WP/04/185

R&D, Innovation, and Economic Growth: An Empirical Analysis
Hulya Ulku

© 2004 International Monetary Fund

WP/04/185

IMF Working Paper Research Department R&D, Innovation, and Economic Growth: An Empirical Analysis Prepared by Hulya Ulku1 Authorized for distribution by Arvind Subramanian September 2004 Abstract This Working Paper should not be reported as representing the views of the IMF.
The views expressed in this Working Paper are those of the author(s) and do not necessarily represent those of the IMF or IMF policy. Working Papers describe research in progress by the author(s) and are published to elicit comments and to further debate.

This paper investigates the main postulations of the R&D based growth models that innovation is created in the R&D sectors and it enables sustainable economic growth, provided that there are constant returns to innovation in terms of R&D. The analysis employs various panel data techniques and uses patent and R&D data for 20 OECD and 10 Non-OECD countries for the period 1981–97. The results suggest a positive relationship between per capita GDP and innovation in both OECD and non-OECD countries, while the effect of R&D stock on innovation is significant only in the OECD countries with large markets. Although these results provide support for endogenous growth models, there is no evidence for constant returns to innovation in terms of R&D, implying that innovation does not lead to permanent increases in economic growth. However, these results do not necessarily suggest a rejection of R&D based growth models, given that neither patent nor R&D data capture the full range of innovation and R&D activities. JEL Classification Numbers: O30; O31; O33; O47 Keywords: Innovation; R&D, patents; economic growth; total factor productivity; panel data; generalized method of moment (GMM) Author(s) E-Mail Address: hulya.ulku@man.ac.uk

1

This paper was prepared while the author was working in the Research Department. The author is grateful to Adam Jaffe, Tito Cordella and Ayhan Kose for their invaluable comments and suggestions.

-2Contents Page

I. Introduction ............................................................................................................................4 II. The Model .............................................................................................................................6 III. Description of Data and Methodology.................................................................................7 IV. Statistical Analysis of Data and Stylized Facts ...................................................................9 V. Empirical Analysis..............................................................................................................13 A. Estimation of Innovation Function .........................................................................17 B. Estimation of Production Function .........................................................................23 VI. Conclusion .........................................................................................................................27 References................................................................................................................................34 Appendices I. Regression Tables.................................................................................................................29 II. Correlation Table and Summary Statistics..........................................................................31 Text Tables 1. Levin-Lin-Chu Panel Data Unit Root Test ..........................................................................10 2. Durbin-Watson and Heteroskedasticity Test, 1982–971 ......................................................10 3. Rankings of Countries by Aggregate GDP, Investment, Patents, and R&D Expenditure, 1981–97 ......................................................................................................11 4. Rankings of Countries by per capita GDP, Investment, Patents, and R&D Expenditure, 1981–97 ......................................................................................................11 5. List of the Countries in Each Sample ..................................................................................12 6. Fixed Effects Regression Analysis of Per Labor Patent Flows, 1981–97 ...........................19 7. General Methods of Moments (GMM) Regression Analysis of Per Labor Patent Flows, 1981–97............................................................................................................................21 8. General Methods of Moments (GMM) Regression Analysis of Per Labor Patent Flows, 1981-1997 .................................................................................................22 9. Fixed Effects Regression Analysis of Per Labor GDP, 1981–97 ........................................25 10. General Methods of Moments (GMM) Regression Analysis of Per Labor GDP, 1981–97..................................................................................................................26 11. General Methods of Moments (GMM) Regression Analysis of Total Factor Productivity (TFP), 1981–97 ...........................................................................................28

........................................ Ordinary Least Square (OLS) Regression Analysis of per Labor GDP...................................... Summary Statistics of the Variables . 1982–97....... Ordinary Least Square (OLS) Regression Analysis of Per Labor Patent Flows...................... 1981–97............................................14 3...........-3- Appendix Tables 12...... Average Growth Rates of Per Capita GDP and Patents.................... Average Per Capita GDP and Patents.................. 1981–97..................................15 4.............................................. 1981–97 ........................ Time-Series Plots of Per Capita GDP and Patent Flows......31 15................. 1982–97 ......30 14........16 ................................. Average Per Capita Patents and R&D.........................................................32 16................15 5.......................... Time-Series Plots of Per Capita Patent and R&D Expenditure................... 1981–97........ 1981-97...... Country Codes .. 1981-97 .............. Summary Statistics of the Growth Rates of the Variables.................16 6.......................33 Figures 1..... Pairwise Correlation Table ........33 17........................................................................................................................ Average Growth Rates of Per Capita Patents and R&D..14 2...................29 13.................................................

economy R&D investment and TFP growth are positively related. This paper uses various panel data techniques and the data of 20 OECD and 10 non-OECD countries for the period 1981–97 to investigate the following postulations of R&D based endogenous growth models: (1) R&D investment increases innovation and there are constant returns to innovation. has a smaller proportional spillover effect on the aggregate stock of knowledge. (2) innovation leads to permanent increases in per capita GDP. The empirical studies of endogenous growth models generally involve testing the effect of R&D variables on total factor productivity (TFP) growth. The positive relationship between countries’ own R&D and productivity growth has been also confirmed by studies using international panel data. Helpman and Hoffmaister (1995). According to these so-called endogenous growth models. the increasing complexity of technology makes it necessary to raise R&D over time just to keep the innovation rate constant for each product. one needs data on both the input (R&D) and the output of an innovative activity.-4I. and therefore. technological innovation is created in the research and development (R&D) sectors using human capital and the existing knowledge stock. However. they alone do not allow us to analyze these models in depth. and Zachariadis (2003) provide strong evidence that in the U. he finds no evidence that these variables are positively related. First. as the number of products increases. Japan and United States to test the validity of R&D based growth models. Porter and Stern (2000) is one of the first studies that utilized aggregate level patent data to examine the . It is then used in the production of final goods and leads to permanent increases in the growth rate of output. In a more recent study Savvides and Zachariadis (2003) show that both domestic R&D and foreign direct investment increase the domestic productivity and value added growth. Scherer (1982). given that there are constant returns to innovation in terms of human capital employed in the R&D sectors. Although R&D data have enabled growth economists to shed some light on endogenous growth theories. such as Frantzen (2000) and Griffith. to examine the determinants of innovation that is the heart of endogenous growth theories. At the heart of these models is their postulation that endogenously determined innovation enables sustainable economic growth. Second. Redding and Reenen (2002). an innovation in any one product affects a smaller proportion of the economy. Jones (1995b) uses the time series plots of the TFP growth and the growth rate of the numbers of scientists and engineers in France. INTRODUCTION Recent theories of economic growth draw attention to endogenous technological change to explain the growth patterns of world economies. Aghion and Howitt (1998). Aghion and Howitt (1998) provide explanations for the contradicting results of Jones (1995b). Redding and Reenen (2002)). For example. Griliches and Lichtenberg (1984). pioneered by Romer (1986). They then argue that instead of the number of the scientists and engineers. GDP share of R&D investment should be used to take into account the size of the economy. There is also strong evidence that R&D spillovers from industrialized countries to developing countries have positive effects on the TFP growth of the latter (Coe. In particular.S. Germany. Zachariadis (2003) compares the effect of R&D on aggregate and manufacturing output and finds that the effect of R&D is much higher for aggregate economy than the manufacturing sector. Griffith.

In particular.05 percent in both OECD and non-OECD countries. Section III explains the data and methodology. they lack the support for constant returns to innovation with respect to R&D. and Trajtenberg (2001. Ariel Pakes (1985).2 percent only in large market OECD countries. This implies that innovation. They find that innovation is positively related to human capital in the R&D sectors and national knowledge stock.-5determinants and the effects of innovation. like capital stock. and is not able to explain perpetual economic growth. However. The remainder of the paper is organized as follows: Section II introduces the model. their use in microeconomic analysis is very common. second. and trade liberalization. and Section VI concludes. its contribution to the existing literature is twofold: first. Jaffe (1986. 2 . such as variation in the intrinsic value of patents and inability of patents to capture the whole range of innovation. Griliches and Schmookler (1963). leads to only short term increases in the growth rate of output. which includes the G-7. and Griliches (1990) for further details on the analysis of patent data as a measure of innovation. Griliches and Hausman (1986) and Hall. Although these results provide strong evidence for R&D based growth models as they suggest that innovation is endogenously created in the economy and it promotes economic growth. our analysis controls for a larger number of both innovation and production determinants such as technology spillovers. as neither patent nor R&D data are complete measures of innovation. The findings of the paper suggest that innovation has a positive effect on per capita outputs of both developed and developing countries. In addition. only the large market OECD countries are able to increase their innovation by investing in R&D and the remaining OECD countries seem to promote their innovation by using the know-how of other OECD countries. while the existing studies use only a small sample of OECD countries. In addition. human capital. Schmookler and Brownlee (1962). detailed analysis on the characteristics of R&D data can be found in Griliches (1994). institutional quality. and Scherer (1969). Hall. Section V presents the regression results. Jaffe. while a 1 percent increase in R&D stock increases innovation by about 0. For example. 2002) are the main contributors to the analysis of innovative activity using sector level patent data.2 They also show that there is a significant but weak relationship between innovation and TFP growth. we use various econometrics techniques to increase the accuracy of our results. a 1 percent increase in innovation raises per capita income by around 0. However. these results should not be interpreted as a rejection of the R&D based growth models. Section IV documents the statistical properties of data and stylized facts. we examine the effect of innovation on per capita GDP of both developed (OECD) and developing (non-OECD) countries. Although there are some disadvantages of employing patent data to measure innovative activity. Although our study shares some common features with Porter & Stern (2000). Though the use of patent data in macroeconomic analysis of innovation is new. they can be accounted for in the econometric models. See Comanor. 1989).

