1
Innovation and Protection of Intellectual Property Rights
With Purchase Price Allocation data as proxy for Innovation.
The last several years’ intellectual property has grown to one of the most important sources of creating value. This thesis analyses the possible relationship between the level of protection of intellectual property rights (IPR) and innovation. The enforcement differs a lot worldwide and literature is still ambiguous whether strong protection is favorable or not. In our investigation we use Ordinary Least Squares (OLS) to estimate the model parameters since all assumption are satisfied. With OLS we find a positive and significant relationship between IPR enforcement index and innovation for Non-OECD countries. However, for the OECD sample and all countries together the IPR enforcement is of insignificance.
Daniel Drewes 10631453
Date: 31 January 2016 University of Amsterdam Supervisor: Ron van Maurik
2 This document is written by Student Daniel Drewes who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
3
Table of Contents
1.0 Introduction ... 4
2.0 Literature Review ... 5
2.1 Intellectual property rights and innovation ... 5
2.2 The positive effect of Intellectual Property enforcement on innovation ... 6
2.3 The negative effect of Intellectual Property enforcement on innovation ... 8
2.4 Differences in intellectual property rights enforcement for developing countries. ... 9
2.5 Conclusion ... 10
3.0 Methodology ... 11
3.1 Variables ... 11
3.2 Origin of the variables and modification of the data panel ... 14
3.3 Econometric model ... 17
3.4 Econometric method ... 18
4.0 Results ... 21
5.0 Robustness checks on empirical model ... 24
6.0 Conclusion ... 25
7.0 References ... 27
4
1.0 Introduction
The last decades the importance of knowledge and human capital for companies has risen significantly. Almost a century ago the value of a company was based on buildings, production plants, land and equipment. This old valuation of worth stands in strong contrast with our current valuation of firms. The importance of buildings, land and other tangible assets has declined during the years and intellectual property or knowledge has gained a lot in importance (Bontis, 1998).
The increasing importance of the knowledge economy resulted in various kinds of research on how to improve knowledge and thereby increasing the value of a firm. According to Schumpeter (1912) innovation is one of the main sources of economic growth and we have to stimulate innovation by all possible means to achieve a higher level of economic welfare. One of the most interesting and controversial ways of stimulating innovation is Intellectual Property Rights (IPR) protection or enforcement. The theories behind IPR enforcement show ambiguous results and even differ between OECD countries and developing countries. Most empirical research about this topic supports the positive effect because it protects innovation for copying, therefore making it worth for an entrepreneur to make the initial investment. However, the counterparts of IPR enforcement are even as strong. Other research shows for example that IPR enforcement is expensive to implement and difficult to maintain and therefore hinders the economy.
In this thesis we will investigate if the level of IPR enforcement has an effect on the level of innovation in a country. The relationship between IPR enforcement and innovation will be tested by analyzing a set of explanatory variables which contributes to the level of innovation. We use an IPR enforcement index made by Property Rights Alliance and this index will be set of against a proxy for innovation. This particular proxy is what distinguishes our research of earlier ones. The proxy is estimated with purchase price allocation data retrieved from PPAnalyser. Purchase price allocation data will provide us with a ratio of how total assets of a firm are allocated between intellectual property and tangible assets like buildings or land. This proxy has never been used before to estimate the level of innovation and is therefore interesting for analyzing the relationship between innovation and IPR enforcement.
5 The empirical research will be done by using the Ordinary Least Squares (OLS) estimator. Regression results with the OLS approach are consistent and efficient estimators of the parameters if the OLS assumptions are satisfied.
The outline of this thesis consists of the following parts. In chapter 2 a literature review is given about the advantages and disadvantages of IPR enforcement. Chapter 3 discusses the empirical methodology. The functioning of the Ordinary Least Squares model is explained thoroughly and the theory behind the assumptions and essential conditions is provided. Chapter 3 discusses further the used dataset in great detail. This section is quite relevant because of the use of the unique dataset of purchase price allocation. Chapter 4 will discuss the empirical findings. In chapter 5 some robustness checks are conducted and in chapter 6 a general conclusion is drawn based upon the empirical findings.
2.0 Literature Review
In this part of the thesis the literature behind IPR enforcement as an indicator of innovation is discussed. We start in section 2.1 with a short introduction about innovation and intellectual property rights themselves. In section 2.2 and 2.3 the advantages and disadvantages are provided. Section 2.4 gives attention to the differences between OECD and developing countries. In section 2.5 a conclusion is drawn based upon the literature discussed in the sections before.
2.1 Intellectual property rights and innovation
This sections provides the ideas and definitions behind intellectual property rights and innovation. Intellectual property rights are defined by the World Trade Organization as follows: ‘’Intellectual property rights are the rights given to persons over the creations of their minds. They
6 usually give the creator an exclusive right over the use of his/her creation for a certain period of time’’. (WTO, 2015)
When a person wants to register a new patent it relies completely on the legal system of the country where it tries to register the invention. In some countries the enforcement is strong and notion of patent infringement is taken really seriously. However, when the legal system in a country is weak and the power of the government is limited copying of products is difficult to prevent. In most western countries the level of IPR enforcement is high with Finland and Sweden leading the ranks. A more detailed explanation is given in the methodology chapter.
The definition of innovation is a much harder one to define. Several different of interpretations exist but the following is most accurate for our research. ‘’Innovation is a process that follows invention, being separated from invention in time. Invention is the creative act, while innovation is the first or early employment of an idea by one Organization or a set of organizations with similar goals’’ (Becker & Whisler, 1967).
2.2 The positive effect of Intellectual Property enforcement on innovation
In this section the literature is reviewed which explains the positive relationship between IPR enforcement and innovation. We start by why Intellectual property rights (IPR) give entrepreneurs an incentive to innovate. Second the effect of IPR on development and economic growth will be discussed. And as last a brief description of Romers economic growth model and why state-development related IPR protection can be an incentive for firms to innovate.
When an entrepreneur wants to exploits the possibilities of a new developed product of idea it has to make an initial investment to realize the research and development costs. When the product is developed, the entrepreneur sells the product and has to earn the initial investment in research and development costs back to be profitable. When his patent is not protected new entrants can enter the market and start producing the same product. However, they don’t suffer the high costs of R&D investments since they have already been made by the first mover. Intellectual property rights insure the first mover to be able to sell his product above marginal cost and thereby cover the initial R&D investment.
7 In the last decades developing countries are trying to carry out western Macro-economic policy to increase their welfare. Despite the promising facts that they dropped trade barriers, the old large budget deficits are gone and state subsidized companies barely exist anymore the expected increase of wealth hasn’t occurred yet (De Soto, 2002). According to De Soto institutions have to be integrated under one rule of law to create a sustainable environment for economic agents to trade goods and services and create new capital, hereby stimulating development and economic growth. In for example the United States a strong rule of law exists and this is in line with a survey of 100 U.S. manufacturing firms which shows that property right enforcement stimulates innovation (Mansfield, 1986).
