Index

1. INTRODUCTION

Innovation is the source of competitive advantage and prosperity for business and countries. That’s why it is necessary to study its determinants. Financial capital and human capital were often the main determinants of innovation. This suggests that the more resources devoted to research and development (R&D), the more innovative is the country. However, the economic world is characterized by the scarcity of resources. R&D expenditure cannot be increased indefinitely, but it is necessary to think about making good use of existing resources and using them in an efficient way. In this context, the research questions of our study are: What is the efficiency level of R&D expenditure? Does R&D intensity reflect R&D efficiency level?

The purpose of this paper is to estimate the efficiency of research and development expenditure and to classify ten emerging countries in order to identify particularly efficient countries and particularly inefficient countries.

Many authors have been interested in the issue of R&D efficiency (Wang, 2007; Wang and Huang, 2007; Li, 2009; Chen et al., 2011; Hu et al., 2014). This study aims to enrich the existing literature by focusing on ten emerging countries.

To estimate efficiency scores, we use the stochastic frontier model of Battese and Coelli (1992). The estimation results allow us to quantify R&D efficiency levels and to compare them to R&D intensity levels. Our findings suggest that innovative countries are not necessarily efficient in the use of their resources and vice versa.

The rest of the paper is organized as follows: Section 2 introduces the methodology. Section 3 presents and interprets the results. Section 4 concludes.

2. METHODOLOGY

This paper considers a production framework of R&D activities based on production theory. Each country is considered as a decision unit that employs human and physical resources of R&D as inputs to produce an output such as patents and scientific publications. The stochastic frontier approach is applied to estimate the relative efficiency of R&D of each country.

The R&D production function

Following Griliches (1990) our study considers the international research and development activity in the context of a production function in different countries.
R&D production function of different countries has the following general form:

Ykt = f(Xkit)                (1)

where: k=1,….., K (country), i=1,…., N (inputs), t=1,…..T (years)

Ykt is the R&D output of country k at time t.

Xkit is the input i of country k at time t.

The wide variety of inputs resulting from the use of data from different countries requires the use of a flexible functional form. The most used is the translog form.

A translog production function at time t can be written as:

ln Ykt = β0 + ∑ βi (ln Xkit) + ½ ∑ βii (ln Xkit)2 + ½ ∑ ∑βij (ln Xkit) (ln Xkjt)                   (2)

Where: Y is the output quantity, Xi and Xj are the inputs i and j, respectively. ln is the natural logarithm. βi βii and βij are parameters to be estimated and are independent of countries.

The stochastic frontier model

This research aims to measure the R&D efficiency scores of ten countries. To do this, we use the frontier model of technical efficiency of Battese and Coelli (1992).

The model of Battese and Coelli (1992) may be expressed as:

Ykt = f(Xkt, β) exp (Vkt – Ukt),                                                      (3)

Ukt = ƞkt Uk= {exp[-ƞ (t - T)]} Uk, t ϵ Ῥ(k); k = 1, 2, . . ., K;       (4)

where: k = 1,…, K country ; t = 1,…, T (year)

Ykt is the production output of kth country at time t.

Xkt is a vector (1*i) of production inputs relative to kth country at time t.

β is a vector (i*1) of parameters to be estimated.

Vkt’s are assumed to be independent and identically distributed N(0, σ2v) random errors;

Uk’s are assumed to be independent and identically distributed non-negative truncations of the N(μ, σ2) distribution;

ƞ is an unknown scalar parameter;

and Ῥ(k) represents the set of Tk time periods among the T periods involved for which observations for the k th firm are obtained.

The stochastic production frontier model to be estimated in this paper is defined as follows:

ln Ykt = β0 + ∑ βi (ln Xkit) + ½ ∑ βii (ln Xkit)2 + ½ ∑ ∑βij (ln Xkit) (ln Xkjt) + Vkt – Ukt    (5)   

       
where: Ykt is the production output of kth country at time t.

Xkt is a vector (1*i) of production inputs relative to kth country at time t.

β is a vector (i*1) of parameters to be estimated.
Parameters are estimated using FRONTIER 4.1 program developed by Coelli (1996).

2.1. Sample Selection

This study considers two empirical models with two different outputs. The first model (Model n°1) will focus on ten emerging countries: Bulgaria, Croatia, Hungary, Latvia, Mexico, Poland, Romania, Russia, South Africa and Turkey. The period covered is 2001- 2010. The second model (Model n°2) will be applied to the same sample during the 2001-2009¹ period. The sample of the study and the periods are selected according to data availability.

2.2. Data

The output of R&D spending is measured by two indicators. (PAT) is the annual number of patent applications filed by the inventors of each country with the United States Patent and Trademark Office (USPTO). The second measure of output is the number of scientific publications (PUB). It is the annual number of papers published in Science Citation Index international journals and the annual number of papers published in Social Sciences Citation Index international journals. These publications were selected by the indicators of science and engineering in 2014 from the National Science Foundation (2014)² .

The inputs of R&D spending are financial capital and human capital.

The R&D manpower (MRD) measured in full time equivalent is all persons employed directly in R&D as well as those providing direct services related to R&D, such as managers, and administrators (OCDE, 2002).

Concerning the stock of knowledge, the variable introduced reflects the cumulative R&D. For each country, we calculate the stock of research and development (SRD) using the method of perpetual inventory (OCDE, 2001) and based on gross domestic R&D expenditure (in thousands Power Parity Purchase).

Data on inputs are from UNESCO database.

Like Wang (2007) we take two years of difference between the inputs and output data. Thus, in models 1 and 2, data for patents and scientific publications are taken for the periods 2003-2012 and 2003-2011, while data on R&D human capital and R&D financial capital relate to the periods 2001-2010 and 2001-2009, respectively.

