R&D Expenditure and Economic Growth: New Empirical Evidence

Abstract

This article provides an empirical analysis of the relationship between research and development (R&D) expenditures and economic growth, and determines whether this relationship differs with respect to the degree of development. In this regard, the study utilises data from 52 countries from 1996 to 2010 and employs a dynamic panel data model. The research finds that R&D expenditure has a positive and significant effect on economic growth for all countries in the long run, which is consistent with the relevant literature. For developing countries, the effect is weak in the short run but strong in the long run, as expected. The study adds new empirical evidence to the literature.

JEL Classifications: E24, O11, O32, O47

Keywords

R&D Expenditure Economic Growth Dynamic Panel ECM Developing Countries Developed Countries

1. INTRODUCTION

An increase in economic growth yields wealth, on average, to everyone. Therefore, factors effecting economic growth are important. One such factor is research and development (R&D) expenditure. The effect of R&D expenditure on economic growth has been the subject of many studies, and the outcomes from these studies are varied. Theoretical and empirical works of literature on economic growth have drawn our attention to the fact that investments in R&D are one of the most important factors of sustainable economic growth and, with increasing innovation, have a positive effect on productivity. As a matter of fact, countries that give priority to R&D activities have been producing higher added value and obtain better economic performance. The real question is whether this positive relationship between R&D expenditure and growth exists when we use a large sample of 52 countries and whether the results differ when levels of development in countries in the sample differ.

Therefore, the present study has two objectives. The first objective is to investigate the relationship between R&D expenditures and economic growth by utilizing data from 52 countries over the period 1996–2010. The second objective is to determine whether the results obtained differ with respect to the economic development of these countries. There are many studies that investigate the relationship between R&D expenditure and economic growth, but none have used panel cointegration tests employing mean group and pooled mean group estimators proposed by Pesaran, Shin and Smith (1999). This article attempts to fill this gap. This article estimates short- and long-term coefficients of the R&D effect on growth, using panel cointegration techniques in a large panel of countries. It is distinct from previous studies since it (a) applies panel unit root and panel cointegration tests and (b) uses a large sample of developing and developed countries. The results we obtain in this article are consistent with the theoretical and empirical findings in the literature. Our results provide new empirical evidence in terms of covering a large number of countries. The rest of article is organized as follows: the literature on the relationship between economic growth and R&D expenditure is discussed in Section 2. The methodology and research hypotheses are provided in Section 3. In Section 4, data and findings are presented, and the last section offers concluding remarks.

2. THE LITERATURE

The activities of entrepreneurs in relation to innovation enable the development of a variety of new ideas, production of new consumption goods and emergence of new markets (Pessoa, 2010: 153). Indeed, Schumpeter argues that in a capitalist economic system, activities performed within the scope of innovation have a direct effect on economic growth (2003: 83). Similarly, Romer (1986, 1990) defines technological change as the primary component of economic growth, and he bases endogenous growth theory primarily on investment in R&D capital. Romer considers that R&D activities generate knowledge that prevents decreasing returns to scale to occur for capital as a factor of production. Grosman and Helpman (1990) argue that R&D expenditures are vital for economic growth. In their empirical study, Aghion and Howitt (1992) employ US data and find that the share of R&D expenditure in gross domestic product (GDP) affects economic growth. Lichtenberg (1992) utilizes data for the period 1964–89 from 74 countries to examine R&D expenditures’ effect on economic growth from the perspectives of the public and private sectors. He finds that although there is a positive effect of R&D expenditure on growth, private sector R&D expenditures are found to be more efficient and effective compared with public sector expenditures. Similarly, Van Pottelsberghe and Guellec (2004) employ 16 Organisation for Economic Co-operation and Development (OECD) member countries’ data covering the period 1980–98 and find a positive and significant effect of R&D expenditure on economic growth regardless of expenditure types, whether originating from the private, public, domestic or foreign sectors. Ülkü (2007) also utilizes data from 17 OECD countries in various sectors and finds that the R&D concentration in the chemicals, electrical and electronics and drugs and medicine sectors determines the innovation potential in these sectors, while innovation potential in turn has a positive effect on output in the investigated sectors.

