementation of laws definitely has an impact on a firm’s financial decisions, and in this case the financing of R&D. Therefore, it is concluded that policymakers have the power to foster economic growth by building a law system that enhances R&D investing for firms. This can be achieved by developing a legal system that is orientated towards protection mechanisms of which potential stakeholders can benefit. By facilitating investor protection, holistic transparency is stimulated so that informational asymmetry is decreased. Furthermore, a country needs juridical effectiveness and government integrity so that, for example, moral hazards by firms’ insiders can be fought and outside investors can be protected.
31
take into account whether the country's law is able to protect investors or not, so that it is known whether they may be able to benefit from outside investors or not. Whenever a country's government is not able to provide investors protection, multinationals may not be agitated to build a plant in that specific country. Thus, managers should consider the legal environment in setting their strategic direction. Additionally, this study may also be helpful to angel investors that seek their profits by investing in corporations. By learning from this study, it is recommended to angel investors to pursue business in a country with a well-functioning legal system. Finally, this research contributes to the institutional theory by showing evidence of external pressures having impact on how decisions are made for financing R&D.
Limitations
This study contains several limitations. Firstly, only firms are considered within the pharmaceutical industry, and thus sectorial differences are excluded. Thus, if one decides to replicate this research in a different industry, different result may be validated. As one can argue R&D effects of the factors across industries are heterogeneous, it is recommended to do further research in this phenomenon by considering several industries.
Additionally, in the methodology section it is found that the sample is clearly dominated by firms located in the United States (53,79%). By acknowledging this issue, the findings with regards to institutional differences per country are considered to be less generalizable. Next this, this study only includes developed countries (with the exception of China). It would be interesting to see whether different result may be validated if one decides to conduct a similar study that includes undeveloped counties.
Furthermore, this study may leave out one or more important variables, and thus omitted-variable bias may occur in this research. For instance, control omitted-variables that may be considered as important determinants for R&D are ignored due to their unmeasurable (complex) character. Examples are a firm’s strategy, culture, willingness of employees to take risks and inclination to innovate are factors which have not been taken into account. As addressed in the discussion section, the control variable firm age may access the problem of sample bias. Therefore, this result is interpreted with caution.
32
that firms have their R&D located in their home country, which may not always be the case (Shefer & Frenkel, 2005). For instance, some multinationals have their R&D located across different countries. Thus, by automatically assuming this, some data used for measuring R&D expenditures may be misleading.
As a longitudinal study observes data without manipulating it, it is important to recognize that a longitudinal study may be less effective in finding causal relationships than experiments do. Therefore, an experimental study in the influence of institutional factors on firm level may be advisable to conduct.
References
Acs, Z.J., & Audretsch, D.B. (1991). R&D firm size, and innovative activity. In: Acs, Z.J., Audretsch, D.B. (Eds.), Innovation and Technological Change: An International Comparison. Harvester Wheatsheaf, New York.
Acs, Z.J., Audretsch, D.B., (1988). Innovation in large and small firms: an empirical analysis. American Economic Review 78(4): 678–690.
Aken J.E. van, Berends, H., Bij, H. van der (2012). Problem solving in organizations – A methodological handbook for business and management students.
Akerlof, G.A. (1970). The Market for ‘Lemons’: Quality, Uncertainty, and the Market Mechanism. Quarterly Journal of Economics, 84: 488–500.
Almeida, H., Campello, M., &. Weisbach, M.S. (2004). The cash flow sensitivity of cash, Journal of Finance 59: 1777–1804.
Almus, M., & Czarnitzki, D. (2003). The effects of public R&D subsidies on firms' innovation activities: the case of Eastern Germany. Journal of Business & Economic Statistics, 21 (2): 226-236.
33
Ascioglu, A., Hegde, S. P., & McDermott, J. B. (2008). Information asymmetry and investment– cash flow sensitivity. Journal of Banking & Finance, 32(6): 1036–1048.
Bakker, G. (2013). ‘Money for nothing: how firms have financed R&D-projects since the Industrial Revolution’, Research Policy 42: 1793-1814.
Baumol, W.J. (2002). The Free-Market Innovation Machine. Princeton: Princeton University Press.
Beck, T., & Levine, R. (2002). Industry growth and capital allocation: Does having a market- or bank-based system matter? Journal of Financial Economics, 64(2): 147–180.
Becker, B. (2013). “The Determinants of R&D Investment: A survey of the Empirical Research”, Economics Discussion Paper Series, School of Business Economics, University of Loughborough UK.
Bloch, C. (2005). R&D investment and internal finance: The cash flow effect. Economics of Innovation and New Technology 14: 213-223.
Blundell, R., Griffith, R., & Reenen, J. V. (1999). Market share, market value and innovation in a panel of British manufacturing firms. Review of Economic Studies, 66(3): 529-554.
Bougheas, S., Görg, H. & Strobl, E. (2003). Is R&D financially constrained? Theory and evidence from Irish manufacturing. Review of Industrial Organization 22: 159-174.
Brown, J.R. & Petersen, B.C. (2011). Cash holdings and R&D smoothing. Journal of Corporate Finance 17: 694-709.
Brown, J.R. Petersen, B.C. (2009). Why has the investment-cash flow sensitivity declined so sharply? Rising R&D and equity market developments. Journal of Banking and Finance 33: 971-984.
Brown, J.R., Fazzari, S.M & Petersen, B.C. (2009). Financing innovation and growth: Cash flow, external equity, and the 1990s R&D boom. Journal of Finance 64: 151-185.
34
Cleary, S., Povel, P., & Raith, M. (2007). The U-shaped investment curve: Theory and evidence, journal of Financial and Quantitative Analysis, 42(1): 1-40.
Coad, A., Rao, R. (2010).“Firm growth and R&D expenditure”. Economics of Innovation and New Technology.19: 127-145.
Coase, R. (1937). The nature of the firm. economics, 4(16): 386-405.
Coase, R. (1961). “The Problem of Social Cost.” Journal of Law and Economics 3, 1-44.
Cohen, W. & Levinthal, D. (1989). Innovation and learning: the two faces of R&D. The Economic Journal, 99: 569-596.
Cohen, W.M., & Levinthal, D.A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly 35(1): 128152.
Cohen, W. M., & S. Klepper. (1996). "Firm size and the nature of innovation within industries: The case of process and product R&D." Review of Economics and Statistics, 78: 223-243.
Chatterjee, S., Hadi, A. S., Price, B. (2000). Regression analysis by example. John Wiley & Sons.
Coxe, S., West, S. G., & Aiken, L. S. (2009). The analysis of count data: A gentle introduction to Poisson regression and its alternatives. Journal of Personality Assessment, 91: 121–136.
De Man, A. P., & Duysters, G. (2005). Collaboration and innovation: a review of the effects of mergers, acquisitions and alliances on innovation. Technovation, 25(12): 1377-1387.
Defond, M. L., & Hung, M. (2004). Investor protection and corporate governance: Evidence from worldwide CEO turnover. Journal of Accounting Research, 42(2): 269–312.
Demirgüc¸-Kunt, A., & Maksimovic, V. (1998). Law, finance, and firm growth. Journal of Finance, 53(6): 2107–2137.
35
Dosi, G. (1988). Sources, procedures and microeconomic effects of innovation. Journal of Economic Literature 25: 1120–1171.
Durnev, A., Morck, R., & Yeung, B. (2004). Value-enhancing capital budgeting and firm-specific stock return variation. journal of Finance, 59(1): 65-105.
Easterbrook, F., Daniel, F. (1991). The Economic Structure of Corporate Law. Cambridge, Mass: Harvard University Press.
Fazzari, S.M., Hubbard, R.G., Petersen, B.C. (2000). Investment-cash flow sensitivities are useful: A comment on Kaplan and Zingales. Quarterly Journal of Economics, 115: 695-705. Fazzari, S.R. Hubbard, G. & Petersen, B.C. (1988). “Financing Constraints and Corporate Investment.” Brookings Papers on Economic Activity, 1: 141–195.
Fisher, F.M., Temin, P., (1973). Returns to scale in research and development: what does the Schumpeterian hypothesis imply? Journal of Political Economy, 81: 56–70.
Gardner, W., Mulvey, E. P., & Shaw, E. C. (1995). Regression analyses of counts and rates: Poisson, over dispersed Poisson, and negative binomial models. Psychological Bulletin, 118: 392–404.
Gilchrist, S. & Himmelberg, C.P. (1995). Evidence on the Role of Cash Flow for Investment. Journal of Monetary Economics, 36: 541–572.
Griliches, Z. (1992). "The Search for R&D Spillovers." Scandinavian Journal of Economics 94 (Supplement): 29-47.
Hall, B.H. (1992). “Research and Development at the Firm Level: Does the Source of Financing Matter?” NBER Working Paper No. 4096.
Hall, B.H. (2002). The financing of research and development. Oxford Review of Economic Policy 18: 35-51.
36
Hausman, J. A., Hall, B. H., & Griliches, Z. (1984). Econometric models for count data with an application to the patents-R&D relationship.
Hilbe, J. M. (2007). Negative binomial regression. New York: Cambridge.
Hillier, D., Pindado, J., De Queiroz, V. & De La Torre, C. (2011). The impact of country level corporate governance on research and development. Journal of International Business Studies 42: 76-98.
Holbrook, J., & Squires, J. R. (1996). Firm-level analysis of Determinants of Canadian Industrial R&D performance. Science and public policy, 23: 369-374.
Holmstrom, B., Tirole, J. (1997). Financial Intermediation, Loanable Funds, and the Real Sector. Quarterly Journal of Economics, 112: 663–691.
Hubbard, R. G. (1998). Capital market imperfections and investment, Journal of Economic Literature 36: 193.
Hyytinen, A., Toivanen, O. (2005). Do financial constraints hold back innovation and growth? Evidence on the role of public policy. Research Policy 34: 1385-1403.
Islam, S. S., & Mozumdar, A. (2007). Financial market development and the importance of internal cash: Evidence from international data. Journal of Banking & Finance, 31(3): 641–658. Jensen, M., & Meckling, W. (1976). “Theory of the firm: Managerial behavior, agency costs, and ownership structure.” Journal of Financial Economics 3: 305-60.
Journal of Industrial Economics 50: 369-390.
Kaplan, S,. Zingales, L. (1997). Do investment-cash flow sensitivities provide useful measures of financing constraints? Quarterly Journal of Economics 112: 169-215.
Klette, T.J., Griliches, Z., (2000). Empirical patterns of firm growth and R&D investment: a quality ladder model interpretation. Economic Journal 110: 363–387.
