Scientometric Secrets of Efficient Countries: Turkey, Greece, Poland, and Slovakia R. D. Shelton 1 and Hassan B. Ali 2 1 shelton@wtec.org WTEC, 46 N. Fairfax Dr., Arlington, VA 2223, USA 2 hali@wtec.org WTEC, 46 N. Fairfax Dr., Arlington, VA 2223, USA Abstract This paper examines why some countries are very efficient in the publication of scientific papers. Various measures of efficiency can be considered, but the focus here is on the relative efficiency of the national share of publications in the Science Citation Index divided by the national share of R&D investment. Four nations excel in that relative efficiency: Turkey, Greece, Poland, and Slovakia. Turkish policies have been analyzed qualitatively to attempt to identify drivers for these beneficial results. Here a quantitative approach is taken to complement this work, and extend the analysis to other efficient countries. It is now known that certain components of R&D investment tend to be more effective in encouraging papers: the government funding component and the higher education spending component. With this insight, it can now be seen that these four countries do well in scientific paper production at least partly because they focus their investments on the most productive components. Background and Purpose The number of scientific publications of a nation is an important reflection of the scientific activity and productivity of that country. However, gross numbers in themselves are not necessarily a good measure of the efficiency of a nation in producing scientific papers. Clearly, in terms of gross numbers alone, richer nations with large economies will likely out produce nations with smaller economies. However, if one compares the efficiencies of scientific publications, the picture is quite different. Hence, establishing indicators of efficiency in publishing can offer smaller or less wealthy countries a way to better compete. There are a variety of ways to define an efficiency indicator. One can divide the number of publications by population, by GDP (Gross Domestic Product), or by GERD (Gross Expenditure for R&D). The focus here will be on a normalized measure called the relative efficiency. Some smaller countries were identified by Shelton and Foland (28) as having high and increasing relative efficiencies. The question posed in the 28 paper was, What policies in these countries led to these beneficial results? Here we provide some evidence of answers to this question. Methods and Results The analysis in this paper mostly uses paper counts in (NSF 21). The papers are based on fractional counts in the Science Citation Index. Research and Development investment data comes from OECD (21). The overall GERD investment is available for about 39 countries for decades, and data for components of GERD is available for most countries. One especially useful efficiency indicator comes from dividing publication share by GERD share (Shelton 28). This relative efficiency k i is calculated on a base of 39 countries in the OECD database, which account for some 9% of the world R&D investment and publication. Another interpretation of the relative efficiency is that it is just the ratio of papers to R&D investment, normalized by the value for the whole dataset. The mean of this normalized measure is about 1.5. For larger economies like the US, PRC, and the EU as a whole, this relative efficiency has been fairly constant (Fig. 1), which permits forecasts (Shelton & Foland (29). For example, this model shows that China will likely come to challenge U.S. and EU leadership within 1 years, because of its rapidly increasing share of world R&D investment.
Relative Efficiency Ki 1.6 1.4 1.2 1..8.6 US EU15 PRC.4.2. 199 1991 1992 1993 1994 1995 1996 1997 1998 1999 2 21 22 23 24 25 26 27 Fig. 1. Relative efficiency in producing papers from R&D investment for the largest countries. Although the relative efficiency is fairly constant for these larger countries, some smaller countries have high and even increasing relative efficiencies (Fig. 2.) Some of these efficient countries were identified by Shelton and Foland (28). In Europe some of these include Turkey, Greece, Poland, and Slovakia. In the Far East, New Zealand is also very efficient. At the time it was not known why these countries were so efficient, but it was recognized that answering this question might uncover policies that could improve the efficiency of other countries. Last year Foland and Shelton (21) analyzed a similar question to determine why the EU greatly increased its efficiency in the 199s, leading it to pass the US to become the world s leader in scientific publication. By multiple regression they determined that some components of research investment seem to be much more effective in encouraging papers, as suggested by Leydesdorff and Wagner (29). These are government investment (rather than industrial investment) and spending in the higher education sector. More recently Shelton and Leydesdorff (211) have extended this analysis to patents, and have determined that there is a trade-off between investment allocations: some are best at encouraging papers and others encourage patents. While regressions cannot prove cause and effect, they provide evidence of these connections. This paper returns to the question of why some countries are very efficient and are increasing their efficiencies. In particular the conference host country, Turkey, stands out because it has been rapidly increasing its publication share, and its efficiency at doing so. Several Turkish scholars have contributed to a better understanding of this phenomenon, and its causes: Akpinar (21); Yurtserver (1999); Uzun (26); Gokceoglu (28); Baskurt (211); and Erkan (29). (We were unsuccessful in finding similar analysis for the other efficient countries.) 4. 3.5 3. 2.5 2. 1.5 1..5. 1995 1996 1997 1998 1999 2 21 22 23 24 25 Poland Slovakia Greece Turkey Fig. 2. Relative efficiency in producing papers from R&D investment. Some countries with
