ISSI 2011 Microsoft Word Template for Full and Research in ...



Scientometric Insight on a Bottom Line of Innovation:

High-Technology Exports

R. D. Shelton1

1shelton@

ITRI, 518 S. Camp Meade Road, #6, Linthicum, MD, 21090 (USA)

Abstract

Indicators like counts of papers and patents provide useful insight into the success of national innovation ecosystems. However, they are only proxies for the quantities that the public cares most about: jobs, strength of the economy, plus survival and growth of flagship companies and industries. Some scientometric measures of innovation that come closer to these concerns are the performance of high-technology industries, including their exports and international market share, which are complied by the OECD in its Main Scientific and Technical Indicators series. Here this data is analyzed to determine connections between these innovation system outputs and input indicators like R&D investment and the number of researchers, plus intermediate ones like papers and patents. The overall input that was most predictive of these exports was R&D investment in business (BERD), using logarithmic scales to reduce the distortions caused by outliers like the US and China. The results also show that the traditional indicators of science, papers and patents, can indeed be good predictors of high-technology exports.

Introduction

A nation’s science and technology (S&T) establishment (or innovation ecosystem) can be considered to be an economic system that needs inputs of resources like labor and capital to produce outputs such as jobs and exports, which contribute to national prosperity (Fig. 1). Inputs to and outputs from such a system can be measured using indicators. Intermediate indicators like scientific paper publication and grants of patents are also helpful. Previous work has shown that there is a strong relation between inputs and these intermediate indicators.

Leydesdorff (1990) regressed world share of publications in the Science Citation Index (SCI) as output on government expenditure for R&D (GERD) data as an input. Later, the measurement was refined by focusing on components of GERD funding as independent variables (Leydesdorff & Gauthier, 1996; Leydesdorff & Wagner, 2009).

Shelton (2008) compared which of the other national inputs were most important in encouraging papers; for example, the number of researchers was less significant than investment. From the structure of the publication process, he built a model for individual countries. This model suggested that changes in the GERD share have been the driver of changes in paper share, accounting for the rise of China since 2001 (Moed, 2002; Jin and Rousseau, 2005; Shelton & Foland, 2010), and the inevitable associated decline of the U.S. and EU. More accurate models then helped explain Europe’s increase in efficiency during the 1990s (Foland & Shelton, 2010). The government component of R&D funding and the higher education spending component were found to be relatively effective in encouraging papers, helping to account for the EU passing the U.S. in SCI papers in the mid-1990s.

[pic]

Fig. 1. Linear model of an innovation enterprise with some indicators.

However, the opposite side of this coin is that industrial funding was shown to be more effective in encouraging patents than government funding. (Shelton & Leydesdorff, 2012). While lags between funding and paper publication seem to be short enough to neglect, lags are essential for modelling the patenting process (Shelton & Monbo, 2012). Here similar analysis will be applied to export of high-technology products as a dependent variable.

First, however, some economic literature should be reviewed, because high-technology exports and R&D investment are also economic indicators. Economists have long been interested in connecting technology and trade, but mostly they consider the problem of how innovation affects trade, not how do input resources affect exports. One authority who has bridged the gap between economics and scientometrics is Luc Soete (1987).

While there are some economic papers on factors that best explain overall international trade, there are relatively few that focus on the high-technology sector. One economic analysis of whether a country's high-tech exports (as a share of its overall exports) could be explained by R&D investment and country size was done by Braunerhjelm and Thulin (2008). They used OECD data for 19 countries during 1981-1999. From their economic model, they concluded that overall R&D investment was significant, but for their case the size of the market was not.

Tebaldi (2011) used panel data to analyze factors that are most explanatory of high-technology trade. This approach adds data from more than one year to the usual cross-country analysis. Human capital, inflows of foreign direct investment, and openness to international trade were found to be the most significant of the factors he analyzed.

Classical economists have long taken as gospel Ricardo's (1817) observations on the comparative advantages that nations have in manufacturing products: wine in Portugal, and cloth in Britain. Thus the optimum policy was held to be free trade between these two by manufacturing only what they did best. This faith was not shaken by long debates over Ricardo's simple models, or even when nations like Japan clearly succeeded with mercantilism: policies that nurtured targeted industries. Once they had achieved economies of scale, they became manufacturing superpowers. Finally Krugman (1991) constructed a theory called the New Economic Geography that explained how this can happen, winning the Nobel Prize for his work in 2008. His model explains why manufacturing is often located near large markets like China.

Modelling of economic systems can still yield results that seem to clash with reality. Ekholm and Hakkala (2007) have a model which suggests that the need by high-technology manufacturing for skilled labor, may encourage R&D to move elsewhere. However in China at least, government policy of encouraging "indigenous innovation" to move up the technology ladder is encouraging multinationals to locate their labs there (Buckley, 2004), (Barboza & Markoff, 2011) (McGregor, 2011). Thus in the nation that is by far largest and by far the most rapidly growing, there is a connection between location of research and location of high-tech manufacturing, driven by government policies as well as economics. And those government policies also strongly encourage patenting and publication in journals indexed by the SCI, making the connections found here more plausible.

The Data and Its Overall Trends

The Main Science and Technical Indicators (MSTI) series from the Organization for Economic Co-operation and Development has been widely used in scientometrics (OECD, 2012). It has data from about 40 industrialized countries going back for decades. Some broader data is available from the World Bank (2012). However, the OECD nations account for more than 95% of industrial output.

Perhaps best known are the MSTI data on national input resources for research, including many components of gross expenditures on research and development (GERD) and counts of researchers. It has some coverage of patents, but does not cover papers; here fractional count data from (NSB, 2012) is used. MSTI does have a set of truly output indicators in its set of series on the high-technology product sector, including exports, imports, and international export share. These are presented separately in five sub-sectors; export share for the one for what is for now mostly the computer sub-sector is shown in Fig. 2.

Many people are interested in where their computers are made, and have followed this rapid shift from dominance of the sector by the U.S. and Japan, to dominance by China. China's market share is well over twice that of the next three major players combined! Taiwan's role is interesting, in that it also gained market share in the 1990s until it shifted much of its production to the mainland.

This measure of industrial output does not capture all the nuances of where manufacturing really takes place (Xing, 2012). For example, the Apple iPad is assembled in China, but many of its components come from Japan, the U.S. and elsewhere. Later a new report from the OECD and World Trade Organization (WTO) will be discussed, which has some data for overall trade on a value-added basis (OECD-WTO, 2013). For the moment, the OECD series on the electronics sub-sector sheds some light on this issue, since many of these electronic components are assembled into computers. Figure 3 shows that China's domination of these components is only slightly less complete. Its share is only about equal to the sum of those of its three leading competitors--not twice that figure.

[pic]

Fig. 2. Export share in office machines and computers.

[pic]

Fig. 3. Export share in electronics.

Analysis

Similar charts will be presented for the other three sub-sectors in the Appendix. First, however, the five series can be aggregated to provide an overall picture of the export market share for the overall high-technology product sector for the nations with OECD data (Fig. 4). The most dramatic trend confirms the well-known rise of China at the expense of earlier manufacturing leaders like the U.S. and Japan. In less than a decade, China went from being a minor player to the world leader in this indicator. There are many reports that western and Japanese companies located their manufacturing plants (sometimes called "transplants") in China in this period to take advantage of relatively cheap labor, and to take advantage of a huge and rapidly growing market (Scott, 2012).

Some people in the U.S. and Europe are concerned about these trends and want their governments and industries to respond constructively to this competition (Weller and Wheeler, 2008). After all, national investments in R&D can only be recovered through manufacturing and sale of the resulting products. Policy responses have included focused R&D programs in an attempt to encourage domestic industry. It seems obvious that such investments in R&D and in encouraging more domestic researchers should pay off in better results, such as greater international market share in high-technology products. Here, some analysis is presented on which of these policy actions might be most effective.

[pic]

Fig. 4. Export shares in the overall high-technology sector. The base is the countries with OECD data.

Simple correlation over the 40 or so countries in the OECD database of input resources can provide insight into which investments might be most productive in encouraging high

technology exports. Tables 1 and 2 show the square of the sample correlation coefficient r2, (the coefficient of determination in percent) for a selection of a half-dozen such inputs, plus some intermediate indicators, e.g. papers and patents. Table 1 uses exports in 2009 as the output variable for inputs from several years to examine the effects of lags. Table 2 uses exports in 2007 with inputs from the same year, under several approaches to dealing with outliers. Generally the two tables show similar patterns, providing some reassurance that the findings are stable. Table 1 suggests that lags are not so important in modelling the aggregate export figure, but later we will see that this depends on the individual sub-sectors.

Some caveats: all the correlations are fairly high partly because all variables tend to increase with the size of the country; one needs to find measures that have stronger correlations with exports than just sizes measured by GDP or population. Further, slight differences between the values can be important; in a regression with two input variables, only the one with the higher correlation will usually be found to be significant by comparison. There is a huge range of sizes for these countries; unless one is careful with data from the US and PRC, most of the results will come from these outliers, with small countries having little influence on the connections between exports and policies that one is seeking. Using the logarithmic scale in the rightmost column of Table 2 seems to yield the most realistic results: all countries included, no outliers, fairly strong correlations with inputs and intermediate variables, which are all well above those for the measures of country size.

Some traditional scientometric indicators will be discussed first. Papers indexed in the Science Citation Index (Row 1a. in both tables) are quite effective at explaining the national variation in high-technology exports—usually far better than the size measures in Rows 4a. and 4b.

Several international patent series are available from the OECD. Triadic patents (1b.) involve applications to all three patent offices for the U.S., EU, and Japan. Since this process is expensive, only the most promising inventions get this treatment, so the term "high value patents," is sometimes used. The Patent Co-Operation Treaty (PCT) series (1c.) for applications for international patents has some of the highest correlations in the tables. Thus patent measures can also be a useful proxy for downstream processes like exports.

Table 1. Coefficients of determination (in %) of high-tech exports in 2009 with explanatory variables in year shown. Linear Scale and US and PRC omitted as outliers.

| |2007 |2008 |2009 |

|1a. Papers SCI |46.6 |47.9 |48.7 |

|1b. Patents Triadic |34.5 |36.0 |36.0 |

|1c. Patent PCT Apps |49.1 |49.6 |46.1 |

|2a. GERD |42.3 |44.2 |47.6 |

|2b. BERD |40.4 |41.1 |44.6 |

|3a. Researchers |25.0 |26.8 |29.3 |

|3b. Business Researchers |27.1 |28.4 |30.3 |

|4a. Size GDP |36.8 |34.3 |35.2 |

|4b. Size Population |13.6 |13.5 |13.4 |

Table 2. Coefficients of determination (in %) of high-tech exports in 2007 with explanatory variables in the same year.

| |Linear w/o |Linear with |Log with |

| |US & PRC |US & PRC |US & PRC |

|1a. Papers SCI |49.1 |55.1 |53.8 |

|1b. Patents Triadic |39.6 |36.1 |63.6 |

|1c. Patent PCT Apps |53.4 |49.3 |60.8 |

|2a. GERD |47.5 |56.0 |62.1 |

|2b. BERD |45.4 |56.1 |65.2 |

|3a. Researchers |28.9 |74.3 |50.5 |

|3b. Business Researchers |32.0 |73.8 |59.2 |

|4a. Size GDP |39.8 |67.1 |43.8 |

|4b. Size Population |15.0 |54.6 |26.7 |

GERD (2a.) is national R&D investment from all sources. The BERD series (2b.) is business expenditures on R&D from all sources, but has very similar correlations to GERD with exports. In the final column it is slightly better than overall GERD.

Researchers (3a.) are in terms of full time equivalents at all levels and all institutions: business, government, and academia. Business Researchers (3b.) is focused on companies, and mostly has slightly higher correlations with exports than the overall count. One reason for focusing on 2007 is that the U.S. has not supplied data for Researchers since then, and the PRC radically changed its counting methods after 2007, causing a large decrease in its data.

The tables also include two series that measure the overall size of the countries: Gross domestic product (GDP) and Population. Exports are a component of GDP, so a strong connection would be expected. GDP is strongly correlated with all the input variables (usually more than 0.9). Since these variables increase with the size of the overall economy of the country, it might be thought that the effects of inputs like GERD are simply mimicking this effect. However, both size measures have correlations below the other series in the tables.

Of the truly input variables, GERD or BERD is most explanatory, and Researchers is well behind in correlations, indicating that capital is more important than labor for the smaller countries--once the US and PRC are removed as outliers.

A single variable regression for the best input variable is BERD in 2007 (Row 2b., last column in Table 2:

log Exports(2007) = 1.183 + 0.839 log BERD(2007) (1)

For this case N= 40 countries, R2 = 65.2%, the p significance value is 0.000. A scattergram is shown in Fig. 5 for this regression.

A regression with both BERD and Researchers is:

Log Exports(2007) = 1.64 + 1.04 log BERD(2007) - 0.250 Log Researchers (2007) (2)

[pic]

Fig. 5. Scatterplot of high-technology exports vs. BERD in 2007 for

N =40 countries in the OECD dataset. Log scales used.

The fit of this multiple regression is only improved to R2 = 66.5%, and BERD has a p value of 0.000. The Researchers variable is not significant with p = 0.435, and its sign in the regression equation is negative, indicating that for a given value of BERD, addition of more Researchers tends to decrease exports. Thus a simple, single input variable, regression using BERD seems to be suitable for modelling.

The MSTI also has series for the BERD investment in each of the five sub-sectors, although data is not available for some countries. In particular there is not much data from China. Fig. 6 uses lagged correlation to shed some light on whether R&D investments can be said to encourage higher exports. The results differ for the sub-sectors. The aerospace field has a very high correlation, which increases even further when the lag is increased, up to a peak at seven years. This suggests that there may well be a causal relationship between R&D investments and exports here.

The pharmaceutical sub-sector has a similar pattern, albeit with a much lower correlation, perhaps reflecting the multinational distribution of research and manufacturing in this field.

The three IT sub-sectors are intermediate and show all patterns of correlation with increasing lags: declining, flat, and increasing.

[pic]

Fig. 6. Lagged correlations of exports in 2009 with BERD investments years earlier for each of the five sub-sectors in the OECD data.

Further Work

As mentioned earlier, OECD has been reporting international trade for decades on a cumulative basis, where imported components of finished goods are counted more than once. Five days before this paper was due, they released a new report and database on a value-added basis (OECD-WTO, 2013). Several sectors are reported, but they do not include the high-technology one considered here. The overall results are quite different. For example the reported U.S. trade deficit with China is now much lower, and almost half of EU trade is in services, up sharply from previous reports. This data will be analyzed and compared to the measures here when time permits.

Conclusions

Analysis of scientometric indicators can be extended to measures like high-technology exports that better capture the overall success of a nation's scientific enterprise. The connection of exports to traditional scientometric indicators like papers and patents might be reassuring, since they can be as good explanatory variables as upstream resource inputs. However these connections and others to capital and labor are not as tight as one would like, since external factors seem to be especially important. These may include the decisions of multinational companies to locate manufacturing where labor is cheap and markets are growing, since the intellectual property can usually be imported from abroad.

The strongest connection between resource inputs and overall high-technology exports was with the R&D investment in the country’s business sector (BERD). The correlation was adequate to show that useful models could be built, and that the value of this input was much more than merely a measure of the size of the country.

However, the sub-sectors of high-technology exports display quite different behaviors. Manufacturing of information technology products has suddenly shifted to China. This seems to have been the result of decisions by companies to locate manufacturing plants where labor costs are low and where governments provided incentives to do so. These decisions are hard to associate with upstream research resources, since their intellectual property results can easily be imported.

In contrast the U.S. and Europe continue to dominate the aerospace sub-sector, where exports seem to be closely tied to R&D investments in that country by its industry--some years earlier. The pharmaceutical industry is dominated by a dozen or more large multinationals who have R&D and manufacturing located somewhat by accident, because they grew by mergers and acquisitions.

Finally, while high-technology exports are one bottom line of national innovation, there are many others. Science and technology help produce much more than the latest gadgets: overall economic output, jobs, improved medical care and quality of life generally. And to a great degree, these benefits can be enjoyed by everyone, regardless of where the R&D and manufacturing are located.

Acknowledgments

Funding from NSF coop agreement ENG-0844639 is much appreciated. The opinions stated in this paper are the author's, and not necessarily those of the funding agency.

References

Barboza, D., & Markoff, J. (2011), Power in numbers: China aims for high-tech primacy, New York Times, Dec. 5, 2011 Accessed Jan. 4, 2013 from:

Bradsher, K. (2013), China's focus on aerospace raises security concerns, New York Times, Jan. 22, 2013 Accessed Jan. 22, 2013 from:



Braunerhjelm, P. & Thulin, P. (2008) Can countries create comparative advantages? R&D expenditures, high-tech exports, and country size in 19 OECD countries 1981-1999. International Economic Journal 22: 95-111. Accessed on 1/14/13 from

Buckley, C. (2004) Research labs power China's next boom. International Herald Tribune, September 13, 2004 Accessed Jan. 5, 2013 from:

Ekholm, K. and Hakkala, K. (2007) Location of R&D and high-tech production by vertically integrated multinationals. The Economic Journal 117: 512-643.

Foland, P. & Shelton RD. (2010). Why is Europe so efficient at producing scientific papers, and does this explain the European Paradox? 11th International Conference on S&T Indicators, Leiden, Sept. 2010. Accessed Jan. 4, 2013 from .

Jin, B. & Rousseau, R. (2005) China’s quantitative expansion phase: Exponential growth, but low impact. Proceedings of the 10th International Conference on Scientometrics and Informetrics, Stockholm, July, 2005.

Krugman, P. (1991) Geography and Trade, Cambridge: MIT Press.