The first economic growth model developed in 1990 describes that the level economic growth can be addressed to physical capital and technological knowledge (Romer, 1990). According to Romer long term economic growth is mostly due to internal endogenous investments in human capital and knowledge instead of external forces. Therefore IPR enforcement stimulates the possibilities and of developing and protecting human capital and knowledge.
Kanwar and Evenson (2003) investigated the determinants of innovation and found positive statistically significant results for IPR enforcement, credit provision, openness of trade and human capital and negative statistically significant results for political instability and real interest rates.
Another question is which level of IPR enforcement is most efficient. Acemoglu and Akcigit (2012) concluded in their paper that full IPR protection is not the best choice when looking at the effect on innovation. However, a state-dependent IPR protection might be most effective when implemented correctly. With a state-dependent IPR protection system inventions from companies which are far ahead of competition are granted a higher level of protection then ones where followers are close. The result of this so called ‘’trickle down’’ effect is that the R&D investments of all companies, even the ones without technological leadership increased. On top of that their research shows that with the implementation of state-dependent IPR protection economic growth increases on average from 1.86% to 2.04%.
8 The impact of IPR enforcement on economic growth is also investigated by Falvey, Foster and Greenaway (2006). In a panel of 79 countries they found a positively and significant relationship between economic growth rates and IPR enforcements for low and high income countries, though not for middle income countries. The reason behind these results are that in high income countries innovations are protected and for low income economies the technology flows towards them have positive spillover effects. However, in middle income economies the reduced income out of copying patented technology offsets the positive effect of IPR enforcements and results in a negative relationship. Interesting to notice is that also Gould & Gruben (1996) found a positive effect of IPR enforcement on economic growth and this effect is even slightly stronger for countries with relatively open economies.
2.3 The negative effect of Intellectual Property enforcement on innovation
In this section several disadvantages of IPR enforcement are discussed. The problems addresses higher cost of knowledge, monopoly power, patent thicket and more.
The most obvious argument against IPR enforcement is pointed out by Stiglitz (2008). In his paper he argues that a legal system to protect intellectual property is expensive to maintain and changes the allocation of resources towards maintaining the legal system instead of developing actual knowledge, which results in a decline of knowledge supply.
Another argument comes forward in the paper of McCalman (2001). He investigates the effect of the TRIPS agreement which came effective on 1 January 1995. In the TRIPS agreement participants agreed to a set of rules in order to provide extensive protection of intellectual property administered by the World Trade Organization. The results of McCalman’s research are twofold. The United States and some western countries seem to benefit most of the agreement and other countries are paying the price of signing the agreement. Especially Canada, Japan and developing countries are suffering of large income transfers into the United States and other western countries because of better protected IPR’s. The result is that a lower national income reduces the possibilities to invest in R&D investment and thereby reducing innovation.
9 Intellectual property rights give the entrepreneur the possibility to reap the fruits of his invention without disturbances of competitors. This can result in a monopoly where the monopolist charges a price where marginal revenue equals marginal costs and creating a deadweight loss for society. Besides the deadweight loss Geroski (1990) shows in his research that monopolies do not seem particularly innovative or progressive and instead small firms and new entrants play a role in stimulating innovation. A monopoly hinders new entrants and small firms and thus hinders indirect innovation as well.
In the same line of reasoning Boldrin and Levine (2008) argue that intellectual property rights should be eliminated to promote the free market concept. They illustrate in their paper that competition is essential for innovation and economic growth. Market competition results in the incentive to innovate as in most industries the first mover earns the highest profits while late entrants just earn enough to recoup their investment.
Nowadays, patent thicket is a quite commonly known problem in the discussion about IPR’s. Patent thicket occurs when an entrepreneur wants to develop a new technology and needs to obtain an overlapping set of licenses or patents. The patents or licenses are belonging to multiple parties making it time consuming and difficult not to infringe any patents. In particular in industries such as semiconductors, biotechnology, internet and computer software patent thicket is a considerable dilemma, in that bringing new innovative technologies on the market is their core business model (Jaffe, Lerner, & Stern, 2001).
2.4 Differences in intellectual property rights enforcement for developing countries.
The above argumentation about the level of IPR enforcement might be to general for developing countries. Some arguments are strong for developed countries but work counterproductive for third world countries. An explanation for this can be for example differences in size of the already existing R&D sector or if the investments are made by foreign MNE’s or domestic firms.
There are existing many different determinants of innovation and according to Léger (2006), we have to be cautious choosing which set of determinants is relevant for a selected country. The correct set of determinants mostly depends on the level of development of a
10 country, where IPR enforcement is of little significance for developing countries while of high significance for western countries.
In the discussion about differences between western countries and developing countries it comes forward that western countries are best in innovating, while the comparative advantage of developing countries are the low labor costs, as result strong IPR enforcements hurts the developing countries and on the other hand benefits western countries (Deardorff, 1992). More research shows almost equal results, with the small difference that developing countries with high productive R&D sectors benefit from international IPR protection. Nevertheless, there are not much developing countries with highly productive R&D sectors so the results are still roughly the same. (Chin & Grossman, 1990)
Helpman (1993) argues that strong IPR enforcement benefits developed countries because MNE’s are mostly located in rich western countries, they extent their market opportunities in developing countries and the rents on intellectual property flows back into western countries instead of the developing country. The counterpart of this argument is given by Falvey, Foster, & Greenaway (2006) who stated in their research that strong IPR enforcement results in higher foreign direct investments producing technology spillover which has a positive effect on the domestic economy. In line with this argument is that countries with a weak level of IPR enforcement aren’t attractive for MNE’s or foreign investors, thereby missing the knowledge transfers and technology spillover and thus solely rely on their own intellectual resources (Maskus, 2000).
2.5 Conclusion
As illustrated in the sections before the existing theories of IPR enforcement and the effect on innovation are widespread and remain ambiguous, even for developing countries the theories are bilateral and although it seems like developed countries benefit more of strong IPR enforcement then developing countries we can’t draw a general conclusion. As result of the large advantages of a sustainable and favorable investment environment for innovation, created by strong IPR enforcements we might forget the discussed counterarguments. The rise in cost of knowledge,
11 patent thicket and the possible existence of monopolies due to strong enforcement are problems which can’t be neglected and we therefore have to do more empirical research to obtain a better understanding on this topic. In the following part a new empirical research is conducted and this will hopefully lead to further insights.
3.0 Methodology
This chapter provides the empirical framework and methodology of the research. In the first subsection we discuss several variables which might influence the level of innovation. Next the origin of the variables and modifications to the data is explained in more detail. When we have specified our variables we set up the model to investigate the relationship between innovation and IPR enforcements. As last we find an econometric estimator for our empirical model and include the test of heteroscedasticity and conditions for an unbiased and consistent result.
3.1 Variables
The literature review showed that there’s no general consensus about the existence of a relationship between IPR enforcement and innovation, besides this the differences between OECD and non OECD countries are not clear yet and might be interesting to investigate. Therefore, we conducted a new research which will hopefully lead to more useful knowledge about IPR’s.