In Romer’s model R&D firms obtain patent rights for their designs. This has two implications: first. After a new design is produced. THE MODEL Our empirical model builds on the R&D based growth model of Romer (1990). labor and producer durables. The most crucial postulation of the Romer’s model that leads to sustainable economic growth is the fact that production of new designs is linear in human capital employed in the R&D sectors and knowledge stock (i. L.e. it enters into an economy in two distinct ways: a new design enables the production of a new intermediate good that can be used to produce output. (2) technological change arises as a result of intentional actions taken by people who respond to market incentives. Romer’s model explains technological change by deliberate actions of profit motivated agents. the higher the productivity of an engineer working in the research sector will be. (3) blue prints (designs) used to produce new products are nonrival. and then sell these patent rights to intermediate goods sector. second. i. Since the entry to R&D sector is free the profits of R&D firms are equal to zero. . it also increases the total stock of knowledge and the productivity of human capital in the research sector. x are human capital. respectively. 1988). the larger the total stocks of designs and knowledge are. L. devoting more human capital to research leads to a higher rate of production of new designs. Romer’s model is based on three premises: (1) growth is driven by technological change.3 The creation of new designs in R&D sector evolves according to the following equation = δH A θ A A (2) is the new where HA is total human capital in R&D sector.-6- II. they can be replicated with no additional cost. The final output is produced according to Cobb-Douglas production function α β Y ( H r . For more details on the derivation of the demand and the profit function of the monopolistic intermediate goods sector and the solution of the model see Romer (1990) and Romer (1994). 1970) and spillover model (Lucas. θ = 1). Total capital evolves according to the following equation: 3 Unlike learning by doing model (Arrow. x ) = H Y L ∫ x(i )1−α − β di 0 ∞ (1) where. firms in the intermediate goods sector can have positive profits as the entry to this sector is restricted by monopoly power. Each producer durable is produced by a monopoly in the intermediate goods sector using η units of forgone consumption and the design of that durable bought from R&D sectors. The model has three sectors: research and development (R&D) sector.e. On the other hand. H. A is knowledge stock and A designs (technological change or innovation). intermediate goods sector and final output sector.

It includes all utility patent applications in manufacturing sectors made in the U.S. Since A determines the range of durables that can be produced. gross R&D expenditure. The most important implication of this model is that countries can attain perpetual economic growth by promoting R&D sectors and investing in human capital. all available durable goods are supplied at the same level and can be denoted as x . pipes and joints. III.-7 (t ) = Y (t ) − C (t ) K (3) Because it takes η units of forgone consumption to create one unit of any type of durable. receptacles and the miscellaneous. 4 The category “others” include: agriculture-husbandry-food. Patent and Trademark Office by the inventors residing in different countries. which will be the main focus of this paper. as an input. this accounting measure of K is related to durable goods that are actually used in production by the rule K = η ∑ xi = η ∑ xi . i =1 i =1 ∞ A Because of the symmetry in the model. Substituting x=K/ηA into the production function in equation (1) results in the final form of the production function in Romer’s model Y ( H Y . it is possible to solve for x from the equation that K=ηAx. . A enters into the R&D sector directly and into the final good production process indirectly through knowledge spillovers. earth working and wells. A. x) + ( H Y A)α ( LA) β ( K )1−α − β η α + β −1 (4) Increasing returns to scale arises in both R&D and final output sectors because both sectors use non-rival knowledge stock.4 The patent counts of each country for a given year are constructed by summing up all utility patent applications made by inventors’ of that country. computers and communication. electrical and electronic and others. apparel and textile. and other macroeconomic data. Patent data is obtained from the NBER Patent Citations database. amusement devices. drugs and medical. furniture house fixtures. heating. and since η units of output are required per unit of durable goods. Utility patents are classified according to five main categories: chemical. DESCRIPTION OF DATA AND METHODOLOGY The data consists of patent applications. L.

trade is calculated adding up the total exports and imports of United States to and from each partner country.S. which takes high values for low level of risk of expropriation. openness in current prices (PWT. which reduces the amount of noise in data.S. regardless of age. gross fixed investment.S. and the U. In other words. It comprises of R&D expenditure in business enterprises. each country’s GDP share of U. The gaps in the data are interpolated by averaging the observations in the preceding and succeeding years. 2002). patent applications such as distance from the United States. dollars. r is the growth rate of patent flows. The potential disadvantages of using U. to the population of the age group that officially corresponds to the level of secondary school education. we used patent applications instead of granted patents as the time lag between the application year and grant year can be very long. International Country Risk Guide). It is defined as total expenditure on R&D performed on national territory during a given period. labor population. government sector. with the exception of the variables that are the share of GDP. and δ is the depreciation rate of patents. The R&D stock has been constructed using 20 percent depreciation rate.S. imports and exports of manufacturing goods (OECD.S. Pt is patent flows at year t. expropriation risk index (World Bank. The stock of R&D expenditure (R&D stock) has been employed in the regression analyses to take into account the effects of past research efforts on innovation.5 The main reason for using the patent applications made in United States is to isolate the effects of different regulations in each country on the number of patent applications.-8The patent stock is calculated using 20 percent depreciation rate as suggested in the literature. in the U. The expropriation risk index ranges from 1 to 10. 2002).S. Trade in manufacturing goods is calculated by summing up the values of imports and exports in constant 1995 U. and dividing this total by each country’s GDP. secondary school enrollments (WDI. which includes the R&D performed within a country and funded from abroad but excludes payments made abroad for R&D. Patent Office all inventors face the same regulations and quality control for their innovation. The series are deflated using the 1995 implicit price deflator and converted to US dollars using the monthly averages of the exchange rates obtained from the OECD database. higher education and non-profit firms. In addition. or the economic alliance with United States have been taken into account throughout the analysis. . The patent stock for subsequent years is calculated using the formula Ps t = Pt + (1 − δ ) Ps t −1 . Gross R&D expenditure (GERD) data are obtained from the OECD Main Statistics and Technology Indicators database. trade share (IMF Direction of Trade Database (IMFDOT)). patent counts and the expropriation risk index. Similarly. dollars. The remaining macroeconomic variables are obtained from the following databases: GDP.6). All variables are in constant 1995 U. 5 The formula used to calculate initial patent stock level is Pst −1 = Pt /(r + δ ) where Ps is patent stock. The gross ratio of secondary school enrollment is measured as the ratio of total enrollment.

non-OECD. The gaps are interpolated using moving averages of the five-year observations. investment. while Portugal. R&D and patent applications. if the observations in 1980 and 1985 are equal to 100 and 120. As seen from Table 3. As seen from Table 1 and 2. data do not have unit root and heteroskedasticity (in most of the countries). investment. then the observation for 1981 is calculated as (100+120)/2=110. STATISTICAL ANALYSIS OF DATA AND STYLIZED FACTS This section examines the statistical properties of the data and presents some stylized facts about the main variables of the Romer’s model. Specifically. Throughout the analysis. In addition. patents and R&D. namely. the G-7 countries are in the highest rank. investment. R&D and patents. which suggests a positive correlation among these variables. Spain. Table 3 documents the rankings of the countries in terms of their average levels of GDP. eight countries also rank high in per capita R&D expenditure. Our results are not sensitive to different types of interpolation. R&D and patents behave across country groups with different income and market sizes. Ireland. the first order autocorrelation problem has been eliminated by either differencing the data or using Prais-Winsten estimation technique. As Table 4 shows.-9The series for secondary school enrollments are available for every five years. per capita GDP. we also documented the averages of these variables in Figures 1 to 4 for the following country groups: OECD. Similarly. Greece. investment. respectively. large and small-market OECD. In addition. non-OECD sample has been included in the analysis wherever R&D data are not used. This again suggests positive relationships among per capita GDP. To take into account the size of the economy. while seven countries rank high in per capita GDP. the countries are also ranked for per capita levels of the above variables. New Zealand and Iceland are in the lowest rank of all four variables. eight of them also have higher levels of R&D expenditure and patent applications. R&D and patent applications. To see how per capita levels and the growth rates of GDP. countries ranking the first nine (last nine) according to their aggregate GDP and investment levels have been referred to as larger market (smaller market) countries. while Greece.6 IV. Portugal. and high income and low-income OECD countries. investment. R&D expenditure and patent applications. countries ranking the first nine (last nine) in terms of per capita GDP and investment have been referred to as higher income (lower income) countries. The remainder of this section provides some stylized facts about the relationships among GDP. Of the ten countries ranking high in per capita patent. and Ireland rank the lowest in per capita levels of GDP. investment. though they exhibit first order autocorrelation. Switzerland and Japan rank the highest. 7 . The lists of the countries in each sample are presented in Table 5. These samples are constructed using the rankings presented in Tables 3 and 4.7 6 For example. out of nine countries having both higher levels of GDP and investment. and the observation for 1982 is calculated as (110+120)/2.