There are many empirical studies in the literature on this issue and each of these studies focuses on a different aspect. We have, therefore, grouped selected studies and mention the important ones here. While Goel and Ram (1994) use data from a cross-section of 52 countries, Gittleman and Wolff (1995) use time-series of cross-country data to find out whether levels of development of a country make a difference to the effect of R&D expenditure on economic growth. Both studies find that the degree of development of a country makes a difference by speeding up economic growth from R&D expenditure.

There are also studies that use OECD data to empirically investigate the relationship between R&D expenditure and economic growth. Among them Sylwester (2001), Saraç (2009), Freire-Seren (1999) and Gülmez and Yardımcıoğlu (2012) are worth mentioning. Sylwester (2001), for instance, examines 20 OECD countries and uses the aggregate data employing a multivariate regression. He finds a statistically meaningful relationship between R&D expenditure and economic growth when only G-7 countries of a sub-sample are considered. The full sample finds no association. On the other hand, Saraç (2009), Freire-Seren (1999) and Gülmez and Yardımcıoğlu (2012) find a positive association. In his panel data analysis covering data for the period 1983–2004 from 10 developed OECD countries, Saraç (2009) finds that R&D expenditures have a positive effect on economic growth. Similarly, Freire-Seren (1999) uses data for the period 1965–90 from 21 OECD countries and concludes that a 1 per cent increase in aggregate R&D expenditure raises the GDP growth by 0.08 per cent. A similar study was undertaken by Gülmez and Yardımcıoğlu (2012) who employ data from 21 OECD member countries and find that a 1 per cent increase in R&D expenditures may, in the long run, increase economic growth by 0.77 per cent.

Two studies use developing countries’ data in searching the association between R&D expenditure and economic growth. Samimi and Alerasoul (2009) study 30 developing countries’ data covering the period 2000–06 and claim that R&D expenditures have no direct effect on economic growth. They note that the underlying reason is insufficient resources allocated to R&D activities in the mentioned countries (Samimi & Alerasoul, 2009: 3469). On the contrary, Göçer (2013) finds a positive association between R&D expenditure and economic growth in developing countries. He uses 11 developing Asian countries’ data covering the period 1996–2012 and finds that the contribution of a 1 per cent increase in R&D expenditures increases modern technology goods exports by 6.5 per cent and information-communication technology exports by 0.6 per cent and accelerates economic growth by 0.43 per cent.

One important study looks also at the externalities of R&D expenditure between developed and developing countries. Coe, Helpman and Hoffmaister (1997) estimate the spill-over of R&D expenditure in developed countries to developing countries. They claim that in developed countries, R&D expenditures positively affect not only national economies but also the economies of developing countries and that high-tech goods exported from developed countries to developing countries and capital goods fuel an efficiency boost in developing countries (Coe et al., 1997: 134). They conclude that an increase in R&D expenditure in industrial countries positively and substantially affects outputs of developing countries. Keller (1998) reinvestigates the effect of R&D spending on total factor productivity (TFP) growth in developing countries. His study raises some doubts about the finding of Coe et al. (1997).¹

These findings maintain that to ensure sustainable economic growth performance, it is necessary to allocate more resources to R&D activities.

3. METHODOLOGY

In the study, panel error correction methods are employed to determine the effects of R&D expenditures on economic growth in the short term and long term. In this context, first, the stability of the data set is examined by unit root tests (Fisher-augmented Dickey–Fuller [Fisher-ADF] test and Fisher-PP test; Breitung, 2000; Hadri, 2000; Harris & Tzavalis, 1999). Harris and Tzavalis, Breitung and Hadri-Lagrange multiplier (LM) tests are based on the assumption that the panel units have common autocorrelation parameters. However, the Fisher-ADF and Fisher-PP tests are based on the assumption that the panel units have different autocorrelation parameters.

In the analysis process, four error-correction-based panel cointegration tests proposed by Westerlund (2007) are used for the detection of a cointegrated relationship in the panel data sets. Two of these are called group mean statistics (Gα and Gτ). The null hypotheses of these tests are that ‘there is no cointegration between the variables for all units’. If the null hypothesis is rejected, there is the presence of cointegration for at least one cross-sectional unit in the panel. The other two tests, which reflect the panel overall, are called panel statistics (Pα and Pτ). The null hypotheses of these tests are that ‘there is no cointegration between the variables for all units.’ However, if the null hypothesis is rejected, there is the presence of cointegration for the panel as a whole.