The relationship between IPR enforcement and innovation is investigated in earlier research by using R&D investment as proxy for the dependent variable innovation (Léger, 2006) or by using total granted patents per country as proxy (Kim, Lee, Park, & Choo, 2008). In this research however, we use the percentage of intellectual property of the total value of a company as proxy. The higher the ratio the more innovative the company.
The following list of variables all influence innovation to a certain level according to economic theory and earlier research. There might exist other factors, which are not included, but of possible importance for innovation. However, we use the data available to construct the
12 best possible model and some data isn’t suitable for us. The subsequent eight variables are publically accessible and have the correct timespan for our research, while other variables don’t have the correct time span or are in other ways not appropriate.
Table 3.1
Variable Explanation
Innovation The ratio of intellectual property to total assets per year and country.
IPRIndex Intellectual property rights enforcement index where the level of enforcement is rated with a rank of range 1 to 10. R&D expenditure Total research and development expenditure in
percentages of GDP
Openeconomy Sum of import and export of goods and services in percentages of GDP.
HumanCapital Gross secondary school enrollment ratio
FDI The net inflow of foreign direct investments as percentages of GDP.
GDP growth The annual GDP growth per capita
OECD Dummy variable which indicates of the country is an OECD member.
The proxy we use for measuring innovation in a country is a new created variable. We make use of the ratio of intangible assets (intellectual property) to total assets allocated to the worth of a company by an acquisition or a takeover. As already mentioned earlier a more detailed explanation about the origin is given in the next section. We argue that this ratio is a good proxy for innovation because it measures how much of the total worth of a company exists of intellectual property. This measurement gives a prediction in what kind of industry the host country of the companies is most active in. Some industries like the information technology
13 industry or telecommunications rely largely on innovations to be successful and therefore have a high intellectual property ratio.
The IPRI index is an index with a range from 1 to 10. A country with rank 1 has the lowest possible intellectual property enforcement, while 10 means perfect intellectual property enforcement. The index is made out of three components. The first is legal and political environment which measures the independence of the judges, the rule of law, political stability and corruption. The second pillar is the level of physical property rights and is based on property rights, registering property and ease of access to loans. The last pillar is intellectual property rights and exists of intellectual property protection, patent protection, and the level of copyright piracy.
R&D expenditure is one of the main sources of the development of innovation (Shefer & Frenkel, 2005). Firms invest in research and development because they see a profitable opportunity to earn a higher profit and thereby stimulating innovation.
The open economy variable consists of the sum of exports and imports of goods and services (% of GDP). According to Baldwin and Hanel (2003) there is a relationship between the openness of the economy and the level of innovation in a country. The general theory is the more open an economy is the higher the level of innovation. An open economy results in higher competition which is a trigger for firms to innovate in order to keep ahead of competition. There is however a counterpart about an open economy. Entrepreneurs are able to use more foreign intellectual property because of the high supply, which is due to the high competition in the open economy. This opens opportunities for patent infringement and therefore R&D expenditure could be lowered, because entrepreneurs need to invest more in protecting their patents at the expense of R&D investment (Aghion, Harris, Howitt, & Vickers, 2001).
Several theories argue that human capital is essential for the development of innovation. Marvel and Lumpkin (2007) found in their research that radical innovations transform and create new markets and stimulate economic growth. More findings in their paper are that general capital although mostly human capital is of vital importance for entrepreneurs to create radical innovations. These radical innovations are on top of that positively related to formal education and prior knowledge of technology. However, when the entrepreneur has prior knowledge of how to serve the market this might work counterproductive for the reason that the entrepreneur
14 is not capable of seeing new possibilities and therefore can’t create radical innovations anymore. The proxy we choose for human capital is the gross secondary school enrolment ratio and we think that it is a good substitute for the formal education and knowledge of technology as described by Marvel and Lumpkin.
Foreign direct investment (FDI) is an important channel for the mediation of knowledge spillovers such as technological information or trade secrets. Branstetter (2006) examined in his empirical work what drives knowledge spillovers and found that international trade has a positive relationship. In his research he measures the knowledge spillovers of Japanese firms in the United States and found evidence that foreign direct investment has a positive effect on knowledge spillovers from Japanese firms into the United States and vice versa.
Annual gross domestic product growth per capita and innovation are correlated with each other and innovations in a country foster economic growth in several ways. Often innovations are made by small startup companies with a high growth potential. In research is found that only high growth rates are possible for relatively small firms and thus the level economic growth is explained by the growth potential of small innovative firms (Wong, Autio, & Ho, 2005).
As last an OECD dummy is added. The level of innovation differs a lot worldwide and the best way to divide it in two parts is using the OECD membership as requirement. The OECD is an international organization focused on promoting policies which tackle the challenges of the globalized economy. The most members of the OECD organization are countries who are far ahead in social and economic development and have a well-organized legal system.
3.2 Origin of the variables and modification of the data panel
In this section the variables are further analyzed to provide a transparent as possible research method. We start with the origin of the variables, then we discuss the adaption process of the data, the number of observations, countries involved and the time span covered by the data.
15 Table 3.2
Variable Source
Innovation Purchase price allocation, retrieved from www.PPAnalyser.com
IPRIndex Property Rights Alliance Data, retrieved from http://internationalpropertyrightsindex.org/ R&D expenditure World Bank national accounts data, and OECD
National Accounts data files.
Openeconomy World Bank national accounts data, and OECD National Accounts data files.
HumanCapital World Bank national accounts data, and OECD National Accounts data files.
FDI World Bank national accounts data, and OECD National Accounts data files.
GDP growth World Bank national accounts data, and OECD National Accounts data files.
OECD http://www.oecd.org/
The dependent variable and proxy for innovation is the ratio of intellectual property to total assets. The ratio is retrieved from analyzing worldwide purchase price allocation data. When a firms completes a merger or acquisition a purchase price allocation (PPA) is required in the United States by the Generally Accepted Accounting Principles (USGAAP) and outside the United States by the International Financial Reporting Standards (IFRS). According to those accounting standards the company has to disclose an allocation of the purchase price to various assets and liabilities. This allocation has to happen for USGAAP in accordance with the Financial Accounting Standards Board's (FASB) Statement of Financial Accounting Standards No. 141 Business Combinations (SFAS 141r) and SFAS 142 Goodwill and Other Intangible assets (SFAS 142) and for IFRS in according to the guidelines of IFRS 3 Business Combinations.
16 In this research we use the database of PPAnalyser which covers worldwide almost 5800 records of purchase price allocation data. Out of these records we made a selection of 30 countries with the most observations and a timespan from 2006 till 2014. The reason for the limited timespan is that the coverage of PPAnalyser before 2006 is limited with only for the United States, China and the United Kingdom enough records. For a list of included countries we refer to table 2 in the appendix. In table 2 the total amount of purchase price allocation (PPA) is given and the United States is leading the ranks with 3312 PPA’s, in comparison with Austria who only has 8. With the purchase price allocation data we construct the intellectual property ratio. We take the sum of goodwill and intangible assets and divide this by the total sum of goodwill, intangible assets and tangible assets for every record. From all these ratios per year per country we use the average ratio to derive our data input for the regression model. In the appendix several tables are attached which offer some more information about the purchase price allocation data. In the second table the total observations per country and the average intellectual property ratio per country from 2006 to 2014 is illustrated. The third table gives the observations per country per year. The third table in the appendix already shows a problem for several years, since the sample of observation Is too small or sometimes there isn’t even one purchase price allocation during the year for a specific country. Austria and Romania suffer the most of missing PPA’s, seeing that 5 years of the timespan are not disclosed for them. To solve the problem of missing data we use interpolation to create new data points. Linear interpolation is a simple mathematical method to insert new data points linear within two known discrete data points.
The International Property rights index is first introduced in 2007 by Hernando de Soto, who believes property rights are one of the most important factors of prosperity and freedom of economy. Every year the Property Rights Alliance publishes an executive summary of their findings of last year. We collected all the data from the reports 2007 till 2014 and paired them with the 30 countries in our dataset. In the appendix table 1 is disclosed and it illustrates all countries and their average IPR index over 2007 till 2014. There was no need to interpolate any of the data between 2007 and 2014 because all the countries where covered in the IPR studies for all years.
17 The data on R&D expenditure, the exports and imports of goods and services (% of GDP), human capital (gross secondary school enrollment), foreign direct investment and GDP growth per capita are all retrieved from the World Bank and cover the period 2006 till 2014. Some of the data however, is incomplete for years between 2006 and 2014 and therefore we use linear interpolation to cover the data points in between.
The OECD country dummy is constructed by allocating a 0 value to non OECD countries and a 1 to members. The information is retrieved from http://www.oecd.org/. In the appendix table 1 is disclosed which indicates if a sample country is OECD member or not.
In the appendix a summary of the variables is included in table 5. The intellectual property ratio has 205 observations and a mean ratio of 0.6. R&D expenditure has with 168 the lowest amount of observations, while foreign direct investment and openness of economy have the highest amount of observations. All observations are within in acceptable range of the mean and no large outliers are detected.
3.3 Econometric model
In this subsection the econometric model is constructed by which we hope to find if IPR enforcement has a relationship with innovation. We have evaluated all the possible variables which might influence innovation according to theory and earlier done research. For these variables we made adjustments and modifications necessary for a useful econometric data panel. For our research we have conducted to following econometric model.
𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛 = 𝛽1 𝐼𝑃𝑅𝐼𝑛𝑑𝑒𝑥 + 𝛽2 𝑅&𝐷𝑒𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒 + 𝛽3 𝑜𝑝𝑒𝑛𝑒𝑐𝑜𝑛𝑜𝑚𝑦 +
𝐵4 𝐻𝑢𝑚𝑎𝑛𝐶𝑎𝑝𝑖𝑡𝑎𝑙 + 𝛽5 𝐹𝐷𝐼 + 𝛽6 𝐺𝐷𝑃𝑔𝑟𝑜𝑤𝑡ℎ + 𝛽7 𝑂𝐸𝐶𝐷 + 𝛽8 𝑌𝑒𝑎𝑟 + 𝜀
With this equation we hope to find a relationship between the IPR index and innovation. The other variables are control and explanatory variables which are necessary to avoid biased coefficients.
18 The Year dummy is added to the equation to control for any unobserved year effect such as the financial crisis of 2007 and 2008. Including the year dummy will not affect the consistency of our model but might reduce efficiency.
3.4 Econometric method
In this section we justify and explain the use of the econometric model we have chosen to investigate the relationship between innovation and intellectual property right enforcement. We find that the most efficient method is the general Ordinary Least Squares (OLS) regression. The OLS assumptions are discussed below, we conclude that they are not violated and thus the OLS estimator is unbiased and consistent.
The OLS estimation results are consistent when all regressors are exogenous and is known as the 1e OLS assumption. The second assumption holds when the dependent and independent variables are independently and identically distributed. The third assumption is that large outliers are unlikely. In our research we construct a multiple regression model and therefore the additional assumption of no perfect multicollinearity has to hold as well (Stock & Watson, 2003). To prove that all regressors are exogenous we have to show that the independent variables have an effect on the model but those variables themselves aren’t affected by the model. We use the Two Stage Least Squares estimator to investigate if the regression suffers from endogeneity. Another name for Two Stage Least Squares is the instrumental variable (IV) method and allows consistent estimation results when the models suffers from reverse causality between the dependent variable and the independent variables. The IV method uses an instrument to remove the reverse causality and only shows the real effect of the endogenous variable on the dependent variable. An instrument is a variable which isn’t covered in the original model but correlated with the endogenous variable and has to satisfy two conditions. The first is the instrument has to be correlated with the endogenous variable, conditional on the other exogenous explanatory variables. The second condition holds when the instrument is not correlated with the error term of the original model.
19 In our investigation we use the lagged versions of the possible endogenous variables as instruments. We find that all instruments are relevant, valid and do not correlate with the error term of the regression model. The lagged versions are strong instruments and all tests result in statistics far above the rule of thumb of 10. With these instruments we test the model for endogeneity using the post estimation commands of Stata12 after running the instrumental variable regression. The null hypothesis holds when the explanatory variables are exogenous. The result of the Durbin-Watson and Wu-Hausman test statistics are for all explanatory variables not significant and we therefore don’t reject the null hypothesis. We conclude that the regression model doesn’t suffer from endogeneity. The test results are disclosed in the appendix table 4.1 till 4.6. Besides that a correlation matrix is constructed which shows that the residuals of the original model are not correlated with the used instruments. The correlation matrixes can be found in the appendix table 5.1 till 5.6.
We assume the used dataset collected is identically and independently distributed. Each random variable is mutually independent and all have the same probability distribution. Therefore the second OLS condition is satisfied.
The third OLS condition holds when large outliers are unlikely to occur. As we see in table 3.3 on the previous page none of the variables have large minima or maxima and therefore we conclude that large outliers don’t occur in the dataset.