88 0. investment and US trade/GDP all of which include a trend.63 0.30) 0. openness (PWT 6) expropriation risk index (World Bank.45 0.03 -3.53 1.08(0.01 -2. Levin-Lin-Chu Panel Data Unit Root Test1 Full Sample2 t-star P>t -3. GDP.03 ---3.98 1.16) 0. None of the regressions include a constant except for patent stock.21(0365) 0.15(0.02 -1. Table 2. All regressions except for GDP and secondary schools (which include trend and no constant.01 -3.35 0.08) 1.82 0.90(0.09 1.60 Italy Japan Malaysia Netherlands New Zealand Norway Philippines Portugal Singapore South Africa Spain Sweden Switzerland United Kingdom Venezuela 0.26(0.08 2.16(0. 1/ t-star statistics is distributed as standard normal under the null hypothesis of nonstationarity. 2002).10 2.78) 1.02 2.00 -1.00) 1.32) 1. and all variables are in natural logarithms. 1/The heteroskedasticity test is applied to the predicted values of GDP.20 2.04 0.15(0.10) 0.08) 0.42(0.00 Non-OECD Sample4 t-star P>t -2.16 0. and patent stock are normalized by labor.97 1. which are obtained from the regression equation of production function: GDP= α0 + α1 (investment) + α2 (patent stock_2) + α3 (secondary school_2) + α4(openness) + α5 (expropriation risk index) + et.56 -3.00 OECD Sample3 t-star P>t -2. secondary school enrolments (WDI.09 2.98 2. patent applications (NBER Patent Citation Database).79(0.94) 8.42 0.88 0. Durbin-Watson and Heteroskedasticity Test.04 -1.63 2. 1982–971 Heteroskedasticity Test H0: Constant Variance Durbin Watson (d-statistic)2 Heteroskedasticity Test H0: Constant Variance Durbin Watson (d-statistic) Country Country Argentina Australia Austria Belgium Brazil Canada Denmark Finland France Greece Hong Kong Iceland India Indonesia Ireland 3. 2/ The values of d-statistics below or above 2 indicate the presence of first order autocorrelation.44 0.00 -1.00 -2.15) 0.08) 2.03) 0.36 2.00 -5.32 2.78 0.01(0.23 1.00 -5.29 0.00(0.28) 2.29(0.01(0. 2/All regressions are augmented by one lag and none of them include a constant except for US trade/GDP.00 -5.31(0.10 Table 1.61) 0.46 2.34) 0. 4/All regressions are augmented by one lag except for patent stocks.87 0.07) 0.92(0.52 0.28) 3.17 2.34) 1.84) 3.49 1.94(0.76 0.00 -3.05(0.62(0. 2002).05 -2.03(0.70) 0. International Country Risk Guide).86 Notes: Figures in parenthesis are p values.29 0.28(0.91 0.00 -4.33 0.07) 0.44 0.31) 3.61) 0.00 -8. GDP. investment.11 2.36) 4.16 0.72 1.84(0. import/trade in manufacturing sector (OECD. 2002).85) 3.22) 2.18(0.07(0.81 0..60 2.03 -5. employment (WEO.07(0. .01 -4. investment. respectively) include a constant and have no trend.00 -2.14 1. 3/ All regressions except for investment and openness (which are augmented by four and two lags.65 1.00 -3.83(0.01 -3.58(0. .019 0.81 0.17 1. All variables are in natural logs. Patent.54 1.00 -8.16(0.20 0.62 0.04(0.39 2.00 -1.98(0.43) 1. respectively) are augmented by one lag.25 2.24(0.05 GDP Secondary school_2 Patent stock_2 Investment Openness Patent flows_2 R&D stock_2 Expropriation risk index Import/trade in manufacturing US trade/GDP Sources: GDP.00 0.45) 1. Investment and R&D are normalized by labor series.60) 0.17 0.51 0.84 1.69 0.07 0.02 2.01 -1.

866 1. Patents. 1981–97 Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Investment Japan 1. Investment. .005 Patents Switzerland Japan Sweden Canada Finland Netherlands France United Kingdom Austria Denmark Belgium Norway Australia Italy New Zealand Iceland Ireland Spain Greece Portugal 176 157 95 68 64 58 49 45 44 42 36 29 26 20 16 15 14 3 1 1 R&D Expenditure Switzerland 870 Sweden 730 Japan 705 France 466 Iceland 442 Norway 436 Finland 436 Denmark 390 Netherlands 376 United Kingdom 375 Belgium 306 Canada 297 Austria 272 Australia 267 Italy 216 New Zealand 150 Ireland 137 Spain 88 Greece 61 Portugal 52 Sources: GDP and Investment (WDI (2002)).177 2.347 3.061 GDP Switzerland Japan Denmark Norway Austria Iceland Sweden Belgium France Finland Netherlands Canada Australia Italy United Kingdom New Zealand Ireland Spain Greece Portugal 42.094 Ireland 8.539 4.507 23. R&D (OECD (2002)).236 Switzerland 5.980 4.886 Canada 91.378.536 4.863 7.952 2. Rankings of Countries by per capita GDP.983 Italy 180.039 3.546 4. dollars. Investment and R&D Expenditure are in millions 1995 U.412 France 26.509 2.417 5.183 17.106 2. dollars and averaged over 1981–97.048 Finland 2.255 Canada 8. Investment.889 Iceland 1.334 Switzerland 59.S.152 26.500 159.442. Patents (NBER Patent Citation Database). and R&D Expenditure. Patents.851 Sweden 34.056 17.842 3.599 Australia 4.S.244.551 Italy 12.423 13.286 2.134 Sweden 6.578 France 255.012 Norway 1.445 Patents Japan France United Kingdom Canada Switzerland Italy Netherlands Sweden Australia Belgium Austria Finland Denmark Spain Norway New Zealand Ireland Greece Portugal Iceland 19.100 213.169 Spain 96.507 Norway 31.824 36.875 4.752 2.476 Spain 3. Rankings of Countries by Aggregate GDP.561 1.100 106.460 Greece 21.000 304.138 30. and R&D Expenditure. Notes: All series are in 1995 U.112 Denmark 2.830 48.200 245.700 505.340 Portugal 20.800 117.996 24.781 Denmark 23.889 29.500 123.366 Austria 45.397 23.153 8.132 858 809 433 358 339 318 217 134 121 55 51 10 5 4 R&D Expenditure Japan 86.395 Belgium 3.469 New Zealand 10.177 1.700 Netherlands 67.700 489.271 United Kingdom 21.100 287.401 18.096 4.539 Australia 64.617 10..229 GDP Japan France United Kingdom Italy Canada Spain Netherlands Australia Switzerland Belgium Sweden Austria Denmark Norway Finland Greece Portugal New Zealand Ireland Iceland 4.400 89.700 976. 1981–97 Rank 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Investment Japan Switzerland Norway Austria Finland Iceland Denmark France Netherlands Belgium Sweden Australia Canada Italy New Zealand United Kingdom Ireland Spain Greece Portugal 10.410 52.000 980.700 2.729 12. Patents are in per million people.149 Belgium 40.848 Greece 621 Portugal 515 New Zealand 511 Ireland 485 Iceland 111 Note: GDP. Table 4.169 Austria 2.492 24.11 Table 3.487 9.700 202.825 Netherlands 5.823 4.000 346.504 2.502 United Kingdom 155.173 18.000 1.419 Finland 24.555 24.800 6.040 15.054 25.

countries with high (low) aggregate GDP and investment are referred to as large (small) market countries.. and countries with high (low) per capita income are referred to as high-(low) income countries.12 - Table 5. as it does not have the openness variable. China is not included in the analysis. . List of the Countries in Each Sample Full Sample Argentina Australia Austria Belgium Brazil Canada Denmark Finland Hong Kong France Greece Iceland India Indonesia Ireland Italy Japan Malaysia Netherlands New Zealand Norway Philippines Portugal Singapore South Africa Spain Sweden Switzerland United Kingdom Venezuela Non-OECD Argentina Brazil Hong Kong India Indonesia Malaysia Philippines Singapore South Africa Venezuela OECD Italy Japan Netherlands New Zealand Norway Portugal Spain Sweden Switzerland United Kingdom Australia Austria Belgium Canada Denmark Finland France Greece Iceland Ireland Large-Market (OECD) G-7 Canada France Japan Italy United Kingdom Others Spain Netherlands Australia Switzerland Small-Market (OECD) Austria Denmark Norway Finland Greece Portugal New Zealand Ireland High-Income (OECD) Switzerland Japan Denmark Norway Austria Iceland Sweden Belgium France Low-Income (OECD) Canada Australia Italy United Kingdom New Zealand Ireland Spain Greece Portugal Note: When dividing the OECD sample in sub samples.

These results are consistent with the premises of R&D based growth models that there is a positive association between R&D and innovation. Within the OECD countries the market size is positively associated with both per capita GDP and patent applications. The time series plots of these variables also indicate that these variables move very closely over time in the majority of the countries (Figures 5 and 6). According to Figure 2. and innovation and per capita GDP. In particular. High income and large market OECD countries have the highest. The average levels and the growth rates of per capita patent applications and GDP for the OECD and non-OECD countries are documented in Figures 3 and 4. while they are negatively associated with the growth rate of patents. OLS estimation has been carried out as well to provide a benchmark model. and that per capita R&D and patents are positively correlated. high-income OECD countries have the highest per capita GDP and patents. Each of these techniques has their own merits. while lowincome OECD and non-OECD countries have the lowest per capita GDP and patent applications. both the levels and the growth rates of per capita GDP and patents are positively related across country groups. while low income and small market OECD countries have the lowest per capita R&D expenditure and patent applications. The following section employs various panel data regression techniques.. we still observe a positive relationship between the growth rates of per capita R&D and patents. However. R&D and patent applications reveal that both the levels and the growth rates of these variables are positively correlated across countries and over time. to 8 In addition to the fixed effects and GMM analyses. V. However. According to Figure 3. as seen from Figure 4.13 Figure 1 compares the average per capita levels of R&D expenditure and patent applications across the country groups. provided that there is no endogeneity problem and the lagged dependent variable is not included in the analysis. In addition. The fixed-effects regression analysis accounts for country fixed effects and yields consistent estimators of the coefficients. both the cross sectional and time series comparisons of per capita levels and the growth rates of GDP. the high income and the large market OECD have the lowest growth rates of these variables. EMPIRICAL ANALYSIS The estimations of both innovation and production function have been carried out using fixed-effects and Arellano-Bond GMM estimators8. by including instrumented lagged dependent variable in the analysis. This indicates that income level and market size are positively correlated with per capita R&D and patents. investment. In brief. Nonetheless. and provides more comprehensive analysis of the relationships among R&D. income level and the market size are positively associated with the growth rate of R&D expenditure. innovation and per capita GDP. the growth rates of these variables are negatively related with the income level and market size. which documents the average growth rates of per capita R&D and patents for each country group. The GMM analysis accounts for country fixed effects and yields consistent estimators in the presence of lagged dependent variable. .

Average Growth Rates of Per Capita Patents and R&D.000 4. OECD Total Non – G7 G–7 Large – Market OECD Small – Market OECD High – Income OECD Low – Income OECD .000 6.06 0. patents (NBER Patent Citation).000 12. Note : Growth rates are calculated as log differences of the series.000 2. patents (NBER Patent Citation) Figure 2.04 0.14 Figure 1.02 0.000 8..03 0. 1982–97 0.01 0 Per Capita Patent Growth Per Capita R&D Growth Sources: R&D Expenditure (OECD).000 0 OECD Total Non – G7 G–7 Large – Market OECD Small – Market OECD High – Income OECD Low – Income OECD Patents (per 100 mn) R&D(per 10 mn) Sources: R&D Expenditure (OECD). Average Per Capita Patents and R&D.000 10.05 0. 1981-97 14.

and GDP (WDI. Non – G7 G–7 Large – Market OECD Small – Market OECD High – Income OECD Low – Income OECD Non-OECD . Average Per Capita GDP and Patents. and GDP (WDI.04 0. 2002).08 0. Investment.000 14. 2002).000 7. Note: Growth rates are calculated as log differences of the series.. Average Growth Rates of Per Capita GDP and Patents. Figure 4. 1981–97 42.15 Figure 3.000 28. 1982–97 Per Capita Patent Growth Per Capita GDP Growth Sources: Patents (NBER Patent Citation).000 0 OECD Total 0.02 0 OECD Total Non – G7 G–7 Large – Market OECD Small – Market OECD High – Income OECD Low – Income OECD Non-OECD Patents (per 100 mn) Per Capita GDP Sources: Patents (NBER Patent Citation). Investment.000 21.000 35.1 0.06 0.

6 10.6 -7.4 -8.4 10.4 9.5 -8 -7.2 10.8 -10.8 10 9.5 -14 -13.8 -8.3 -8.2 10.5 -11 -10.9 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 10 10.4 -10 -9.4 -8.6 -7.6 -15 -14.1 10.5 -10-9.6 -8.75 9.5 -10 -9.2 -10.2 10.5 -9 -9.6 -7.7 -7.2 -8 -7.7 -9.7 9.2 -7 -6.55 10.5 -12 -10.6 -8.2 -8 -7.8 -7.55 9.8 -7.5 -10.5 -14 -13.8 -10 -9.05 10. Note: See Appendix II.5 JP GB SE CH JP 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 GB SE CH 1980 1985 1990 1995 10.8 -8.5 -9-8.6 -8.4 10.4 10.2 -8 -8.5 -11 -10.8 -9.65 10.4 -9.75 -7.5 -8.5 -7 -6.3 -10.4 -10.5 -10 -9.6 9. population (WDI).79.2 9.6 -10 -9.2 -8.5 -9 -8.6 -10.5 -11 -10.15 9.5 -10.2 9.2 -10.4 -8.3 10.5 FI IS NZ BE FI IS appyear NZ BE 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 10. -16 -15 -14 -13 -11.8 -9.7 -8.2 -10.7 -8.5 -9 -11 -10.4 -9.6 -8.5 -10-9.8-9.5 -8.5 -8.8 -8.16 Figure 5.8 -8.5 -9 -8.9 1010.4 -7.5 -8 -7.4 9.5-8-7.2 9.1 10.6 AT DK GR NL AT DK 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 GR NL Graphs by country 1980 1985 1990 1995 9.8 lnpcpat lnpcpat -12.5 -8.5 -10 -11 -10.5 -11 -10.5 9.6 -7.2 -11 -10.2 -8.5 -12 -11.1 10.9 -9.3 10.5 -13. Patents (NBER Patent Citation).8 -8.6 -10 -9.3 9.9 10 10.5-10 -9.5 8.5 -12 -10.2 -8 -7. Time-Series Plots of Per Capita Patent and R&D Expenditure.2 ES IE NO AU ES IE 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 NO AU 1980 1985 1990 1995 9.8 -9.89.9 10 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 10.5 -8 -7.8 -8. R&D Expenditure (OECD).1 -8 -9.8 9.7 -9.5 -10 -9.4 -8.8 -8. Table 17 for the list of the country codes.999.8 -9.5 -10 -9.1 10.5 -13 -10.5 -10.5 10.91010.8 9.8 -7.2 -7 -6.8 Sources: GDP.5 -8 -10.9 -9.6 -8.2 10.5 -13.3 10.2 9.15 10.6 -9.8 -7.5 -12.9 -9.5 -10-9.75 -9 -8.8 -10.1 -10 -9.5 -9 -9.9 9.05 10.5 -13 -10.5 1010.2 -8 -7.6 -10.8 9.2 -10-9.5 -11 -10.5 9.5 -11.6 -10.5 -10 -12-11.1 10.3 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 Sources: Patents (NBER Patent Citation).6 9.8 -10.1 9.1 10.1 10.2 -10-9.2 -16 -15 -14 -13 IT PT CA FR IT lnpcgerd PT CA FR 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 .4 -10.1 -10 -9.95 10 10.15 10.2 9.7 10.4 -8.2 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 appyear 9.5 -9 -8.5 -7 -6.6 -9.35 10.15 10.5 -9 -11 -10.5 -9 -8.5 -10.9 -8.6 -10.4 -7.5 -13 -12.9 -9.05 10.85 9.25 9.6 -7.5 -8.2 10.6 9.4 -9.1 10.2 -11 -10.5 -11.65 9. Time-Series Plots of Per Capita GDP and Patent Flows.5 -13 -12.9 -9.8-9. R&D (OECD).25 9.6 9.5 -8 -7. -8.8 . 1981–97 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 1980 1985 1990 1995 lnpcgdp 9.1 -8 -7.9 -8.6 -8.8 -10.6 10.2 10.lnpcp at -15 -14. 1981–97 Figure 6.9 -9.6 -9.7 -8.2 -8 -7.5 -10-9.5 -12 -11.9 -7.8 -9.7 9.2 -10.5 -10.9 -8.69.9 -9.1 10.2 10.7 9.1 -10 -9.2 9.5 -10.5 Graphs by country -12 -11.8 lnpcpat -10 -9.5 -11 -10.7 9.7 9.2 -9 -8.8 -7.2 -11 -10.6 9.3 9.1 -10 -9.4 -8.2 9.3 -11.

The results of the AR (2) test and the Sargan test (which tests the correlation between regressors and the error term) are reported at the end of each GMM estimation results. Non-OECD. The log linearized version of the above model is Log(A) = Log(A) +θ Log(H). and OECD. the OECD countries are grouped within themselves as low income.9 However. Moreover. small market and large market OECD countries. . The reason for using these sub samples is to find out whether or not countries with different per capita incomes and market sizes behave differently in terms of the model.10 A. the first order autocorrelation problem has been accounted for either by employing the Prais-Winsten estimation technique. though more focus will be given to GMM results. or using the first differences of the series. In addition. Therefore. For more technical details of this estimation see Arellano and Bond (1991). and the common shocks to all countries. GMM accounts for endogeneity problem. high income. In addition. 10 . As mentioned in section II. A stock of innovation (or knowledge stock) and H is human capital devoted to R&D. 9 The GMM yields consistent estimators provided that data do not have AR (2) and the regressors are not correlated with the error term. A is the where. and there are constant returns to innovation. Estimation of Innovation Function This section investigates the postulation of R&D based growth models that technological innovation is created in the R&D sectors. and Arellano-Bond Linear GMM Estimator in the reference book of STATA 8. This section is allocated for the estimation of θ and other determinants of innovation using international panel data from 19 OECD countries for the period 1981-1997. to obtain accurate estimations of coefficients the empirical analysis has been undertaken for nine samples. (5’) Equation (5′) tells us that a 1 percent increase in A and H increases innovation by 1 percent and θ percent. The time dummies have been included in all regressions to control for the time trend. Furthermore. Due to the scarcity of R&D data. suggesting a separate analysis for each sample. the analysis has been conducted for 19 OECD countries from 1981 to1997. the magnitude of θ should be equal to 1 in order for the growth rate of output to grow continuously. the results of both estimations will be reported throughout the paper. The regression equation is derived from equation (2) in Section II = AH θ A (5) is flows of innovation (or knowledge flows.. respectively. in Romer’s model. as in the Romer’s model).17 some extent. These samples are: Full Sample. since GMM uses the first differences of the series it might cause loss of information in the data. See footnote 7 for the details on the construction of these samples. the results of the Chow test reveal that the coefficients are not constant across these groups.

S. trade controls for the effect of economic alliance with the United States on the numbers of the patent applications made in the U. Expropriation risk is significant only in the large market OECD countries. share of GDP is significant in the G-7. In addition. In addition. (3) the differences in the initial levels of knowledge stock across countries are accounted for in both fixed effects and GMM analyses. Patent Office by the inventors of different countries. inclusion of the patent stock as a regressor leads to a multicollinearity problem. the imports of manufacturing goods account for technology spillovers. However.50 percent in low-income OECD countries.11 However. it has been taken into account using different techniques: (1) instead of R&D expenditure we include stock of R&D expenditure (R&D stock) to proxy for knowledge accumulation over time as well as the human capital in the R&D sector. including G-7. other large market OECD countries.S. Similar to the fixed effects results. respectively.18 In the empirical analysis of equation 5’ the flows of innovation ( A) and human capital in the R&D sectors (H) are measured by patent applications (patent flows) and stock of R&D expenditure (R&D stock). and 0. the coefficient of R&D stock is positive and significant in the G-7. According to these results a 1 percent increase in per capita R&D stock increases innovation by 0. 12 . while the U. the relationship between R&D expenditure and innovation is positive only in the G7 and large market countries. 11 . However.S. (2) the first lag of patent flows in the GMM analysis can serve as a an instrument for knowledge stock. and the share of U. Other control variables included in the regression analysis are expropriation risk. secondary school enrolments are included in the analysis to capture the effect of overall human capital of a country on its innovation level. Table 12 for OLS regression results. and the low-income OECD countries. and the large market countries.40 percent in the G-7. trade in each country’s GDP. it is not significant in any of the samples. The expropriation risk index captures the overall institutional quality of countries which takes higher values for lower risk of expropriation. The patent stock variable could be a good candidate for knowledge stock. which could be attributed to the the fact that OLS results are biased in the presence of country fixed effects. in equation 5’ has not been accounted for directly in our regressions due to a lack of an independent measure for A. See Appendix I. imports of manufacturing goods as share of total trade in manufacturing goods. The knowledge stock. large market OECD and low-income OECD countries only. non-G-7 and the small market samples.. Both patent flows and R&D stock have been normalized by labor series to control for the size of the economy. different from the fixed effects results. and the share of U. the coefficient of the import share of trade in manufacturing goods is positive and significant the countries that do not have significant coefficient on R&D. As seen from the Table. A.S. the coefficient of R&D stock is negative and significant in the full. since patent stock and the R&D investment are highly correlated. This might suggest that countries that do not have effective R&D sectors are using the know how of other countries to increase their innovation.12 Although secondary school enrollments have high t values in the large market and low-income OECD countries. The results of the fixed-effects regression analysis are reported in Table 6.

288 (3.406 (4.46) -0.67) -0.082 (0. International Country Risk Guide).677 (3. 2002).153 (0.37) -1.158 (2.649 (2.10) 0.53) -0.121 (1.95) -0.99 120 8 HighIncome 0.99 210 14 G-7 0.668 (1.30) -0.52) -0.375 (1.49) 1.S.135 (1.88) -0. Fixed Effects Regression Analysis of Per Labor Patent Flows.86) 3.414 (8.389 (7.754 (1.925 (3.74) 0.81) -0.30) 0. patent applications (NBER Patent Citation Database).392 (1.52) 2..359 (0.27) 0.77) 3.70) -0. Greece is not included in the analysis.286 (0.248 (0. 2002).00) -0.35) 0.19 Table 6.05) 1.46) 0.66) 0.51) -0.29) -0.165 (0.S. trade/GDP Openness R squared Observations Number of ifs 0.056 (2. import/trade (OECD.151 (0.04) 0.126 (1.26) 0.089 (0.136 (1.157 (0.35) -0.30) 0.219 (0. . as it does not have data on imports in manufacturing sector.41) 0.043 (0. expropriation risk index (World Bank.19) 0.76) -0.55) 0. trade (IMF Direction of Trade Database). 1981–97 Full Second lag of R&D stock Second lag of Secondary school Expropriation risk Index Manufacturing import/trade U.99 120 8 Sources: R&D stock (OECD.36) 0. openness (PWT 6) employment (WEO.162 (1.99 135 9 Small Market 0.89) 0.59) -0.32) -0.74) 0.590 (1.252 (0.11) 0.022 (0.396 (2.99 135 9 LowIncome 0.36) -0.516 (4.99 285 19 Non-G-7 0.542 (2.724 (0.57) 0.030 (0. Notes: z statistics in parentheses.99 75 5 Large Market 0.086 (0.423 (1. 2002).090 (5. All variables are in natural logs and normalized by labor. U.122 (0. All regressions include time dummies.23) 0.292 (1.

3 percent. none of these measures yield significant coefficients. Although the sample size is not large enough to draw strong conclusions. Australia and Spain) and small market (New Zealand. The coefficient of R&D stock is significant only in large-market OECD. in Romer’s model. This might imply that these countries import the know-how of other OECD countries to promote their innovation. and 0. which. secondary school enrollments seem to have an important effect on innovation in the G-7 countries only. imports of manufactured goods as share of GDP. the coefficient of the first lag of patent flows is positive and significant in all of the countries. with the magnitude of around 0. Only the imports in manufactured goods as share of trade in manufactured goods have a significant coefficient and expected sign. 13 Among the other control variables. As seen from Table 8.14 13 Different measures of technology spillovers across countries have been employed as well. such as foreign direct investment inflows. Table 8 reports the results of the regression analysis for the sub groups of low-income OECD countries in terms of their market sizes. as suggested by Romer’s model. In addition. Italy.20 percent increase the large market OECD countries including the G-7. only the large market countries have significant coefficient on R&D stock. economic alliance with the United States and the openness do not seem to matter in terms of the numbers of patent applications made in the United States. Another important observation from Table 7 is that the most of the countries that do not have effective R&D sectors have significant coefficient on import share of trade in manufacturing goods. the results reported in Table 8 are consistent with our hypothesis that market size is an important factor for the effectiveness of R&D sectors. However. This suggests that the knowledge flows of the previous year have a strong positive effect on the current innovation level. including G-7 and low-income OECD countries. instead of investing in formal R&D sectors. then it means that a one percent increase in knowledge stock leads to 0. A. takes into account endogeneity problem by including instrumented lagged dependent variable in the analysis. to some extent. while institutional quality matters only in the OECD countries with large markets. a 1 percent increase in R&D stock leads to around 0. According to these results. the R&D and patent stock of other countries multiplied by the trade share of each country. Canada. As seen from Table 7 the results are very similar to the ones obtained in the fixed effects analysis.3 percent increase in innovation. Ireland and Portugal) OECD countries.20 - Table 7 reports the results of the Arellano-Bond GMM estimation. If the first lag of patent flows in this regression is taken as an instrument for knowledge stock. However. The fact that the coefficient of R&D stock is significant in the G-7 and the large market OECD countries provides support for the models that emphasize the importance of market size in promoting R&D sectors and innovation. driving force behind the significant relationship between R&D stock and innovation in low-income OECD countries is not clear as this group includes both large market (UK.30 percent increase in low income OECD.. 14 . Griliches (1957) and Acemoglu and Linn (2003).

047 (0.340 (3.83) -0.17) -0.024 (5.178 (0.191 (1.315 (1.162 (2.72) 2.05) 0.08) 0.231 (3. 3/ H0: errors in first difference regression exhibit no second order serial correlation.89) -0.00 0. patent applications (NBER Patent Citation Database).362 (0.06) 0.56 70 5 Large Market 0.359 (3.522 (0.06) 0. General Methods of Moments (GMM) Regression Analysis of Per Labor Patent Flows.304 (3.71) 0.297 (4.228 (2.062 (3.01) 0.09 56 182 14 G-7 0.354 (2.00 0.44) 0.35) -0.080 (0.23) 1.89) 1.44) 0.633 (2.844 (2.72) -0.034 (0.188 (2.89) --0.164 (0.29) 4.187 (1.339 (1.94) 0.29) -0.005 (0.42) 0.00 0.178 (1.93 0.63) 1.43) 0. import/trade (OECD.S.317 (3.314 (0. For example.455 (1.066 (0.72) -0.169 (0.095 (0.80) 0. 2002).05) 0.20) 0.130 (1.184 (1.169 (0.51) 0.133 (1.154 (1.17) -0.26) -0.89) -0.37 112 8 Sources: R&D stock (OECD. .02) -0.53) -0.309 (2.583 (1.046 (0. Notes: Absolute values of z statistics are in parentheses.66) 0. as it does not have data on imports in manufacturing sector.289 (0.38 117 9 Low Income 0.650 (2.039 (0.297 (1. 1/ To eliminate the first order autocorrelation problem and the serial correlation between residuals and the regressors different lag lengths of the dependent variable are included in the analysis of different groups.99 0.96 0.21 Table 7.715 (0.073 (0.09) 0. 2002). trade (IMF Direction of Trade Database).71) 0. expropriation risk index (World Bank.356 (0.04) -0.27) 0.09) 0.071 (2. in large market and low income country groups it was sufficient to include one lag of the dependent variable.14) -0.298 (3.62 112 8 High Income 0. U. It should also be noted that the inclusion of the different lag lengths is important only to solve autocorrelation and endogeneity problem and it does not have an effect on the sign and the significance of the coefficients of the explanatory variables. International Country Risk Guide).89) 0.226 (3.99) ---0. 1981–971 Full Second lag of R&D stock Second lag of Secondary school Expropriation risk Index Manufacturing import/trade U.32) -0.75) -0.42) 0.135 (2.171 (1.92) 0.38 247 19 Non-G-7 0.28) -0.323 (0.52) 0.28) -0.697 (2.66) 0.057 (0.81) 0.104 (1..05) 0. trade/GDP Openness First lag of per labor patent Second lag of per labor patent Third lag of per labor patent Constant Sargan test2 (p-value) AR(2) test3 (p-value) Observations Number of countries 0.84) 0.36) -0.27 126 9 Small Market 0.96) -0.94) -0.178 (2.04) 0.52) 0.576 (6.015 (0.17) 0.80) -0. 2/ H0: regressors are not correlated with the residuals. openness (PWT 6) employment (WEO.77) -0.S. whereas in the small market countries three lags were needed to eliminate first order autocorrelation and endogeneity problem.59) 1.38) -0. All variables are normalized by labor and log differenced.122 (1.010 (0.030 (3. 2002). Greece is not included in the analysis.012 (0.761 (2.90) 0.

01) -0.715 (1.516 (3. U.299 (2. and normalized by labor. All variables are in natural logarithms.176 (1.423 (1.17) 1..001 (0.20) 0. Greece is not included in the analysis. General Methods of Moments (GMM) Regression Analysis of Per Labor Patent Flows.34 42 3 Sources: R&D stock (OECD.292 (1. 2002).378 (1.808 (0. import/trade (OECD. 2002). openness (PWT 6) employment (WEO.00 0.S.19 70 5 Small Market -0.15) 0.056 (0. expropriation risk index (World Bank.42) 0.513 (4.195 (0.41) 4.09) -0. 2002).S. as it does not have data on imports in manufacturing sector.24) 0.22 - Table 8.01) 0.00 0. International Country Risk Guide). trade (IMF Direction of Trade Database).59) 0.22) 0. trade/GDP Openness First lag of per labour patent Constant Sargan test1 (p-value) AR(2) test2 (p-value) Observations Number of countries 0.04) 1.40) -0. patent applications (NBER Patent Citation Database).686 (2. 1/ H0: regressors are not correlated with the residuals 2/ H0: errors in first difference regression exhibit no second order serial correlation. Notes: Absolute value of z statistics in parentheses.28) -0. 1981-1997 Low Income OECD Large Market Second lag of R&D stock Second lag of Secondary school Expropriation risk Index Manufacturing import/trade U.50) -0. .001 (0.993 (1.072 (0.90) -0.