Leydesdorff, L. (1990). The prediction of science indicators using information theory. Scientometrics 19, 297-324.

Leydesdorff, L., & Wagner, CS. (2009). Macro-level indicators of the relations between research funding and research output. Journal of Informetrics, 3(4), 353-362.

McGregor, J. (2011) China's drive for indigenous innovation: A web of industrial policies. Washington: U.S. Chamber of Commerce. Accessed 1/7/2013 from:



Moed, H. F. (2002), Measuring China’s research performance using the Science Citation Index. Scientometrics 53: 281- 296.

NSB (2012), Science and Engineering Indicators 2012. Arlington: National Science Foundation. NSB 12-1. Accessed Jan. 4, 2013 from

OECD (2012), Main Science and Technology Indicators. Paris: OECD. Volume 2012/1.

OECD-WTO (2013) OECD-WTO Trade in Value-Added Initiative Report. Accessed on 1/19/13:

Ricardo, D. (1817), On the Principles of Political Economy and Taxation. Accessed on 1/19/13 from:



Scott, RE (2012), The China toll. Washington: Economic Policy Institute Accessed 1/7/13 from



Shelton, RD & Monbo, SD. (2012) Input-output modeling and simulation of scientometric indicators: A focus on patents. Proceedings of the 17th International Conference on S&T Indicators, pp 756-767. Montreal, September, 2012.

Shelton, RD. & Leydesdorff, L. (2012) Publish or Patent: Bibliometric Evidence for Empirical Trade-offs in National Funding Strategies. Journal of the American Society for Information Science and Technology. 63:3, 498-511.

Shelton, RD., (2008), Relations between national research investment and publication output: Application to an American paradox. Scientometrics, 74(2), 191-205

Shelton, RD. & Foland, P. (2010), The race for world leadership of science and technology: Status and forecasts.  Science Focus 5, pp. 1-9 (Feb. 2010) in Chinese.  Also, Proceedings of the 12th International Conference on Scientometrics and Informetrics, pp. 369-380, Rio de Janeiro, July, 2009.  

Soete, L. (1987), The impact of technological innovations on international trade patterns: The evidence reconsidered. Research Policy, 16, 101-130.

Tebaldi, E. (2011), The determinants of high-technology exports: A panel data analysis. Atlantic Economic Journal 39:343-353.

UNESCO (2010) UNESCO Science Report 2010: The Current Status of Science around the World. Paris: The United Nations Educational, Scientific, and Cultural Organization.

Weller, CE. & Wheeler, H. (2008), Our Nation's Surprising Technology Trade Deficit, Center for American Progress, March 2008. Accessed 1/4/13 from:

World Bank (2012) World Development Indicators. Accessed 1/12/13 from:

Xing, Y. (2012) The PRC's High-Tech Exports: Myth and Reality, ADBInstitute Working Paper #357. Accessed 1/13/13 from

Appendix

For completeness here the remaining three high-technology sub-sectors are now discussed. As already seen in Fig. 6, their behaviors are quite different from one another, something to be kept in mind in drawing conclusions about the aggregate high-technology sector.

Figure 7 presents the trends in the aerospace sub-sector, which is very different from the earlier two in information technology (IT). Here, the U.S. continues to dominate the market, albeit with a decreasing market share due to European challenges. This may be attributed to the success of the Airbus consortium, which now produces about half of the world's airliners. China had only a 0.85% share in 2010, but this has more than doubled since 2003, and the PRC clearly intends to use access to its airliner market to carve out a larger share (Bradsher, 2013).

The situation in the pharmaceutical industry is also very different from that in the IT sub-sectors. Fig. 8 shows that the U.S. is far from being the major manufacturer in this field. Indeed the field is spread over multinational firms which manufacture at plants located in many countries. The OECD data shows that European firms are the major players, and it is surprising that small countries like Belgium and Switzerland have such leading positions in this sub-sector. The discontinuity in the Belgium curve in 2002 was due to international mergers and acquisition activity, which is common in this industry. The Chinese share is not growing so fast: it had 2.5% of the market in 2009, about the same as in 1995.

Finally, the instruments sub-sector shows a pattern in Fig. 9 similar to those in the IT sub-sectors: the rise of China (and S. Korea) cutting into the shares of the earlier leaders: the U.S., Germany, and Japan. However, the shift in dominance is not nearly so sharp or complete.

[pic]

Fig. 7. Export share in airliners and aerospace hardware of all kinds

[pic]

Fig. 8. Export share in pharmaceuticals

[pic]

Fig. 9. Export share in instruments

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download