The last assumption requires no perfect or no nearly perfect multicollinearity. A high degree of multicollinearity results in less accurate estimates of parameters and therefore should be avoided when constructing a strong model. We used a VIF test in Stata to test the independent variables on multicollinearity. For every VIF score under 4 we don’t have to worry for multicollinearity at all. We provided the VIF outputs in the table below and only for the IPR Index a value of 4.11 is found which might be problematic. However, we assume this won’t result in any problems because all other values are far beneath 4 with an average of 2.25. Therefore, a small value of 0.11 point above 4 will still produce an accurate and efficient model without much multicollinearity. Besides the VIF table we provide a correlation matrix of the coefficients of the regression model illustrated in table 7 of the appendix. Common theory argues that correlation between coefficients above 0.8 is definitely problematic and we should start worrying with a
20 correlation above 0.5. A look at the results of the correlation matrix shows only for the correlation between the IPR index and R&D expenditure a correlation of -0.542, the other correlations are far below 0.5. This high correlation might expose our model to less accurate estimates, nonetheless we argue that it will cause only small trouble, because of the small overvalue of 0.042 and therefore conclude that the econometric model doesn’t suffer from multicollinearity. Table 3.3
Variable VIF 1/VIF
R&D expenditure 2 0.50
IPR index 4.11 0.24
Openness of Economy 1.85 0.54
Secondary School enrollment 2.12 0.47 Foreign Direct Investment 1.75 0.57
GDP growth per capita 3.06 0.33
OECD 2.69 0.37
Mean VIF 2.25
As last we test if the econometric model has constant variances or that we have to correct our estimates for heteroscedasticity. This test is conducted using the Breusch-Pagan / Cook-Weisberg test for heteroscedasticity. The null hypothesis is the model has constant variances. The table below shows the test results for the heteroscedasticity test. A p-value of 0.0554 is higher than the standard significance level of 0.05 and we are thus allowed to use homoscedastic variances.
Table 3.4
Breusch-Pagan / Cook-Weisberg test for heteroscedasticity
Variables
R&D expenditure IPR index
Openness of Economy Secondary School Enrollment Foreign Direct Investment GDP growth per capita OECD
chi2(7) 13.78
21
4.0 Results
In this section we discuss the results of our empirical regression. The regression is done on three different country samples. We start with providing the results of the main regression covering all 30 countries of the sample, then we discuss the findings of the OECD country sample and as last we take a look at the regression results for non-OECD countries.
Ordinary Least Squares
Innovation
Independent Variable All countries OECD Non OECD
R&D expenditure 0.0489*** 0.0565*** -0.1092 (0.018) (0.019) (0.09) IPR index 0.0089 -0.0094 0.1063* (0.02) (0.021) (0.06) Openness of Economy 0.0020*** 0.0018*** 0.0043** (0.001) (0.001) (0.002) Secondary School Enrollment -0.0029* -0.0023* -0.0140* (0.002) (0.002) (0.007)
GDP growth per capita -0.0073 -0.0171* -0.0111
(0.006) (0.009) (0.008)
Foreign Direct Investment -0.0010 -0.0004 0.0971***
(0.002) (0.002) (0.024) OECD -0.0728 (0.058) Constant 0.6567*** 0.6364*** 0.8557 (0.146) (0.170) (0.693) Number of observations 144 117 27 R- squared 0.228 0.236 0.799
Note: figures between parentheses are t-values: *** = significant 1% level, ** = significant 5% level and * = significant 10% level
22 We use the empirical model explained in subsection 3.3 to investigate the relationship between innovation and Intellectual Property Rights enforcement. The econometric method we use for our research is the Ordinary Least Squares estimator. All the required assumptions for the OLS estimator are satisfied and therefore we are able to estimate consistent and efficient coefficients.
The results for our three country samples are presented in the table on the previous page. The independent variables on the left side are for each regression the same with the exemption of the OECD dummy for the OECD country sample and the non-OECD sample.
For the general model with all countries included we find some interesting results. R&D expenditure is highly significant and positively influences the level of innovation. Shefer and Frenkel (2005) predicted this result already since innovative projects are having a relative high profits. However, the IPR enforcement index is not significant at all and implies that worldwide the level of IPR enforcement is not a main source of innovation. It might have a small effect in a particular country, but overall other factors such as human capital or knowledge spillovers are more important for determining the level of innovation. These findings suggest that the negative effects of IPR enforcement are quite strong and this is in line with Stiglitz (2008) who argued that IPR enforcement increases the cost of knowledge and thereby decreasing the level of innovation. This result is in strong contrast with research done by Kanwar and Evenson (2003) who found that IPR enforcement positively influences innovation and technological change.
Openness of economy is highly significant and shows that countries which are more involved in international trade (% of GDP) have a higher inflow of technology and knowledge. With a higher inflow of technology and knowledge a higher level of knowledge spillovers is reached which positively effects innovation. Earlier research about openness of the economy supports this positive relationship because of a higher level of competition between firms to stay market leader which fosters innovation (Baldwin & Hanel, 2003). The positive significant result is however, in contradiction with the findings of Aghion et al. (2001) who argue that a more open economy results in higher costs of protecting intellectual property and therefore less money is available for innovation and R&D expenditure. Further we find that human capital has a significant and positive effect on innovation. The used proxy secondary school enrollment is significant at
23 10% level and has a negative coefficient, although a negative relationship seems strange because all kind of theories indicate a positive relationship (Marvel & Lumpkin, 2007).
GDP growth per capita doesn’t influence the level of innovation according to our empirical model. These findings are not in line with the results of Wong, Autio, & Ho (2005) who argue that there is a positive relationship between GDP growth per capita and innovation. We find no relationship between Foreign Direct Investment and innovation for the general model. This result stands in contrast with the research of Bransteter (2006) who finds that that a higher level of FDI results in knowledge spillovers and thereby positively effects the level of innovation.
The OECD country sample regression shows similar results as the general regression. The reason could be that of the 144 observations 117 observations where noticed for OECD countries. The IPR enforcement index is not significant and R&D expenditure openness of economy are still highly significant. Nonetheless, there is a small difference with the general regression because GDP growth per capita is negatively significant at the 10%. This result implies that a high GDP growth per capita results in a lower level of innovation. This is notwithstanding with the literature and seems like an odd result (Wong, Autio, & Ho, 2005). An explanation for this negative coefficient could be that western countries have relative low GDP growth rates per capita but the level of innovation is from origin higher than in developing countries and therefore a negative relationship might be correct.
For the Non-OECD country sample we find completely different results. Where R&D expenditure is highly significant for the OECD sample and general model it is of insignificance for non-OECD countries. For the non-OECD sample we find a somewhat surprising result because the IPR enforcement index is here significant at the 10% level. The positive coefficient for the IPR index implies that for developing countries a higher level of IPR enforcement positively effects the level of innovation. A possible reason for this result is foreign firms are more willingly to invest in developing countries because their intellectual property is well protected against intellectual piracy of small domestic firms. The openness of Economy is significant at 5% level and indicates that developing countries with a relative open economy benefit of knowledge spillovers from MNE’s. This result is in line with the positive and significant coefficient of Foreign Direct Investment. Theory already indicates that foreign direct investments create knowledge spillovers
24 in developing countries and thereby positively effecting the level of innovation (Bransteter, 2006). Secondary school enrollment is negatively correlated with innovation but only has a significance level of 10. However, this is the same for the other regressions and the negative relationship is not explained by any theory in the literature review.
5.0 Robustness checks on empirical model
In this section several robustness checks for the empirical model are discussed. Robustness checks are necessary to test if the found results won’t change when small modifications are made to the model. The modifications we do are the replacement of explanatory variables by two new variables and by eliminating some countries out of the data sample. We find that most of the results for the new data samples are robust for changes in the dataset or model specifications. The regression outputs of the robustness checks are attached in the Appendix and illustrated in table 9 till 14.
To conduct the first robustness check we replace the variables GDP growth per Capita and Foreign Direct Investment for Annual Population growth and Unemployment. Both variables are retrieved from the World Bank and cover the period 2006 till 2014. A summary of the new variables is found in table 8 of the appendix.
The results for the main regression, with all countries included, are roughly the same as with the old explanatory variables. R&D expenditure, Openness of Economy and Secondary School Enrollment are significant and the IRP index still of insignificance.
The OECD country sample is like the main regression robust for changes in the explanatory variables, with only Secondary School Enrollment that changes to a p-value of just above 10%.
For the Non-OECD sample the results are somewhat different. The IPR index is now highly significant instead of a p-value just below 0.1. The R&D expenditure is significant while in the old model it was of insignificance and the openness of trade changed as well. We conclude
25 that for the Non-OECD sample robust results are hard to find for all variables and only the IPR index seems robust. A possible explanation of this robustness problem is the low amount of observations which result in high standards errors of the coefficients.
The second robustness check we do is eliminating one or two countries randomly out the data sample. The results should be roughly the same when a country is excluded otherwise we haven’t found robust results. The countries excluded from the dataset for the general model are Poland and Finland, for the OECD country sample France and Norway. For the Non-OECD country sample we excluded Colombia out of the sample.
The results for the general OECD sample are not of considerable differences and all old significant results are still significant. Only Secondary School enrollment and the old insignificant coefficients of the model are subject to change.
The Non-OECD sample suffers from the same problem as with the first robustness check. Due to the small amount of observations the elimination of Colombia results in a completely different data distribution and therefore results are not robust.
6.0 Conclusion
This thesis analyzed the relationship between Intellectual Property Rights enforcement and the level of innovation in a country. The literature review gives a short look into to ambiguity about this topic and a general consensus is far from reached. With the unique proxy for innovation this research will provide new information and hopefully will lead to a better understanding of IPR enforcement and the effect on innovation.
In general we find that IPR enforcement is not correlated with innovation. The results for the all country sample and OECD sample showed an insignificant coefficient and other factors then IPR enforcement seem more important in influencing the level of innovation. The results are robust for model modifications and we therefore conclude that IPR enforcement is not a source of
26 innovation. A reason for this result could be the high costs of enforcement or the hassle and time consuming problems around acquiring patent rights, which slows down innovation process.
However, for the Non-OECD country sample we found different results. The IPR enforcement index is now significant and robust for changes in model specifications. This implies that foreign firms are more attracted by well-protected intellectual property rights in developing countries since it reduces the risk of intellectual piracy and copying of products. The problem with the Non-OECD country sample is that the number of observations is too low to get a strong robust model. Although the IPR index is robust all the other explanatory variables are not and they vary a lot when tested for robustness.
The found results are again ambiguous because of the differences found per dataset. In general we conclude that there’s no relationship between IPR enforcement nonetheless for developing countries it is different. A reason for this difference might be the small amount of observations and the limited time span of 2007 till 2014. It is not yet possible to extend the dataset because of scarce data before 2007. To improve this research it is useful to find more purchase price allocation data for small countries since a lot of years between 2007 and 2014 are still missing and to extent the time span of the whole dataset. This will make interpolation unnecessary and increase the accuracy of the proxy.
27
7.0 References
Acemoglu, D., & Akcigit, U. (2012). Intellectual property rights policy, competition and innovation. Journal of the European Economic Association, 10(1), 1-42.
Aghion, P., Harris, C., Howitt, P., & Vickers, J. (2001). Competition, Imitation and Growth with Step-by-Step Innovation. The Review of Economic Studies, 467-492.
Baldwin, J. R., & Hanel, P. (2003). Innovation and Knowledge Creation in an Open Economy. Cambridge: Cambridge University Press.
Becker, S. W., & Whisler, T. L. (1967). The innovative organization: a selective review of current theory and research. The Journal of Business, 40(4), 462-469.
Boldrin, M., & Levine, D. K. (2008). Against Intellectual Monopoly. Cambridge: Cambridge University Press.
Bontis, N. (1998). Intellectual capital: an exploratory study that develops measures and models. Management Decision, 36(2), 63-76.
Bransteter, L. (2006). Is foreign direct investment a channel of knowledge spillovers? Evidence from Japan's FDI in the United States. Jourmal of International Economics, 68(2), 325-344.
Chin , J. C., & Grossman, G. M. (1990). The Political Economy of International Trade: Essays in Honor of Robert E. Baldwin. (R. W. Jones, & R. O. Krueger, Eds.) Oxford: Blackwell. De Soto, H. (2002). Law and Property Outside the West: A Few New Ideas About Fighting
Poverty. Optima Special Issue on Sustainable Development, 48(1), 2-9.
Deardorff, A. V. (1992). Welfare effects of global patent protection. Economica, 59(223), 35-51. Falvey, R., Foster, N., & Greenaway, D. (2006). Intellectual Property Rights and Economic
Growth. Review of Development Economics, 10(4), 700-719.
Geroski, P. (1990). Innovation, Technological Opportunity, and Market Structure. Oxford Economic Papers, New Series, 42(3), 586-602.
Gould, D. M., & William, C. G. (1996). The role of intellectual property rights in. Journal of Development Economics, 48, 323-350.
Helpman, E. (1993). Innovation, Imitation, and Intellectual Property Rights. Econometrica, 61(6), 1247-1280.
Jaffe, A. B., Lerner, J., & Stern, S. (2001). Innovation Policy and the Economy. Cambridge: MIT Press. Retrieved from http://www.nber.org/books/jaff01-1
28 Kanwar, S., & Evenson, R. (2003). Does Intellectual Property Protection Spur Technological
Change? Oxford Economic Papers, 55(2), 235-264.
Kim, Y. K., Lee, K., Park, W. G., & Choo, K. (2008). Appropriate intellectual property protection and economic growth in countries. Research Policy, 41(2), 358–375.
Léger, A. (2006). Intellectual property rights and innovation in developing countries: evidence from panel data. German Institute for Economic Research.
Mansfield, E. (1986). Patents and Innovation: An Empirical Study. Management Science, 173 - 181.
Marvel, M. R., & Lumpkin, G. T. (2007). Technology Entrepreneurs Human Capital and Its Effects on Innovation Radicalness. Entrepreneurship Theory and Practice, 807-829.
Maskus, K. E. (2000). IP rights and economic development. Case Western Reserve Journal of International law, 32, 471-506.
McCalman, P. (2001). Reaping what you sow: an empirical analysis of. Journal of International Economics, 55, 161-186.
Romer, P. (1990). Endogenous Technological Change. The Journal of Political Economy, 98(5). Schumpter, J. (1912). Theorie der Wirtschaftlichen Entwicklung (The Theory of Economic
Development). Cambridge: Harvard U. Press.