fourth. and the secondary school enrollments as share of the population who are in secondary school age. The resulting regression equation is shown below yt = α ht + γ it + (1 − α − γ ) χ t + ε t (6’) where yt. and physical investment. L. second. Production function presents constant returns to scale in its inputs L. Y. instead of population) to normalize data is to eliminate multicollinearity problem caused by the high correlation between labor and investment.. respectively. there are no constant returns to innovation. which include the G-7 and some low-income OECD countries. new product and human capital are measured by the gross fixed investment. and the equation has been log linearized. ht is the human capital as share of population.e. stock of patent applications from the U. seem to increase their innovation by investing in R&D sectors. The investment. H and K. H. the results obtained from the analysis of the relationship between R&D stock and innovation (patent applications) allows us to make the following conclusions: first. R&D intensity changes across countries with different market sizes and income levels15.23 - In brief. and A are total output. risk 15 This has also been confirmed by the Chow test. These results are consistent with the premises of R&D based growth models that innovation is endogenously created. composite investment. When deriving the regression equation from the above model. respectively. they do not provide support for constant returns to innovation. Equation (3) in Section II is the basis of our regression analysis Y ( H Y . According to this test we reject the hypotheses that the coefficients are the same across samples. decomposed into new products. human capital and the knowledge stock of economy. K. Patent Office.S. L. labor. investment and new products. B. and xt are per labor output. it. This could be because our data is not able to capture the full range of innovation activities. third. x) = ( H Y A)α ( LA) β ( K )1−α − β η α + β −1 (6) where. However. Here K includes both physical investment and new products. 16 . all variables are normalized by labor series16. Estimation of Production Function This section is allocated to the analysis of the relationship between innovation and per capita GDP. The reason for using labor series (i. only the large market OECD countries. respectively. technology spillovers have significant effects on the innovation of countries which do not have efficient R&D sectors. increasing returns to scale arises because of the knowledge stock that enters the production function through new products and spillovers from R&D sectors. In addition to the variables in the regression equation above.

the coefficient of patent stock is positive and significant in all samples. For example. with the magnitude of 0. a 1 percent increase in investment increases output by around 0. implying that trade liberalization and institutional quality are important determinants of the per capita income levels of countries.17 Table 10 reports the results of the GMM estimation. Though the coefficient of secondary school enrollment is negative and significant in the full and the OECD sample. returns to investment are positive and significant in all samples. Similarly. non-OECD and large market OECD countries. According to the pooled OLS results. Data include 20 OECD countries and 10 non-OECD countries for the period 1981–97. while the small market OECD countries have the lowest returns.19 17 See Appendix I. which takes higher values for lower risk of expropriation. except for the G-7. Returns to investment are positive and significant in all samples.04 percent). According to these results. In addition.24 in the high income OECD countries to 0. the coefficient of patent stock is positive and significant in all samples. except for the G-7. As seen from the table. presumably as a result of the small variation in this variable over time.06. it is statistically significant only in the low-income OECD countries. as indicated by the low p values of the Sargan test. Similar to the fixed effects results. the risk of expropriation and openness variables are significant with the expected signs in most of the samples. Returns to schooling are not significant in any of the samples. and the openness variable approximates the degree of trade liberalization. with a magnitude of around 0.06 percent) compared to the rest of the samples (0.20 percent in the OECD.17 percent in the non-G-7. Table 13 for OLS regression results. while the secondary school enrollment is not significant in any of the samples..24 of expropriation and openness variables are included in the analysis as well. 0.08. returns to investment are positive and significant in all samples. large market and high-income OECD countries have higher returns to their patent applications (0. non-OECD countries have the highest returns to their patent stock with a magnitude of 0. 18 19 . low income and small market OECD countries.18 Different from the fixed effects regression results.11. All other countries have around 0. and 0. with the magnitudes ranging from 0. The fixed-effects regression results are presented in Table 9.37 in the non-OECD countries. while returns to patent stock are significant only in the full. OECD and non-OECD samples. 0.07 percent increase in their per capita GDP as a result of a 1 percent increase in their patent stock.10 percent in the high income. As expected.25 percent in the G-7 countries. The reason that patent stock is not significant in the G-7 might be because of the small size of this sample. The risk of expropriation index captures the overall institutional development. the model for these samples suffers from the serial correlation between regressors and the residuals. Although the coefficient of secondary school enrollments has the expected sign and high t values in most of the samples. the coefficient of openness and risk of expropriation variables are positive and significant in most of the samples.

99 300 20 NonG-70.074 (4.42) 0.53) 0.94) 0.79) 0.097 (1. z statistics in parentheses.43) -0. Fixed Effects Regression Analysis of Per Labor GDP.04) 0. 2002).91) -0. corruption index (World Bank.140 (4.111 (6.288 (8. import/trade (OECD.063 (1. patent applications (NBER Patent Citation Database).99) 0..90) 0.098 (3. Notes: All variables are in natural logs and normalized by labor.86) 0.012 (0.071 (2.01) 0. trade (IMF Direction of Trade Database).24) 0.99 75 5 Large Market 0.39) 0. 1981–97 Sub-samples of OECD Full Sample Investment Second lag of patent stock Second lag of secondary school Openness Expropriation Risk R-squared Observations Number of countries 0.70) 0.86) 0.87) 0.86) 0.63) 0.S.009 (0.46) 0.39) -0.41) 0.071 (2.076 (4.274 (16. All regressions include time dummies. .99 135 9 Low Income 0.09) 0.54) -0.312 (24.127 (3.50) 0. Greece is not included in the analysis.30) 0. as it does not have data on imports in manufacturing sector.023 (0.67) 0.73) -0.80) 0.51) 0.47) 0.296 (7.198 (2.58) 0. 2002).44) -0.196 (2.268 (14.144 (4.010 (0.99 225 15 G-7 0.009 (0.66) 0.025 (1.99 135 9 Sources: R&D stock (OECD.99 449 30 NonOECD 0.72) 0.239 (9.99 149 10 OECD 0.75) 0.96) 0.99 135 9 High Income 0. U.29) 0.025 (0.72) 0.99 135 9 Small Market 0.49) 0.25 Table 9.025 (1. openness (PWT 6) employment (WEO.104 (9.018 (0.174 (2.262 (9. International Country Risk Guide).63) -0.43) 0.16) 0.18) 0.016 (0.084 (3.131 (4.111 (3.365 (17.044 (1.032 (0.254 (8.026 (0.059 (2. 2002).133 (3.025 (1.47) 0.60) 0.034 (1.040 (1.

76 0.192 (2.45) 1.20) 0.091 (2.05) -0.44) 0.36) 0.048 (1.113 (1.32 0.123 (4.16) 0.025 (4.026 (0.042 (0.75) 0.15) -0.253 (5.029 (0.34) 0.32 126 9 Small Market 0.79) 0.155 (6.010 (0.92) -0.135 (6.99 0. 1981–97 Sub-samples of OECD Full Sample Investment Second lag of patent stock Second lag of secondary school Openness Expropriation risk First lag of GDP Second lag of GDP Third lag of GDP Fourth lag of GDP Constant Sargan test1 (pvalue) AR(2) test2 (pvalue) NonOECD 0.057 (2.037 (1. 2002).51) 0.39) -0.94) --0.67) 0.173 (6.106 (1.006 (0.75) 0.056 (3.049 (4.42) 0.20) 0.001 (0.267 (2. .206 (11. All regressions include time dummies.99) -0.14) ---0.060 (0.042 (1.082 (2.083 (1.09) 0.227 (2.68) 0. 2/ H0: errors in first difference regression exhibit no second order serial correlation.044 (2.80 0.71) 0.004 (1.050 (2.17) -0.81) 0.702 (7.76) 0. General Methods of Moments (GMM) Regression Analysis of Per Labor GDP.019 (0.21) -0.98) 0.75) 0.188 (5.59) 0.35 210 15 G-7 0.03) 0.69) 0.58 126 9 0.57 359 30 Observations Number of countries Sources: GDP.037 (1.41) --0.85) -0.21) 0.31) 0.06) 0.35) -0.169 (2. All variables are log differenced once.032 (1.002 (1.012 (0.02 0.34) 0.15) 0.84 119 10 OECD 0.84) -0.542 (8.140 (4.51) 0. Notes: Absolute value of z statistics in parentheses.29) 0.42) 0.39) -0.029 (1.82) -0.740 (7.781 (8. employment (WEO.447 (7.059 (3. International Country Risk Guide). patent applications (NBER Patent Citation Database).050 (1.098 (4. and patent stock normalized by labor.033 (1.00) 0.212 (1. investment.54) 0.00 0.12 0. openness (PWT 6) expropriation risk index (World Bank.74) --0.36 126 9 High Income 0.048 (2.042 (1.038 (1.27) -0.654 (8.169 (5.94) 0.036 (1.002 (0.48 70 5 Large Market 0.42 280 20 NonG-70..49) -0.47) 0.055 (1.058 (4. investment.99) 0.14) 0.61) 0.89) -0.07) 0.155 (4.52) -0. GDP.013 (0.71) -0.38) --0.00 0.001 (0.30) 0.74) -0.13) 0.02) --0.006 (2.44) 0.048 (3.003 (1.036 (0.40) --0.82) 0.034 (0.268 (4.040 (3.10 0.90 126 9 Low Income 0.023 (0.194 (9.023 (1. 2002).14) 0. 1/ H0: regressors are not correlated with the residuals.081 (1.29) 0.05) 0.63) 0.26 Table 10.74) 0.069 (2. secondary school enrolments (WDI.10) -0.17) 0.60) 0.582 (6.484 (5.

VI. Our results show that there is a strong positive relationship between innovation (patent stock) and per capita GDP in both OECD and non-OECD countries. given that neither patent nor R&D data are complete measures of innovation and research activities. However. and knowledge stock. Spain. and Switzerland.27 Not surprisingly. Aghion and Howitt (1998). it lacks the evidence for constant returns to innovation in terms of R&D stock.20 and 0. Similarly. Netherlands. Specifically. coefficient of patent stock is significant only for non-OECD countries. The results for the full and the OECD samples are not conclusive as the model for these samples have been rejected by the sargan test. Second. As Table 11 shows. innovation. as suggested by our results. the relationship between total factor productivity (TFP) and innovation has been examined as well. The fact that only the large market OECD countries promote their innovation by investing in R&D provides support for the theories emphasizing the importance of market size for effective R&D sectors. i. and about a 0. they are not fully endogenous.e. the effect of trade liberalization on per labor GDP is positive and significant in all samples except for the G-7.11.. as postulated by R&D based endogenous growth models. large market OECD. Australia. this deserves some explanations.09 percent increase in the high income and low-income OECD countries. while our analysis lends support for endogenous growth theories in that it confirms a significant relationship between R&D stock and innovation. This implies that R&D models are not able to explain sustainable economic growth. and non-OECD countries. the coefficient of the expropriation risk index has the expected sign and significant in most of the samples with the highest magnitude in the G-7 and large market OECD countries. In addition. respectively. suggesting that in this sample innovation increases per capita GDP partly through its effect on TFP. which include the G-7. Additional to the analysis of the relationship between GDP and innovation. a 1 percent increase in the openness variable is associated with about a 0. are able to increase their innovation by investing in R&D. our results should not be interpreted as a rejection of R&D models. The results also suggest that the OECD countries that do not have effective R&D sectors seem to promote their innovation through technology spillovers from other OECD countries. while only the OECD countries with larger markets. 0. R&D models can still explain long term growth as long as there are constant returns to produced factors such as capital.14 percent increase in per labor GDP in the non-G-7 and the OECD countries with small markets. . First. and between innovation and per capita GDP. Acemoglu and Linn (2003). even there are diminishing returns to innovation in terms of R&D. CONCLUSION The objective of this paper was to assess whether there is a significant relationship between countries’ R&D efforts and their innovation and between innovation and per capita income.

00 0.035 (2.789 (19.876 (22.66) 0. 1/ H0: regressors are not correlated with the residuals. 2002).29) -0.726 (12.59) 0.10) 0. International Country Risk Guide). Notes: Absolute value of z statistics in parentheses. employment (WEO.28 Table 11.74 280 20 Non-OECD 0.021 (1.019 (1. All regressions include time dummies.40) 0. patent applications (NBER Patent Citation Database).4 log (investment) .0. secondary school enrolments (WDI. which are assumed to be 0.005 (0.28 0.060 (2.6 log (employment).012 (1.50 139 10 0.13) 0.001 (0. 2/ H0: errors in the first difference regression exhibit no second order serial correlation.67) 0.016 (1.67) 0.31) 0.006 (0.029 (1.53 419 30 Observations Number of countries Sources: GDP. respectively.038 (3.64) 0. TFP (author’s calculation: TFP = log (GDP) .28) -0.000 (0.4.. openness (PWT 6) expropriation risk index (World Bank.08 0.53) 0.0.001 (1.54) 0.025 (0.11) -0.00) -0.25) 0. investment.001 (0.6 and 0. 2002). All variables are log differenced once. Patent stock is normalized by labor.25) 0. 1981–97 Full First lag of TFP Second lag of patent stock Second lag of secondary school Openness Expropriation risk Constant Sargan test1 (p-value) AR(2) test2 (p-value) OECD 0. .58) 0. TFP is calculated by subtracting the log levels of labor and investment multiplied by their shares of income. General Methods of Moments (GMM) Regression Analysis of Total Factor Productivity (TFP).

99 135 9 Low Income 0.99 120 8 High Income 1.34) 0.37) 0.11) -0.031 (0.624 (2.171 (4.923 (9.073 (1.413 (3. 1981–97 Full Initial patent flows Second lag of R&D stock Second lag of Secondary school Expropriation risk Index Manufacturing import/trade U.85) 0.039 (0.298 (4.118 (27.046 (26.11) -0. 2002).021 (2.32) -0.572 (11.480 (2.109 (2.640 (9.154 (3. expropriation risk index (World Bank.26) 0.291 (3.67) -0.96) 0.93) -0.73) -0.460 (6.428 (5. All regressions include time dummies.12) -0.19) -0.178 (1. All variables are in natural logs and normalized by labor. as it does not have data on imports in manufacturing sector.229 (2.82) 0.31) 0. 2002).15) 0. U.30) 0.66) -2.99 120 8 Sources: R&D stock (OECD..99 285 19 Non-G-7 1.076 (0. trade (IMF Direction of Trade Database).05) 1.S.464 (3.42) 0.748 (2.194 (3.96) 2. import/trade (OECD.706 (3.056 (14.32) -0.228 (1.951 (1.10) 0.133 (2.08) 0.238 (0.17) 0.02) 0. .153 (2.800 (3.24) 0.23) 0.74) 0.99 135 9 Small Market 1.81) 0. patent applications (NBER Patent Citation Database).034 (0.32) 0.49) 0.161 (0. trade/GDP Openness R squared Observations Number of countries 1.079 (0.366 (1.S.272 (1.511 (1.99 75 5 Large Market 0. openness (PWT 6) employment (WEO.99) 0.26) 0.33) -0.981 (0.32) -0.11) 0.112 (1.99 210 14 G-7 0.48) 0.10) -0.56) 0.010 (0. International Country Risk Guide).438 (2.36) 2. Ordinary Least Square (OLS) Regression Analysis of Per Labor Patent Flows.42) 0. Greece is not included in the analysis.48) 0. Notes: Heteroskedasticity corrected z statistics are in parentheses.40) 0.486 (1.699 (0.29 I.20) 0. 2002). Regression Tables APPENDIX I Table 12.64) 0.05) 0.72) -0.253 (2.322 (0.343 (1.65) 1.193 (19.

58) -0.30) 0.12) 0.95) 0.99 300 20 0.695 (11. and all regressions include time dummies.215 (5.06) -0.13) -0.46) 0.022 (3.77) 0.93) -0.02) 0.99 135 9 0.99 449 30 Sources: GDP.223 (8.60) 0. All variables are in natural logarithm.77) 0.89) 0.101 (2.010 (0. secondary school enrolments (WDI.012 (0.784 (4.015 (0.060 (1.070 (5.90) 0.99 135 9 0. .482 (23.47) -0.104 (2.752 (9.013 (1.597 (12.039 (1.19) 0.70) 0.58) 0.054 (0.013 (1.008 (0.003 (0.239 (7.99 75 5 0.90) 0. Notes: z statistics in parentheses.039 (1.99 135 9 0.17) -0. GDP.243 (11.12) 0.031 (0. employment (WEO.037 (1.49) 0.26) 0.99 225 15 0.204 (8.03) -0.008 (0.018 (1.46) 0.009 (0.44) 0.18) 0..58) 0.29) 0.002 (0.41) 0.30 - APPENDIX I Table 13.046 (1.89) 0.16) 0.91) 0.020 (3.086 (2.021 (1. 2002). investment.99 135 9 Investment Second lag of patent stock Second lag of secondary school Openness Expropriation -0.049 (3.94) 0.479 (6.11) 0. Ordinary Least Square (OLS) Regression Analysis of per Labor GDP.48) 0.276 (8.408 (12.011 (0.13) 0. 2002).111 (4.52) R-squared Observations Number of countries 0.08) 0.029 (1. patent applications (NBER Patent Citation Database).12) -0.94) 0.99) 0. International Country Risk Guide).295 (23.96) 0.43) 0. openness (PWT 6) expropriation risk index (World Bank.114 (2.048 (1.705 (25. 1981-97 Sub Samples of OECD Full Sample Initial GDP NonOECD OECD NonG-7G-7 Large Market Small Market High Income Low Income 0.075 (6. investment and patents are normalized by employment series.018 (0.99 149 10 0.06) 0.66) 0.33) 0.178 (2.29) -0.620 (9.34) 0.88) 0.27) 0.254 (13.65) 0.869 (8.060 (0.73) -0.12) 0.009 risk (0.

17* 0.34* 0.23* R&D stock 1 0.01 Investment Patent stock Import/ Secondary Expropriati US trade/ Openness trade in school on risk GDP manufacturing 1 0.04 1 0. GDP.17* 0.30* -0.48* -0.08 -0. patents and R&D stock are normalized by labor.24* 0.88* 0.09 0.51* 0.29* 0.43* 1 -0.34* 0.03 0. Pairwise Correlation Table1 Full Sample GDP Investment Patent stock R&D stock Secondary school Openness Expropriation risk Import/trade in manufacturing US trade/GDP GDP 1 0.35* 0.33* 0.06 1 OECD Sample GDP Investment Patent stock R&D Stock Secondary school Openness Expropriation risk Import/trade in manufacturing US trade/GDP GDP 1 0.22* -0.86* 0.09* 1 0. openness (PWT 6) expropriation risk index (World Bank.05 0.37* 0.58* 0.18* 1 0. secondary school enrolments (WDI.53* 0.45* -0.44* 0.02 0.97* 0.72* 0.24* 0. patent applications (NBER Patent Citation Database).18* 0.97* 0.81* 0.84* 0. Imports /trade in manufacturing sector (OECD.49* 0.64* 1 -0.70* 0.23* -1 Sources: GDP.25* ----- 1 0.22* 0.086 1 0.22* 1 0.24* 0. Correlation Table and Summary Statistics Table 14.76* 0.06 -0. investment.14* 1 0.70* 0.20* 0.45* 0.55* 0.06 0.40* 0.35* 0.16* 0.14* 0.37* 0. investment.50* -0. 2002). 1/ All variables are in natural logs.06 1 0.90* 1 0.31 - APPENDIX II II. 2002).26* -0.10* Investment Patent stock R&D stock Import/ Expropriati US trade/ Secondary Openness trade in on risk GDP school manufacturing 1 0.05 0.29* 0.23* R&D stock 1 0.35* 0.36* 0.65* 0.20* Investment Import/ Secondary Expropriati US trade/ Openness trade in school on risk GDP manufacturing 1 0.70* 0.26* -0.41* 0.43* -0.06 1 Non-OECD Sample GDP Investment Patent stock Secondary school Openness Expropriation risk US trade/GDP GDP 1 0.41* 0.03 -0.76* 0.30* Patent stock 1 0.57* -0.55* 0.13* 1 0.24* 0.30* -0.30* 1 0. Note: (*) significant at 10 percent significance level. employment (WEO.76* 0.29* 0..92* 0.26* 0.01 0. International Country Risk Guide).92* 0. 2002). .