The relationship between the two variables in the short term and long term can be determined with error correction models (ECM), which are cointegrated. The instruments within the ECM reveal relationships between the examined series in the short and long term (Pesaran et al., 1999: 622). The mean group estimator (MGE) developed by Pesaran and Smith (1995) assumes that panel units are heterogeneous in the short and long terms. Thus, the MGE provides estimates according to all units in the short and long term. However, the pooled mean group estimator (PMGE), developed by Pesaran et al. in 1999, assumes that panel units are heterogeneous only in the short term; they are homogeneous in the long term. Hence, the PMGE provides estimates according to all units in the short term, but in the long term it forecasts a common parameter for all units.

The Hausman specification test can be used to choose between the MGE and the PMGE estimator. The null hypothesis of the Hausman test is that there is no systematic difference between the long-term coefficients of the MGE and PMGE. If the null hypothesis is rejected, the MGE will be preferred because panel units are heterogeneous both in the short run and in the long run. In contrast, if the null hypothesis cannot be rejected, the PMGE will be preferred (Tatoğlu, 2012: 255).

In the study, the basic hypothesis, which has been shaped by the model of endogenous growth and other empirical investigations in the literature, is that ‘There is a significant relationship between R&D spending and economic growth.’ This hypothesis formulates with the following functional equation:

ΔGDP = f (ΔR&D ​ Expenditure, ​ Other ​ Factors)

We have tested this functional relationship with the econometric model which has been estimated on panel data for selected developing and developed countries in logarithmic form.

This is as follows:

ln (G D P_{i t}) =_{0} +_{1} ln (R & D_{i t}) + U_{i t}

(1)

where

i: 1,........, 52(all countries) t:1,........,15(1996–2010)

i: 1,........, 32(developed countries) t:1,........,15(1996–2010)

i: 1,........, 20(developing countries) t:1,........,15(1996–2010)

for i = 1,… . ., N; t = 1,…….,T, where N refers to the number of countries in the panel and T refers to number of observations over time.

Assuming that lnGDP and lnR&D are I(1) and cointegrated, the residual terms should be I(0) for all i, and they are independently distributed across t. The maximum lag length for lnGDP and lnR&D is confirmed by the Akaike Information Criterion (AIC) and Schwarz Bayesian Criterion (SBC), which is one lag. So the autoregressive distributed lag, ARDL (1, 1) model proposed by Pesaran et al. (1999) is as follows:

\ln G D P_{i t} = γ_{i} + λ_{i} l n G D P_{i, t - 1} + δ_{0 i} l n R & D_{i t} + δ_{1 i} l n R & D_{i}_{, t - 1}^{} + Є_{i t}

(2)

and the equilibrium error correction representation of the error correction equation is as follows:

Δ \ln ​ G D P_{i t} = ϕ_{i} (\ln ​ G D P_{i, t - 1} - {\hat{θ}}_{0 i} - {\hat{θ}}_{1 i} \ln ​ R & D_{i, t - 1}) + δ_{1 i} \ln ​ R & D_{i, t} + Є_{i t}

\begin{matrix} where ϕ_{i} = - ​ (1 - λ_{i}), {\hat{θ}}_{0 i} = \frac{γ_{i}}{1 - λ_{i}}, {\hat{θ}}_{1 i} = \frac{δ_{0 i} + δ_{1 i}}{1 - λ_{i}} . \end{matrix}

(3)

Є_it

4. THE DATA SET AND EMPIRICAL RESULTS

A total of 52 countries form the sample in this study. Among them 32 countries are developed countries and 20 are classified as developing countries.² We use annual data covering the period 1996–2010. Data on GDP per capita and R&D expenditure per capita are taken from World Bank, World Development Indicators (WDI), United Nations International Children’s Emergency Fund (UNICEF) statistics and the Statistical Office of the European Union (Eurostat). Due to the unavailability of data we use R&D expenditure and GDP at current US$ in the study.