Shefer, D., & Frenkel, A. (2005). R&D, firm size and innovation: an empirical analysis. Technovation, 25, 25-32.
Stiglitz, J. E. (2008). Economic Foundations of Intellectual Property Rights. Duke Law Journal, 57, 1693-1724.
Stock, J., & Watson, M. W. (2003). Introduction to Econometrics. New York: Prentice Hall. Wong, P. K., Autio, E., & Ho, Y. P. (2005). Entrepreneurship, Innovation and Economic Growth:
Evidence from GEM data. Small Business Economics, 24(3), 335-350.
WTO. (2015, 12 26). https://www.wto.org/english/tratop_e/trips_e/intel1_e.htm. Retrieved from World Trade Organization.
29
8.0 Appendix
Table 1
Countries
OECD
member Average IPR index
Argentina 4.4 Australia X 8.0 Austria X 7.9 Belgium X 7.4 Brazil 5.1 Canada X 7.9 Chile X 6.6 China 5.2 Colombia 5.0 Denmark X 8.2 Finland X 8.5 France X 7.2 Germany X 8.0 India 5.6 Ireland X 7.7 Israel X 6.4 Italy X 6.0 Japan X 7.7 Mexico X 4.9 Netherlands X 8.2 Norway X 8.3 Poland 5.6 Romania 5.0 Russian Federation 4.3 Singapore 8.1 Spain X 6.6 Sweden X 8.4 Switzerland X 8.2 United Kingdom X 7.9 United States X 7.6
30 Table 2
Countries amount of PPAs Region Average allocation to IP
United States 3312 US 67%
Canada 380 Canada 52%
United Kingdom 256 Europe 67%
Germany 131 Europe 61% China 121 Other 64% Australia 84 Other 62% Netherlands 69 Europe 58% France 62 Europe 57% Brazil 54 Other 54% Israel 46 Other 78% Sweden 42 Europe 72% Italy 37 Europe 54% Spain 32 Europe 51% Switzerland 32 Europe 67% Mexico 29 Other 46% Poland 27 Europe 54% India 26 Other 64% Japan 22 Other 42% Norway 22 Other 54% Ireland 21 Europe 72% Denmark 19 Europe 59% Singapore 19 Other 62% Belgium 17 Europe 67%
Russian Federation 17 Other 62%
Colombia 16 Other 48% Argentina 14 Other 29% Chile 14 Other 31% Finland 12 Europe 70% Romania 10 Other 69% Austria 8 Europe 61% Total 4951
31 Table 3
Purchase price allocations per year and country
Countries 2006 2007 2008 2009 2010 2011 2012 2013 2014 Argentina 4 1 3 - 1 4 1 - -Australia 1 7 9 9 17 17 11 10 3 Austria 2 - 2 - - 3 1 - -Belgium - 2 - 2 7 2 2 - 2 Brazil 2 5 7 6 10 9 10 4 1 Canada 11 28 32 37 65 77 71 40 19 Chile - 1 3 3 2 - 3 - 1 China 18 25 30 9 12 7 12 4 4 Colombia 2 2 6 1 1 3 1 - 1 Denmark - 1 1 2 5 4 3 2 1 Finland 1 1 1 2 3 4 - - -France 3 2 13 8 5 15 8 4 4 Germany 5 14 7 17 19 20 20 23 6 India - 5 1 6 4 3 4 1 1 Ireland 3 - - 4 1 5 6 2 -Israel 4 3 4 1 2 10 13 7 2 Italy - 2 2 4 5 9 6 6 2 Japan 1 4 6 1 2 4 2 2 -Mexico 2 6 2 - 3 2 4 9 1 Netherlands 3 10 11 7 3 11 12 6 6 Norway - - 3 3 4 7 1 3 1 Poland 1 1 2 1 6 7 8 1 -Romania - 4 2 - - 3 1 - -Russian Federation 3 7 3 1 - 2 1 - -Singapore - 2 2 3 2 3 2 - -Spain 1 4 6 3 3 3 9 1 2 Sweden 5 5 5 1 8 7 6 3 1 Switzerland 2 3 2 4 4 2 7 4 4 United Kingdom 11 19 46 28 33 43 39 26 11 United States 211 508 528 271 302 396 543 352 201
32 Table 4.1
R&D expenditure
Tests of endogeneity
Ho: variables are exogenous Test statistic P-value
Durbin (score) chi2(1) 0.901 0.343
Wu-Hausman F(1,105) 0.801 0.373
Table 5.1
Correlationmatrix
Residuals of
regression L1.R&D expenditure Residuals of regression 1 L1.R&D expenditure -0.0195 1 Table 4.2 IPR index Tests of endogeneity
Ho: variables are exogenous Test statistic P-value
Durbin (score) chi2(1) 0.264 0.607
Wu-Hausman F(1,106) 0.234 0.630
Table 5.2
Correlationmatrix
Residuals of
regression L1.IPR Index Residuals of
regression 1
33 Table 4.3
Openness of Economy
Tests of endogeneity
Ho: variables are exogenous
Durbin (score) chi2(1) 0.057 0.812 Wu-Hausman F(1,106) 0.050 0.823 Table 5.3 Correlationmatrix Residuals of regression L1.Opennes of Economy Residuals of regression 1 L1.Opennes of Economy 0.0083 1 Table 4.4
Secondary School Enrollment
Tests of endogeneity
Ho: variables are exogenous Test statistic P-value
Durbin (score) chi2(1) 1.152 0.283
Wu-Hausman F(1,106) 1.027 0.313 Table 5.4 Correlationmatrix Residuals of regression L1.Secondary School enrollment Residuals of regression 1 L1.Secondary School enrollment 0.0093 1
34 Table 4.5
Foreign Direct Investment
Tests of endogeneity
Ho: variables are exogenous Test statistic P-value
Durbin (score) chi2(1) 0.003 0.954 Wu-Hausman F(1,106) 0.003 0.957 Table 5.5 Correlationmatrix Residuals of regression L.1 Foreign Direct Investment Residuals of regression 1 L.1 Foreign Direct Investment 0.0083 1 Table 4.6
GDP growth per capita
Tests of endogeneity
Ho: variables are exogenous Test statistic P-value
Durbin (score) chi2(1) 0.001 0.971
Wu-Hausman F(1,106) 0.001 0.973 Table 5.6 Correlationmatrix Residuals of regression L1.GDP growth per capita Residuals of regression 1 L1.GDP growth per capita 0.0083 1
35
Table 6
Variable
Obs Mean
Std.