80 21.00 0. 2002).37 27.09 10.45 0.19 76.02 0.55 0.13 Sources: GDP.95 11.45 1.65 0.01 21.95 10.09 427.00 0. openness (PWT 6) expropriation risk index (World Bank. employment (WEO.50 0.25 0.00 0.84 9.01 Min 17504 3249 0.20 15.38 0. 2002).66 1.87 3. patent applications (NBER Patent Citation Database).55 Max 83442 24888 430 2780 157.61 0.52 Max 52752 20820 71 97.81 12. 2002).92 4.07 Std Dev 12868 4272 12 17. Note: GDP and investment are in per labor.00 0.09 Mean 50583 10084 103 777 100.08 112.95 49.44 64.95 11.55 Mean 38018 8085 69 776 85.25 0. secondary school enrolments (WDI..01 65.64 427.15 62.93 0.51 0.00 0.81 12.08 Std Dev 22330 4985 88 485 27 71 1.05 Median 50576 9123 80 701 99.87 3.02 Max 83442 24888 429 2780 157. .86 100.25 0. import/trade in manufacturing sector (OECD.06 Mean 14674 4063 6 56.61 9.00 0.32 - APPENDIX II Table 15. investment.74 0. Summary Statistics of the Variables Full Sample GDP Investment Patents R&D Expenditure Secondary school enrolments Openness Expropriation risk US Trade/GDP OECD GDP Investment Patents R&D Expenditure Secondary school enrolments Openness Expropriation risk Import/trade in manufacturing US trade/GDP Non-OECD GDP Investment Patents Secondary school enrolments Openness Expropriation risk US trade/GDP N 523 509 523 340 523 509 523 523 N 340 340 340 340 340 340 340 323 340 N 183 169 183 183 169 183 183 Min 505 251 0.09 146. International Country Risk Guide).10 Std Dev 15046 4015 92 485 17.01 Min 505 251 0.91 57.67 8.45 7.00 0.2 66 41.90 8. Patents and R&D stock are in per million-labor.04 Median 10764 2390 1 58.13 Median 40166 7819 39 700 90.

Table 17.576 Min -0.049 0.011 0.263 1.792 0.549 Max 0.046 0.001 Median 0.004 Median 0.031 0.101 0.022 0.047 0.010 -0. employment (WEO.792 0.013 0.016 0.577 0.001 -0.026 0.059 0. openness (PWT 6) expropriation risk index (World Bank. secondary school enrolments (WDI.022 0.452 -0.021 0.962 -0.288 0.403 -1.010 0.146 0.018 0.486 0.007 0.134 -0.455 0. 2002).031 -0.014 0.017 0.263 1.000 -0.132 0. Note: All variables are log differenced. GDP and investment are in per labor.013 0.023 0.102 0. patent applications (NBER Patent Citation Database).079 -0.030 0.017 0.962 -0.073 0. investment.425 -0. 2002).003 Mean 0.303 -0.036 0.018 0.33 - APPENDIX II Table 16.584 -0.023 0.452 -0.015 0.038 0.008 0.157 -0.558 0. International Country Risk Guide). Summary Statistics of the Growth Rates of the Variables Full GDP Investment Patents R&D Expenditure Secondary school enrolment Openness Expropriation risk US trade/GDP OECD GDP Investment Patents R&D Expenditure Secondary school enrolment Openness Expropriation risk Import/trade in manufacturing US trade/GDP Non-OECD GDP Investment Patents Secondary school enrolment Openness Expropriation risk US trade/GDP N 492 479 492 320 492 479 493 492 N 320 320 320 320 320 320 321 304 320 N 172 159 172 172 159 172 172 Min -0.157 -0.121 0.155 Sources: GDP.018 0.030 0.584 -0.357 0.122 0.076 0.021 0.361 0.019 0.122 0.024 0.085 0. Patents and R&D stock are in per million-labor.068 0.000 0.010 Median 0.002 Mean 0.455 Max 0.576 Max 0.239 -1.044 0.549 Mean 0.307 0.425 -0.024 0.022 0.145 0.022 0.006 0. import/trade in manufacturing sector (OECD.452 Min -0.361 0.040 0.036 0.135 0.145 0. 2002). Country Codes AU: Australia AT: Austria BE: Belgium CA: Canada CH: Switzerland DK: Denmark ES: Spain FI: Finland FR: France GB: Great Britain GR: Greece IE: Ireland IS: Iceland IT: Italy JP: Japan NL: Netherlands NO: Norway NZ: New Zealand PT: Portugal SE: Sweden ..116 0.558 0.065 0.020 0.403 -1.455 0.014 0.152 -0.001 0.100 -0.022 0.147 Std Dev 0.001 Std Dev 0.134 -0.009 0.486 0.000 -0.435 0.142 Std Dev 0.235 1.065 0.109 -0.

” (Cambridge. “North-South R&D Spillovers. “Endogenous Growth Theory.” NBER Working Paper 10038. 29. Vol. “Some Tests for Specification of Panel Data.” Econometrica. United Kingdom: The Institute for Fiscal Studies). 1984. “Market Size and Innovation: Theory and Evidence from the Pharmaceutical Industry.” Journal of Political Economy.” Economics Letters Vol 7. 1991. l8387. “R&D. Human Capital and International Technology Spillovers: A Cross Country Analysis. (Cambridge. No. Griliches. Massachusetts: National Bureau of Economic Research). 501-22. R&D and Patents. Vol 77. Acemoglu.M. 725-29. “Hybrid Corn: An Exploration in the Economics of Technological Change.34 References Aghion.” The Institute for Fiscal Studies. Zvi. 155-73. 2000. Manuel. (Cambridge. 25. and Alexander W. pp. “Mapping the Two Faces of R&D: Productivity Growth in a Panel of OECD Countries. and Stephen Bond. 1998. No. 1. Griliches. pp. 1962. pp. “Market Value. Philippe. 1981. Arellano. 392-98. 02/00 (London.. “Patent Statistics as a Measure of Technical Change. Elhanan Helpman. pp. pp. Jacob Schmookler. 58. Zvi. 4.. Zvi.” Review of Economic Studies. “Interindustry Technology Flows and Productivity Growth: A Reexamination. “Inventing and Maximizing. 3. Frank Lichtenberg.. 277–97. Hoffmaister. 1963. Vol. Massachusetts: National Bureau of Economic Research). . pp. and John Van Reenen. pp. 1995. “The Economic Implications of Learning by Doing. Vol. 53 No. 57-75. Daron. Arrow.. 5048.” Review of Economics and Statistics. 2003. Griffith Rachel. Working Paper. Griliches. 2001.” NBER Working Paper. Zvi. 1969. 1957. No. Massachusetts: The MIT Press). Scherer. pp. Joshua Linn. Coe. 66. F. Griliches. Comanor. Vol. Frantzen. Vol. and Peter Howitt.” Review of Economic Studies. 102 No. Kenneth J. David T.” Scandinavian Journal of Economics.” The American Economic Review. William S. Stephen Redding. Dirk. 324–29.

pp. 1. Vol. 7891. Jr.E. 1–24. “On the Mechanics of Economic Development. No. Charles I. 141–54. Adam B.” Development and Comp Systems. Hall. Economics Working Paper Archive. 2002..3–42.” American Economic Review. “Productivity.27. 2002.” Journal of Economic Literature.” American Economic Review.35 Griliches. Insights and Methodological Tools. “Time Series Test of Endogenous Growth Models. “Real Effects of Academic Research. “On Patents. and the Data Constraint. “Technological Opportunity and Spillovers of R&D: Evidence from Firms’ Patents.” NBER Working Paper 8498. ———.. (Cambridge: Massachusetts: National Bureau of Economic Research). ———. 9–21. Michael E. 76. ———. Ariel.” Journal of Econometrics. “Patent Statistics as Economic Indicators: A survey. Porter. R.” Journal of Monetary Economics. 1989. “Measuring the ‘Ideas’ Production Function: Evidence from International Patent Output.. Lucas. Bronwyn H. 390–409. “Patents and R&D: Is There a Lag?” International Economic Review.. Vol. 1986. pp. 108. Vol.” Journal of Political Economy. Bronwyn H.” NBER Working Paper No. “The NBER Patent Citation Data File: Lessons. 1994. pp. Vol. 984–1001. Pakes. “Growth Empirics: A Panel Data Approach. ———.” American Economic Review. xxviii. Islam. C. “Productivity. (Cambridge. and J. Vol. and the Stock Market Rate of Return. 495–525. 1988. 2001.. 84. Vol. 1985. Levin. 2000. Chu. pp. Nazrul. 4. 1661-1707. pp. and Manuel Trajtenberg. pp.. 1986. Massachusetts: National Bureau of Economic Research). 76. Jones.. 957–70. pp. pp. 79.” Quarterly Journal of Economics. Lin.1. Hausman. Vol. Jaffe.” American Economic Review. Vol. R&D. 1127-70.” Quarterly Journal of Economics.165–283. pp. A. CX No.. R&D and Basic Research at the Firm Level in the 1970s. Profits and Market Value. pp. pp. Zvi. 1995. “Market Value and Patent Citations: A First Look. R&D. and Scott Stern. Vol. . Adam Jaffe. Hall. No. and Jerry A. 22. “Unit Root Tests in Panel Data: Asymptotic and FiniteSample Properties. 1990.93. Vol. 1986. Zvi Griliches. 1995b.

No.” (forthcoming...” The American Economic Review. and Dry Holes. pp.” Journal of Political Economy.. Vol. Lags. 1999.M. F. Vol. 1994. Innovation.1002–37. Schmookler. 8. Vol 36. 2. Andreas. Scherer. “Issues in R&D-Productivity Dynamics: Causality. 52. Marios. pp. ———. Vol. Jacob and Oswald Brownlee. pp. Finland: The Research Institute of Finnish Economy).” The Review of Economics and Statistics. pp. Marios Zachariadis. 2003. 566-686.” Journal of Economic Perspectives. 627-34 Zachariadis. Vol.36 Romer. 165-76.” Journal of Political Economy. Savvides. 3–22. 1962. and Technological Progress: A test of the Schumpeterian Framework without Scale Effects. 1986. 94. . 1982 “Inter-Industry Technology Flows and Productivity Growth. 64. pp. Paul M. 2003 “R&D. “The Origins of Endogenous Growth.” Canadian Journal of Economics. 3. Oklahoma: Oklahoma State University. “Increasing Returns and Long Run Growth. pp. “International Technology Diffusion and TFP Growth. 71–102. ———.” (Helsinki. Petri. 1990. Rouvinen. “Endogenous Technical Change. “Determinants of Inventive Activity. No. 98. Department of Economics). Vol.