4.1 Results of the Panel Unit Root Tests

Table 1 presents the results for all the variables in the six panel unit root tests. According to the results, lnGDP and lnR&D are non-stationary in level for all samples. The results of the tests demonstrate that the null hypothesis of the unit roots for the panel data cannot be rejected in I(0).³ However, when the first differences are taken, the null hypothesis is rejected. The results show that the lnGDP and lnR&D series are stationary in the first differences (p < 0.01).⁴

4.2 Results of the Panel Cointegration Test

The cointegrated relationship between the variables is tested using the Westerlund tests because both series are found to be integrated of order one. Table 2 reports the Westerlund cointegration test results. According to the results, the null hypothesis is rejected by the G_τ, P_τ and P₁ tests for all panels of developed countries and developing countries. For all panels, the G_τ statistic rejects the null hypothesis of no cointegration at the 5 per cent level; the P_τ and P_α statistics reject it at the 1 per cent level. For developed countries, the G_τ statistic rejects the null hypothesis at the 10 per cent level; the P_τ and P_α statistics reject it at the 5 per cent level. For developing countries, the G_τ, P_τ and P_α statistics reject the null hypothesis at the 1 per cent significance level. Because the null hypotheses are rejected, we conclude the presence of cointegration for the panel as a whole.

Table 1
Panel Unit Root Tests

H-T
Breitung
Hadri LM
ADF—Fisher
PP—Fisher

Statistic P Value Statistic P Value Statistic P value Statistic P value Statistic P Value

I(0) lnGDP (all countries) 0.6576 0.9963 3.1532 0.9992 24.5840* 0.0000 –3.5712 0.9998 –3.5712 0.9998

lnR&D (all countries) 0.6259 0.9657 1.1177 0.8681 24.5120* 0.0000 –2.0399 0.9793 –2.0399 0.9793

lnGDP (developed countries) 0.6041 0.8323 2.1810 0.9854 14.9794* 0.0000 –2.6205 0.9956 –2.6205 0.9956

lnR&D (developed countries) 0.5868 0.7244 1.1210 0.8689 16.2057* 0.0000 –2.0016 0.9773 –2.0016 0.9773

lnGDP (developing countries) 0.6967 0.9898 2.2855 0.9889 17.8170* 0.0000 –2.4438 0.9927 –2.4438 0.9927

lnR&D (developing countries) 0.6469 0.9310 0.3832 0.6492 16.4730* 0.0000 –0.7574 0.7756 –0.7574 0.7756

I(1) lnGDP (all countries) 0.2784* 0.0000 –4.3069* 0.0000 –5.0678 0.4234 4.6501* 0.0000 4.6501* 0.0000

lnR&D (all countries) 0.1883* 0.0000 –4.9503* 0.0000 –4.7776 0.3891 10.4788* 0.0000 10.4788* 0.0000

lnGDP (developed countries) 0.2914* 0.0000 –2.9254* 0.0017 –3.9616 0.3411 3.2989* 0.0005 3.2989* 0.0005

lnR&D (developed countries) 0.2436* 0.0000 –2.9213* 0.0017 –3.9242 0.1922 4.2519* 0.0000 4.2519* 0.0000

lnGDP (developing countries) 0.2486* 0.0000 –3.4217* 0.0003 –3.1519 0.1257 3.3253* 0.0004 3.3253* 0.0004

lnR&D (developing countries) 0.1445* 0.0000 –4.5831* 0.0000 –2.8884 0.2827 11.5183* 0.0000 11.5183* 0.0000

		H-T	Breitung	Hadri LM	ADF—Fisher	PP—Fisher
I(0)	lnGDP (all countries)	0.6576	0.9963	3.1532	0.9992	24.5840*	0.0000	–3.5712	0.9998	–3.5712	0.9998
	lnR&D (all countries)	0.6259	0.9657	1.1177	0.8681	24.5120*	0.0000	–2.0399	0.9793	–2.0399	0.9793
	lnGDP (developed countries)	0.6041	0.8323	2.1810	0.9854	14.9794*	0.0000	–2.6205	0.9956	–2.6205	0.9956
	lnR&D (developed countries)	0.5868	0.7244	1.1210	0.8689	16.2057*	0.0000	–2.0016	0.9773	–2.0016	0.9773
	lnGDP (developing countries)	0.6967	0.9898	2.2855	0.9889	17.8170*	0.0000	–2.4438	0.9927	–2.4438	0.9927
	lnR&D (developing countries)	0.6469	0.9310	0.3832	0.6492	16.4730*	0.0000	–0.7574	0.7756	–0.7574	0.7756
I(1)	lnGDP (all countries)	0.2784*	0.0000	–4.3069*	0.0000	–5.0678	0.4234	4.6501*	0.0000	4.6501*	0.0000
	lnR&D (all countries)	0.1883*	0.0000	–4.9503*	0.0000	–4.7776	0.3891	10.4788*	0.0000	10.4788*	0.0000
	lnGDP (developed countries)	0.2914*	0.0000	–2.9254*	0.0017	–3.9616	0.3411	3.2989*	0.0005	3.2989*	0.0005
	lnR&D (developed countries)	0.2436*	0.0000	–2.9213*	0.0017	–3.9242	0.1922	4.2519*	0.0000	4.2519*	0.0000
	lnGDP (developing countries)	0.2486*	0.0000	–3.4217*	0.0003	–3.1519	0.1257	3.3253*	0.0004	3.3253*	0.0004
	lnR&D (developing countries)	0.1445*	0.0000	–4.5831*	0.0000	–2.8884	0.2827	11.5183*	0.0000	11.5183*	0.0000

Source: Authors’ estimation.

Notes: * and ** indicate levels of significance at 1 per cent and 5 per cent, respectively.

4.3 Results of the Error Correction Model

The MGE and PMGE are utilized to identify the direction of the relationship between the cointegrated variables in the short and long terms. Table 3 presents the results of the MG and PMG estimations of the short-run and long-run coefficients of lnR&D and the convergence parameter (adjustment coefficient).

Table 2
Error Correction-based Cointegration Test Results

Entire Panel Statistics Z value P value

G_τ –2.028** –1.908 0.028

G_α –7.870 –0.890 0.187

P_τ –15.72* –5.021 0.000

P_α –3.732* –6.730 0.000

Developed Countries

Statistics

Z value

P value

G_τ –2.022* –1.457 0.073

G_α –7.335 –0.140 0.444

P_τ –10.05 –1.682 0.046

P_α –5.782** –1.758 0.039

Developing Countries

Statistics

Z value

P value

G_τ –2.717* –4.648 0.000

G_α –8.503 –1.074 0.141

P_τ –12.28* –5.609 0.000

P_α –7.88* –3.434 0.000

Entire Panel	Statistics	Z value	P value
G_τ	–2.028**	–1.908	0.028
G_α	–7.870	–0.890	0.187
P_τ	–15.72*	–5.021	0.000
P_α	–3.732*	–6.730	0.000
Developed Countries	Statistics	Z value	P value
G_τ	–2.022***	–1.457	0.073
G_α	–7.335	–0.140	0.444
P_τ	–10.05**	–1.682	0.046
P_α	–5.782**	–1.758	0.039
Developing Countries	Statistics	Z value	P value
G_τ	–2.717*	–4.648	0.000
G_α	–8.503	–1.074	0.141
P_τ	–12.28*	–5.609	0.000
P_α	–7.88*	–3.434	0.000

Source: Authors’ estimation.

Notes: (i) G_τ and G_α are group mean statistics that test the null hypothesis of no cointegration for the whole panel against the alternative of cointegration for some countries in the panel. P_τ and P_α are the panel statistics that test the null of no cointegration against the alternative of cointegration for the panel as a whole.

(ii) *, ** and *** indicate levels of significance at 1 per cent, 5 per cent and 10 per cent, respectively.

The Hausman test results indicate that the null hypothesis, which is the restriction of homogeneity in the long run, cannot be rejected in all panels of developed countries. Therefore, the PMG estimator’s results are valid and consistent for these samples. However, in developing countries, the null hypothesis of the Hausman test is rejected at the 5 per cent significance level. Therefore, the MG estimator’s results are valid and consistent for developing countries.

The results of the PMG estimator in Table 3 demonstrate that the error correction coefficient is negative and significant at the 1 per cent significance level for all countries. Thus, there is a long-term relationship between lnGDP and lnR&D for the entire panel. In addition, the short-term and long-term coefficients of lnR&D are positive and significant at the 1 per cent significance level. Based on these findings, R&D expenditures affect GDP positively in the short term and long term. However, R&D expenditures’ long-term effects are greater than their short-term effects on GDP. Indeed, when per capita R&D expenditures increase by 1 per cent in all panels, GDP per capita rises by 0.44 per cent in the short term and by 0.98 per cent in the long term.

Table 3

Mean Group Estimator and Pooled Mean Group Estimator Results

	Entire Panel		Developed Countries		Developing Countries
	MGE	PMGE	MGE	PMGE	MGE	PMGE
lnR&D	0.6821	0.9824*	0.4834	1.0075*	0.9998*	1.1958*
ECM coefficient	–0.2864*	–0.1458*	–0.0311*	–0.1169*	–0.2473*	–0.0663
lnR&D_–1	0.3508*	0.4373*	0.4451*	0.5554*	0.2000**	0.3122*
Constant	0.6885	0.0921**	0.6946*	–0.0405	0.6788*	0.0304
Log likelihood		1177.893		811.686		368.317
Hausman test
χ²		0.20		0.22		6.52
P > χ²		0.6549		0.6378		0.0107
Number of observations	728	728	448	448	280	280
Number of groups	52	52	32	32	20	20

Source: Authors’ estimation.

Notes: * and ** indicate levels of significance at 1 per cent and 5 per cent, respectively.

The results of the PMG estimator in the fifth column of Table 3 demonstrate that the error correction coefficient is negative and significant at the 1 per cent significance level for developed countries. In this context, it demonstrates that there is a long-term relationship between lnR&D and lnGDP in developed countries. According to these results, when per capita R&D expenditures increase by 1 per cent in developed countries, GDP per capita rises by 0.56 per cent (p < 0.01) in the short term and by 1 per cent (p < 0.01) in the long term.

Finally, the results of the MG estimator in the sixth column of Table 3 demonstrate that the error correction coefficient is negative and significant at the 1 per cent significance level for developing countries. When per capita R&D expenditures increase 1 per cent in developing countries, per capita GDP rises by 0.2 per cent (p < 0.05) in the short term and by 0.99 per cent (p < 0.01) in the long term.

From these results one can interpret that in the long term, R&D spending has a positive and almost one-to-one contribution to GDP growth regardless of the development stage of a country. When R&D spending increases, it exponentially expands production capacities across time and across sectors. In the short term, however, our results differ with respect to the level of development of a country. Although R&D spending has a positive effect in the short term both for developed and developing countries, the coefficient (0.56) for developed countries is greater than (0.20) for developing countries. This difference can be attributed to the effect of differences in the stock of capital and productivity on R&D spending activities in both types of countries.

5. CONCLUSION

It is an expectation and a common view of economists that increased R&D expenditures have a positive and significant impact on economic growth. Indeed, qualitative and quantitative changes in R&D expenditures would affect technological developments, economic growth, employment capacity and export and import activities. The empirical findings obtained from this study support and contribute to the literature. Moreover, the study reveals that R&D expenditures have a strong and positive effect on GDP in both the short and long run for developed countries. For developing countries, the effects are strong in the long run; however, they are weak in the short run. Our empirical results suggest proposing that for developing countries more resources should be allocated on R&D activities for speeding up growth and economic performance.

For further research, it would be interesting to explore whether data at the firm level of countries at different stages of development produce high and statistically significant R&D elasticity to verify and support the results obtained in this article.

Footnotes

Notes

References

Aghion

, & Howitt

(1992). A model of growth through creative destruction. Econometrica, 60(2), 323–51.

Breitung

(2000). The local power of some unit root tests for panel data. In Baltagi

(Ed.), Non-stationary panels, panel cointegration and dynamic panels: Advances in econometrics (Vol. 15, pp. 161–78). Amsterdam: Jai Press.

Coe

, Helpman

, & Hoffmaister

(1997). North-South R&D spillovers. Economic Journal, 107(440), 134–49.

Freire-Serén

M. J.

(1999). Aggregate R&D expenditure and endogenous economic growth (UFAE and IAE Working Papers No. WP 43699). Barcelona, Spain: Unitat de Fonaments de l’Anàlisi Econòmica (UAB) and Institut d’Anàlisi Econòmica (CSIC).

Gittleman

, & Wolff

E. N.

(1995). R&D activity and cross-country growth comparisons. Cambridge Journal of Economics, 19(1), 189–207.