Dev. Min Max
Independent Variables
R&D expenditure 168 1.9 1.1 0.2 4.5
IPR index 239 6.9 1.4 3.2 8.7
Secondary School enrollment 188 102.4 15.7 58.7 165.6 Foreign Direct Investment 234 5.2 9.3 -9.2 88.1
Openness of Economy 234 82.4 66.3 22.0 439.7
GDP growth per capita 232 1.3 3.4 -8.7 13.6
OECD 240 0.7 0.4 0.0 1.0
Dependent Variable
36 Table 7
Correlation matrix of coefficients of regress model
e(V) R&D expenditure IPR index Openness of Economy Secondary School enrollment GDP growth per capita Foreign Direct Investment OECD R&D expenditure 1.000 IPR index -0.542 1.000 Openness of Economy 0.028 -0.111 1.000 Secondary School enrollment 0.057
-0.336 -0.106 1.000
GDP growth per capita -0.186 0.121 0.033 0.245 1.000 Foreign Direct Investment 0.178
-0.072 -0.537 -0.154 -0.139 1.000
OECD -0.066
37
Robustness Checks
In tabel 9-11 the regression output of different explanatory variables: Population Growth and Unemployment.
Table 8 variable Summary
Variable
Obs Mean Std. Dev. Min Max
Unemployment 239 6.948368 3.339241 1.95 26.6
Annual Population growth 240 0.786958 0.771272 -1.69 5.32
Table 9 General model with all countries
_cons .5429519 .1152885 4.71 0.000 .3149315 .7709724 2013 .1796919 .1226675 1.46 0.145 -.0629231 .4223068 2012 .0180559 .0548675 0.33 0.743 -.0904624 .1265743 2011 -.0275405 .0479443 -0.57 0.567 -.1223661 .067285 2010 -.0044717 .0473827 -0.09 0.925 -.0981865 .0892431 2009 .0058344 .0471963 0.12 0.902 -.0875116 .0991804 2008 .0139809 .0462761 0.30 0.763 -.0775452 .105507 year oecd -.0422308 .0528188 -0.80 0.425 -.1466971 .0622356 unemployment .0081946 .0051967 1.58 0.117 -.0020836 .0184729 populationgrowth -.0051804 .0200535 -0.26 0.797 -.0448426 .0344819 secondaryschoolenrollment -.0031218 .0014016 -2.23 0.028 -.005894 -.0003497 open .0018446 .0004326 4.26 0.000 .0009889 .0027002 iprirank .0142803 .0187279 0.76 0.447 -.0227604 .0513209 rdinvestmentgdp .0514721 .0173265 2.97 0.004 .0172034 .0857408 ratio Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 4.62200201 147 .031442191 Root MSE = .163
Adj R-squared = 0.1550 Residual 3.56016588 134 .026568402 R-squared = 0.2297 Model 1.06183613 13 .081679703 Prob > F = 0.0005 F( 13, 134) = 3.07 Source SS df MS Number of obs = 148
38
Table 10 OECD country sample
Table 11 Non-OECD Country sample
_cons .4445263 .1593725 2.79 0.006 .1284846 .7605679 2013 .1991825 .1235338 1.61 0.110 -.0457897 .4441547 2012 .0347496 .0592571 0.59 0.559 -.0827595 .1522587 2011 .0103653 .0551395 0.19 0.851 -.0989784 .119709 2010 .0211448 .0541364 0.39 0.697 -.0862097 .1284993 2009 .0406564 .0538497 0.75 0.452 -.0661295 .1474424 2008 .051265 .0524862 0.98 0.331 -.0528172 .1553472 year unemployment .0094562 .0056961 1.66 0.100 -.0018394 .0207519 populationgrowth .020616 .0252248 0.82 0.416 -.0294056 .0706377 secondaryschoolenrollment -.0026843 .0016284 -1.65 0.102 -.0059135 .0005449 open .0016697 .0004434 3.77 0.000 .0007905 .002549 iprirank .009541 .0220598 0.43 0.666 -.0342044 .0532865 rdinvestmentgdp .0531535 .0184173 2.89 0.005 .0166313 .0896757 ratio Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 3.56829229 116 .03076114 Root MSE = .16245
Adj R-squared = 0.1421 Residual 2.74467519 104 .026391108 R-squared = 0.2308 Model .823617094 12 .068634758 Prob > F = 0.0046 F( 12, 104) = 2.60 Source SS df MS Number of obs = 117
_cons .615284 .6157872 1.00 0.330 -.6735734 1.904141 2012 -.0272898 .2166516 -0.13 0.901 -.4807468 .4261673 2011 -.1771776 .1298446 -1.36 0.188 -.4489455 .0945903 2010 -.1593238 .1075467 -1.48 0.155 -.3844216 .0657741 2009 -.24334 .0930445 -2.62 0.017 -.4380845 -.0485955 2008 -.2630979 .0933876 -2.82 0.011 -.4585605 -.0676354 year unemployment .0927608 .0312753 2.97 0.008 .0273008 .1582208 populationgrowth -.2680234 .0989667 -2.71 0.014 -.475163 -.0608837 secondaryschoolenrollment -.0195569 .0077762 -2.51 0.021 -.0358327 -.0032812 open -.0018537 .0046731 -0.40 0.696 -.0116347 .0079273 iprirank .2388462 .0696542 3.43 0.003 .0930582 .3846341 rdinvestmentgdp .2106903 .0807427 2.61 0.017 .041694 .3796866 ratio Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .970967767 30 .032365592 Root MSE = .11936
Adj R-squared = 0.5598 Residual .27068958 19 .01424682 R-squared = 0.7212 Model .700278187 11 .063661653 Prob > F = 0.0022 F( 11, 19) = 4.47 Source SS df MS Number of obs = 31 > ent I.year
39
Removed countries from sample
In table 12-14 the regression output of the model with excluded countries are given.
Table 12 General model: deleted Poland and Finland
_cons .6756087 .1478221 4.57 0.000 .383004 .9682134 2013 .1715758 .1256662 1.37 0.175 -.0771729 .4203244 2012 .0085425 .0580017 0.15 0.883 -.1062683 .1233534 2011 -.0253337 .0510038 -0.50 0.620 -.1262926 .0756251 2010 -.0072519 .0502398 -0.14 0.885 -.1066986 .0921948 2009 -.0679588 .068612 -0.99 0.324 -.203772 .0678544 2008 -.0277897 .0512889 -0.54 0.589 -.1293129 .0737335 year oecd -.1168516 .0644707 -1.81 0.072 -.2444673 .0107641 gdpcapita -.0111974 .0070128 -1.60 0.113 -.0250788 .0026839 fdinetgdp -.0005893 .0018955 -0.31 0.756 -.0043413 .0031627 secondaryschoolenrollment -.0032426 .0015217 -2.13 0.035 -.0062548 -.0002304 open .0019027 .0005258 3.62 0.000 .0008619 .0029436 iprirank .017697 .0203343 0.87 0.386 -.0225536 .0579476 rdinvestmentgdp .0525804 .0189282 2.78 0.006 .0151132 .0900477 ratio Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 4.44888903 136 .032712419 Root MSE = .1659
Adj R-squared = 0.1587 Residual 3.38517466 123 .027521745 R-squared = 0.2391 Model 1.06371437 13 .081824182 Prob > F = 0.0008 F( 13, 123) = 2.97 Source SS df MS Number of obs = 137