high, and increasing, efficiencies. Akpinar and Karçaaltıncaba (21) analyzed scientific papers published by Turkish authors in the field of radiology and medical imaging included in Science Citation Index Expanded database. They noted that the number of such papers has increased significantly in recent years and the driving force leading to increased publications in the field of radiology from Turkey appears to be the rapid adaptation to imaging technology and improved education.in our opinion, the main factor for increased publications from Turkey is the extensive investment in human resources and technology in university, government and private hospitals and clinics Gokceoglu et. al (28) examined the publication trends in the international earth science literature from Turkey in the period of 197-25 using the Science Citation Index Expanded database. They note a sharp increase in the number of earth science publications from Turkey starting in the 199 s, in parallel with the increase in the total scientific output of Turkey. In the last decade the annual growth rate has been 16%, with a concomitant increase in the number of citations. They attribute the sharp increase in the publication numbers to, in order of importance, changes in the rules of academic promotion and appointment, changes in academic attitudes towards publishing, increasing support for research, financial incentives for publishing, and expansion of higher education. These results suggest that, in the case of Turkey, at least, the causes of the increasing efficiency of publications may vary somewhat, depending on the particular scientific field, but it is clearer now that this overall pattern is strongly associated with Turkey's focus on what are now known to be the most effective investments. Data from the Turkish government confirms their science policy strategies and the impressive R&D statistics emerging from Turkey over the past decade (1998 to 29). (EURAXESS Turkey (21). We quote a few of these statistics here:. When a world ranking is calculated in terms of the growth rate more specifically between the years 22-28, the Republic of Turkey is: 3 rd rank in terms of growth rate in GERD, with a position improved from 26 to 21, 7 th rank in terms of growth in GERD as a percentage of GDP, with a position from 5 to 42, 1 st rank in terms of growth in FTE researchers, with a position from 25 to 16, 3 rd rank in terms of the growth in scientific publications, with a position from 22 to 18, 2 nd rank in terms of the growth in PCT patent application, with a position from 33 to 27. Erkan (29) notes that the GERD intensity in Turkey exhibits a significant upward trend after 22, and reached a historical peak in 25, when more than 4% of Turkish GERD was financed by industry and 5% by government. The main economic problem of the S&T sector was the low levels of participation of private domestic and foreign companies in the financing of R&D. We have noted that there are several different indicators of a nation s R&D output. Erkan states that one of the most promising indicators for Turkey is the one measuring scientific publications per million inhabitants. Although Turkey s share is around 1.5% in the OECD area, in the last five years, with a growth rate of 141% Turkey reached the highest rate in the world. Shelton and Leydesdorff (211) have shown that the most effective components for production of scientific papers are the (Gov%) government funding component of GERD, and its higher education spending component (HERD%). The correlation coefficients between the relative efficiency and these two components are.6 for Gov% and.7 for HERD%. Both independent variables are significant in predicting the relative efficiency with p-values of better than.. A concise way of demonstrating this is in the scattergrams of Figs. 3-4, for the countries in the OECD database.
4 Scatter plot of Ki vs Gov% 3 i K2 1 1 2 3 4 Gov% 5 6 7 Fig. 3 Scatterplot of relative efficiency k i vs Gov%. The regression line is: k i =.21 +.349Gov%, the p <.1 for Gov%, and the R 2 is 35.9%. 4 Scatter plot of Ki vs HERD% 3 i K2 1 1 2 3 HERD% 4 5 6 Fig. 4. Scatterplot of relative efficiency k i vs HERD%. The regression line is: k i =.4 +.487HERD%, the p <.1 for HERD%, and the R 2 is 46.5%. These single independent variable (IV) models show a fairly strong relationship with the relative efficiency, but the scatter shows that there are other important factors involved. For example, the scatter can be reduced with a two IV model using both Gov% and HERD%. This regression equation is: k i =.192 +.375 HERD% +.224 Gov% The fit with this more elaborate model is improved to R 2 = 58.8%. The p-values here for HERD% and Gov% are. and.3, respectively.
Thus, to explain why some countries are efficient, we can examine their trends in these components. For example the percent of funding of R&D by government is shown for three very efficient countries (Turkey, Greece, and Slovakia) in Fig. 5. The similar curve for Poland is omitted here for clarity. This can be contrasted with a similar curve for a very inefficient country (the US), where the government funding component is far smaller than the industrial component. Similarly Fig. 6 shows the part of R&D funding that is done in universities (HERD%) for Turkey, Greece, and Poland. The similar curve for Slovakia is omitted here for clarity. Thus it is reasonable that the four countries should be efficient in producing scientific papers from their overall R&D investment, and the US should be inefficient. The most recent data in Fig. 5-6 suggests, however, that Turkey s policy strategies may be changing, and thus it may lose some of its momentum. Part of GERD Funding from Government 8 7 6 5 4 3 2 1 Turkey US Greece Slovakia 1991 1993 1995 1997 1999 21 23 25 27 Fig. 5. Percent of GERD funding from Government (Gov%) for three efficient countries, Turkey, Greece, and Slovakia; compared to an inefficient one, the US. Part of GERD Spending at Universities 8 7 6 5 4 3 2 1 Greece US Turkey Poland 1991 1993 1995 1997 1999 21 23 25 27 Fig. 6. Percent of GERD spending in higher education (HERD%) for three efficient countries, Turkey, Greece, and Poland; compared to an inefficient one, the US.
Conclusions An especially useful indicator of a nation s scientific productivity, as measured by the number of its scientific publications, is obtained by dividing its world publication share by its world GERD share. For larger economies like the US, PRC, and the EU as a whole, this relative efficiency, k i, has been fairly constant, which permits forecasts of output publication share from input GERD share. This k i not only measures efficiency of an individual country, but also compares it to others, with a normalization that shows relative changes to others with time. Unlike the larger economies, however, some smaller countries have high and increasing relative efficiencies. Some of these in Europe these are Turkey, Greece, Poland, and Slovakia. Some light is shed into this seeming mystery if one divides the overall GERD R&D investment sources into two main components: industrial funding and government funding--which can be shown by regression to be much more effective in encouraging papers. Similarly R&D spending in the higher education sector is much more effective than spending in the business and other sectors. The high allocations toward government funding and higher education spending in these four countries seem to account for much of the superior efficiencies in these countries. Acknowledgements This work was sponsored by NSF cooperative agreement ENG-844639. References Akpinar, E. & Karçaaltincaba M. (21). Analysis of scientific papers in the field of radiology and medical imaging included in Science Citation Index Expanded and published by Turkish authors. Diagn Interv Radiol 21; 16:175 178. Baskurt, Ok. (211). Time series analysis of publication counts of a university: what are the implications? Scientometrics 86: 645-656. Erkan E. & Gossartb C. (29). Appraising the integration of sustainable development into sectoral policies: The case of Turkish science & technology policies. Accessed July 9, 211 from (http://www.stps.metu.edu.tr) EURAXESS Turkey (21). R&D Statistics in Turkey. TUBITAK. doi: 1.1787/data-182-en Accessed on 13 March 211 from (http://euraxess.tubitak.gov.tr/rd-statistics-in-turkey) Foland P. & Shelton, R.D. (21). Why is Europe so efficient at producing scientific papers, and does this explain the European Paradox? Eleventh International Conference on Science and Technology Indicators, Leiden, Sept. 21. Accessed July 14, 211 from http://itri2.org/s/ Gokceoglu C. (28). International earth science literature from Turkey - 197-25: Trends and possible causes. Scientometrics 74 : 49 DOI 1.17/s11192-7-1813-x 28 Leydesdorff, L. & Wagner, C. (29). Macro-level indicators of the relations between research funding and research output. Journal of Informetrics 3:4, 353-362. NSB (21), Science and Engineering Indicators -- 21. Arlington, VA: NSF. OECD (211). "Main Science and Technology Indicators", OECD Science, Technology and R&D Statistics (database). Paris: OECD. Shelton, R.D. (28). Relations between national research investment and publication output: Application to an American paradox, Scientometrics, 74:2, 191-25. Shelton, R.D. & Foland P. (28). National Efficiencies in Publishing Scientific Papers, Tenth International Conference on Science and Technology Indicators, Vienna, Sept. 28. Accessed July 14, 211 from http://itri2.org/s/ Shelton, R.D. & Foland P. (21), The Race for World Leadership of Science and Technology: Status and Forecasts, Science Focus, 5:1, 1-9. Accessed July 14, 211 from http://itri2.org/s/ Shelton, R.D. & Leydesdorff, L. (211). Bibliometric Evidence for Funding Trade-Offs in National Funding Strategies, 13 th International Conference on Scientometrics and Informetrics, Durban, July. 211. Accessed July 14, 211 from http://itri2.org/s/
Uzun A. (26). Science and technology policy in Turkey. National strategies for innovation and change during the 1983-23 period and beyond. Scientometrics 66: 551 DOI 1.17/s11192-6-4-1. Yurtsever, E. (1999). The increase in the rate of publications originating from Turkey Scientometrics 46: 321. Source: 211 Collnet Conference, Istanbul, Sept., 211 http://itri2.org/