3. FINDINGS

Table-1. Mean R&D efficiencies during 2001-2010

Source: Prepared by the author on the basis of FRONTIER Output

Table-2. Mean R&D efficiencies in the ten emerging countries

Source: Prepared by the author on the basis of FRONTIER Output

3.1. Average Performance

The analysis of efficiency scores shows that the average performance of innovation systems is low. We also note that the mean efficiency score for scientific publications is the best. In fact, the average value of the scores is 35% for publications and 28.7% for patents, indicating potential improvements in innovation results of 65% and 71.3% respectively, with the same levels of expenditure. Thus, the countries in the sample seem to have more leeway to increase patents than to increase publications. This result is consistent with findings of Wang (2007); Wang and Huang (2007) and Chen et al. (2011) where the average technical efficiency is higher when scientific publications are adopted as output rather than patents.

Moreover, we note from Table 1 that the efficiency of innovation systems has remained relatively stable since 2001 (especially in terms of scientific publications).

The Most Efficient Countries

For each innovation indicator, a group of efficient countries emerges.

For patents, the most efficient countries are: South Africa and Hungary.

For publications, the most efficient countries are Russia and Turkey.

The Most Inefficient Countries

For patents, the most inefficient countries are Latvia, Romania and Poland. For scientific publications, the most inefficient countries are Latvia, Romania and Bulgaria.

Note that Romania is among the least efficient countries, regardless of the orientation of efficiency (patents or scientific publications). This result corroborates that found by Chen et al. (2011).

Figure-1. Mean efficiencies for the ten emerging countries

Source: Prepared by the author on the basis of FRONTIER output

Figure 1 compares the efficiency scores obtained for each of the innovation indicators and shows that some countries such as Turkey, Russia and Poland are much more efficient in terms of scientific publications than patents.

3.2. Performance at the individual level

At the individual level, the level of efficiency varies between countries according to the potential for innovation (see Table 3).

Table-3. Innovation level and efficiency level in the ten emerging countries

Source: Prepared by the author on the basis of FRONTIER Output and EUROSTAT Database

Patent-Oriented Efficiency

Russia and Croatia have the highest levels of domestic R&D spending but have a low average level of efficiency (ranked 7th and 5 th respectively). Mexico is weakly innovative (in terms of share of gross R&D spending in GDP), yet it is ranked among the top countries in terms of efficiency. It is in the 3rd place.

Hungary and South Africa are the most innovative countries in our sample and are also ranked among the most efficient countries, with ranks 2 and 1, respectively.

Scientific Publication-Oriented Efficiency

The results show that Russia is the country with the greatest potential for innovation and is also the most efficient. However, although Hungary and South Africa have strong potential for innovation (in terms of R&D share in GDP), they are ranked sixth and fifth in terms of efficiency.

Romania and Mexico are the least innovative countries in our sample and are also ranked among the most inefficient countries.

4. CONCLUSION

After estimating R&D efficiency scores, our study shows that emerging countries are lowly efficient in the use of their R&D resources. Results suggest that an innovative country is not necessarily efficient. Our analysis has important implications for policy makers. In fact, an effective public innovation policy must aim not only to increase innovation but also R&D efficiency. Our analysis also shows that the efficiency of innovation systems has remained relatively stable since 2001 (especially in scientific publications), indicating that improving efficiency is not a top priority for the government agenda and innovation policies, which is harmful for emerging countries.

REFERENCES

Battese, G.E. and T.J. Coelli, 1992. Frontier production functions, technical efficiency and panel data: With application to paddy farmers in India. Journal of Productivity Analysis, 3(1/2): 153-169. View at Google Scholar | View at Publisher

Chen, C.P., C.H. Yang and J.L. Hu, 2011. An international comparison of R&D efficiency of multiple innovative outputs: The role of the national innovation system. Innovation: Management, Policy & Practice, 13(3): 341-360. View at Google Scholar | View at Publisher

Coelli, T.J., 1996. A guide to Frontier, version 4.1: A computer program for stochastic frontier production and cost function estimation. CEPA Working Paper No. 96/07. Retrieved from https://files.itslearning.com/data/ku/103018/publications/coelli96.pdf.

Griliches, Z., 1990. Patent statistics as economic indicators: A survey. Journal of Economic Literature, 28(4): 1661-1707. View at Google Scholar 

Hu, J.L., C.H. Yang and C.P. Chen, 2014. R&D efficiency and the national innovation system: An international comparison using distance function approach. Bulletin of Economic Research, 66(1): 57-71. View at Google Scholar | View at Publisher

Li, X., 2009. China’s regional innovation capacity in transition: An empirical approach. Research Policy, 38(2): 338-357. View at Google Scholar | View at Publisher

National Science Foundation, 2014. Science and engineering indicators. National Center for Science and Engineering Statistics, and the Patent BoardTM, Special Tabulation (2013) from Thomson Reuters, SCI and SSCI.

OCDE, 2001. La mesure des stocks de capital, de la consommation de capital fixe et des services du capital. Manuel de l’OCDE.

OCDE, 2002. Méthode type proposée pour les enquêtes sur la recherche et le développement expérimental. Manuel de Frascati. Retrieved from http://www.stis.belspo.be/docs/pdf/Frascati2002_finalversion_f.pdf .

Wang, E.C., 2007. R&D efficiency and economic performance. A cross-country analysis using the stochastic frontier approach. Journal of Policy Modeling, 29(2): 345-360. View at Google Scholar | View at Publisher

Wang, E.C. and W. Huang, 2007. Relative efficiency of R&D activities: A cross country study accounting for environmental factors in the DEA approach. Research Policy, 36(2): 260-273. View at Google Scholar | View at Publisher

Footnotes: