Center for Research on Science, Technology and Society

Luiz Ricardo Cavalcante *

There is a widespread consensus that innovation represents a “crucial source of effective competition, of economic development and the transformation of society”.[1] Based upon this perception, several countries adopt innovation policies and fix targets to increase their R&D investments as a way of reaching higher levels of development. The Lisbon strategy and the Europe 2020 strategy, for example, stated that 3% of the European Union’s gross domestic product (GDP) should be invested in R&D. More recently, the Biden administration proposed spending $250 billion on the U.S. research enterprise over the next years “as part of a plan to rebuild the country’s infrastructure, create jobs, and outinnovate the rest of the world”.[2] Developing countries have also established targets for their R&D expenditures or for their business enterprise R&D expenditures, as in the case of Brazil in the 2010s.

Brazil invests more than Latin-American and most BRICS countries in R&D, but much less than developed countries and China

Brazil has been adopting science, technology, and innovation (ST&I) policies for some decades and managed to increase its R&D efforts, although still lagging behind developed countries. In 2017, R&D investments in Brazil as a percentage of GDP reached 1.26%. Although higher than the figures in other Latin-American economies such as Argentina and Mexico and in the so-called BRICS (except China), R&D investments in Brazil are much smaller than in developed economies. Even when compared with countries like Australia and Canada, which are relevant commodities exporters like Brazil, R&D investments in the country are reduced. China is advancing very rapidly towards developed countries indicators of R&D investments. In 2017, that country invested more in R&D as a percentage of GDP than Australia and Canada.

Given its relatively low levels of R&D investments, Brazil must not only raise them, but also increase their efficiency in order to convert local efforts into scientific and technological results. However, the efficiency of R&D investments is not a frequent subject in the literature. As a result, in order to measure and compare R&D efficiency indicators, it is necessary to define them first.

R&D efficiency can be measured as the ratio between R&D outputs and R&D inputs

The efficiency of R&D efforts can be generically defined as their capacity of generating scientific or technological results. Such kind of indicator has been subject of a reduced number of analysis in the literature and has had little impact on ST&I policies, which tend to focus on the increase of investments (and not on their efficiency). However, a few papers did analyze R&D efficiency and defined it as the ratio between some output indicator (such as scientific publications or patents) and some input indicator (such as R&D investments or number of researchers). An average lag between inputs and outputs is usually assumed before computing the ratios.

In this article, we considered the following input indicators:

• Gross domestic expenditure on R&D financed by the government, the higher education and other sources except business enterprises.
• Total R&D personnel full-time equivalent (FTE) employed by the government, the higher education and other except business enterprises.
• Gross domestic expenditure on R&D financed by business enterprises.
• Total R&D personnel FTE employed by business enterprises.

And the following output indicators:

• Number of citable documents.
• Number of patent cooperation treaty (PCT) applications.

These indicators are combined in such a way that i) efforts not financed or employed by business enterprises are associated with citable documents (generating the “scientific efficiency” indicators), whereas ii) efforts financed or employed by business enterprises are associated with patent applications (generating the “technological efficiency” indicators). We also assumed a three-year lag between R&D inputs and outputs. In order to compute these indicators, we collected data from United Nations Educational, Scientific and Cultural Organization (UNESCO), SCImago and World Intellectual Property Organization (WIPO). Other indicators and sources could also have been used. Although eventually leading to different absolute numbers, they do not change significantly the country efficiency rankings discussed in the remainder of this article.

In 2018, Brazil produced 3.76 citable documents per previous US$ million R&D expenditures mainly financed by the government

The scientific efficiency of R&D expenditures calculated on the basis of the previous discussion for the selected countries is indicated in the graph below. We calculated the number of citable documents per US$ million R&D expenditures mainly financed by the government and considered a three-year average lag between inputs and outputs. As a result, data regarding the citable documents refer to 2018, whereas data regarding expenditures refer to 2015. The indicator measures the number of papers generated as a result of US$ 1.0 million R&D investments mainly directed to scientific research.

Scientific efficiency of R&D expenditures (2018/2015)

Source: elaborated by the author.