Göçer

(2013). Ar-Ge Harcamalarının Yüksek Teknolojili Ürün İhracatı, Dış Ticaret Dengesi ve Ekonomik Büyüme Üzerindeki Etkileri [Effects of R&D expenditures on high technology exports, balance of foreign trade and economic growth]. Maliye Dergisi, (165), 215–40.

Goel

R. K.

, & Ram

(1994). Research and development expenditures and economic growth: A cross-country study. Economic Development and Cultural Change, 42(2), 403–11.

Grossman

G. M.

, & Helpman

(1990). Comparative advantage and long-run growth. The American Economic Review, 80(4), 796–815.

Gülmez

, & Yardımcıoğlu

(2012). OECD Ülkelerinde Ar-Ge Harcamaları ve Ekonomik Büyüme İlişkisi: Panel Eşbütünleşme ve Panel Nedensellik Analizi (1990–2010) [The relationship between R&D expenditures and economic growth in OECD Countries: Panel cointegration and panel causality analyses (1990–2010)]. Maliye Dergisi, (163), 335–53.

10.

Hadri

(2000). Testing for stationarity in heterogeneous panel data. The Econometrics Journal, 3(2), 148–61.

11.

Harris

R. D. F.

, & Tzavalis

(1999). Inference for unit roots in dynamic panels where the time dimension is fixed. Journal of Econometrics, 91(2), 201–26.

12.

Keller

(1998). Are international R&D spillovers trade related? Analysing spillovers among randomly matched trade partners. European Economic Review, 42(8), 1469–81.

13.

Lichtenberg

F. R.

(1992). R&D investment and international productivity differences NBER Working Paper Series No. 4161. Cambridge, MA: National Bureau of Economic Research.

14.

Pesaran

M. H.

, & Smith

R. P.

(1995). Estimating long-run relationships from dynamic heterogeneous panels. Journal of Econometrics, 68(1), 79–113.

15.

Pesaran

M. H.

, Shin

, & Smith

R. P.

(1999). Pooled mean group estimation of dynamic heterogeneous panels. Journal of the American Statistical Association, 94(446), 621–34.

16.

Pessoa

(2010). R&D and economic growth: How strong is the link? Economics Letters, 107(2), 152–54.

17.

Romer

P. M.

(1986). Increasing returns and long–run growth. Journal of Political Economy, 94(5), 1002–37.

18.

Romer

P. M.

(1990). Endogenous technological change. Journal of Political Economy, 98(5), 71–102.

19.

Samimi

A. J.

, & Alerasoul

S. M.

(2009). R&D and economic growth: New evidence from some developing countries. Australian Journal of Basic and Applied Sciences, 3(4), 3464–69.

20.

Saraç

T.B.

(2009). Araştırma-Geliştirme Harcamalarının Ekonomik Büyüme Üzerindeki Etkisi: Panel Veri Analizi [The effect of research-development expenditures on economic growth: A panel data]. EconAnadolu 2009: Anadolu International Conference in Economics, Eskişehir, Turkey.

21.

Schumpeter

J. A.

(2003). Capitalism, socialism and democracy. London: T&F.

22.

Sylwester

(2001). R&D and economic growth. Knowledge, Technology & Policy, 13(4), 71–84.

23.

Tatoğlu

(2012). Panel Veri Ekonometrisi [Panel Data Econometrics]. Istanbul, Turkey: Beta Yayıncılık.

24.

Ulku

(2007). R&D, innovation, and growth: Evidence from four manufacturing sectors in OECD countries. Oxford Economic Papers, 59(3), 513–35.

25.

UNESCO. (2014). Science, technology and innovation. Retrieved from http://data.uis.unesco.org/Index.aspx?queryid=115

26.

Van Pottelsberghe del la Potterie

, & Guellec

(2004). From R&D to productivity growth: Do the institutional settings and the source of funds of R&D matter? Oxford Bulletin of Economics and Statistics, 66(3), 353–78.

27.

Westerlund

(2007). Testing for error correction in panel data. Oxford Bulletin of Economics and Statistics, 69(6), 709–48.

28.

World Bank (2014). Research and development expenditure (% of GDP). Retrieved from http://data.worldbank.org/indicator/GB.XPD.RSDV.GD.ZS/countries/1W?display=default