As shown in the graph, Brazil produced 3.76 citable documents per US$ million R&D expenditures mainly financed by the government. As a result, the scientific efficiency of R&D expenditures in Brazil is about ⅓ of Australia’s, 50% of China’s and 75% of the United States. Brazil’s indicator is very close to the other Latin-American countries (Mexico and Argentina). The relatively low efficiency of the United States seems puzzling, even when compared with other non-English-speaking countries like Russia and China (especially taking into account that several papers point to a sort of “language bias” of scientific publications in favor of English-speaking countries). At any rate, the results indicate that the scientific efficiency of R&D expenditures in Brazil is low. It might be associated, for example, with the R&D infrastructure (i.e., some sort of capital stock which can increase scientific product even in the absence of large recent investments) and with increasing returns on scale (i.e., countries that invest more in R&D tend to present higher levels of efficiency). These hypotheses, however, are not tested in the brief analysis that deals only with a reduced sample of countries.

As for the scientific efficiency of R&D personnel (number of citable documents produced by researchers employed mainly by the government), Brazil also ranks among the less efficient countries in the sample. Although more efficient than Argentina and Russia, Brazil showed a scientific efficiency of R&D personnel of about half of several other countries’ (including China and Mexico) and of a little more than one third of Unite States’ and Canada’s. Of course, efficiency (or “productivity”) is related not only to the R&D infrastructure but also to the investment stock. Researchers working in modern labs equipped with modern equipment obviously tend to publish more than researchers working at precarious conditions. Besides, the pulverization of resources might contribute to lowering Brazil’s scientific efficiency of R&D personnel. Although in this article we just point to the problem (and not to its causes), the role of all these factors must be investigated in order to tackle with it.

The performance of Brazil regarding the technological efficiency of both R&D expenditures and R&D personal is even worse than the scientific efficiency

Similar exercises have been conducted for the technological efficiency of R&D expenditures and of R&D personnel. In both cases, the distance between Brazil and the leading countries is higher than in the case of the scientific efficiency. Regarding the technological efficiency of R&D expenditures, we considered the number of PCT applications in 2018 per US$ billion R&D expenditures financed by the business enterprise sector in 2015. The results are shown in the graph below.

Technological efficiency of R&D expenditures (2018/2015)

Source: elaborated by the author.

In Brazil, we estimated about 37 patent applications per US$ billion R&D expenditures financed by the business enterprise sector. This figure is about 12% of Germany’s or South Korea’s, the leading countries in the sample. Countries like Mexico, China, Australia and Canada presented technological efficiencies of R&D expenditures more than five times the indicator estimated for Brazil.

As for the technological efficiency of R&D personnel, we estimated that a thousand researchers employed by business enterprises in 2015 are associated with about 12 patents in 2018. Although more efficient than Argentina (9) and Russia (5), Brazil lags behind the other countries in the sample. Brazil’s technological efficiency of R&D personnel is much smaller than Mexico’s (27), India’s (27) and Canada’s (25), a quarter of several countries’ in the sample and less than 15% of Germany’s (86).

Of course, a series of caveats must be considered when dealing with these data. It is well known that patents are not a measure of innovation (it is basically a measure of invention). Besides, high-tech industries tend to patent more than low-tech industries; as a result, the productive structure influences the number of patents in a given country. In spite of limitations like these ones, several papers use the number of patent applications or concessions as a proxy of the technological performance of countries and firms. At any rate, the results indicate that there is a long road ahead if the country wants to improve the technological efficiency of its R&D efforts.

No sign of improvement over the last years

So far, the analysis focused on cross-country comparisons in a given moment. No time-series have been estimated for the countries in the sample, but specifically in the case of Brazil, we used data available in the web page of the Ministry of Science, Technology, and Innovation in order to compute the scientific and the technological efficiency of R&D expenditures over the last decade. The sources are not necessarily the same used so far (for example, for patents, the data refer to USPTO applications), but the results are reasonably comparable with the ones previously reported in this paper. In short, there is no sign of improvement during the last ten years. In the case of the scientific efficiency, there seems to be a steady (yet slight) decrease, whereas in the case of the technology efficiency no clear trend is to be observed. The erratic nature of the output indicator in this case (patent applications) may suggest that future works could use some sort of moving average instead of spot data in order to take this problem into account.

Some concluding remarks

The main conclusion of this article is that Brazil, besides investing less than other reference countries in R&D, is also less efficient in converting these investments into scientific publications and patents. An immediate consequence of the results reported above is, of course, the analysis of the reasons behind the low R&D efficiency in Brazil. Several possible explanatory variables have been suggested in this article and in the literature. For example, the research infrastructure (labs ad equipment) may affect the number of published papers and a language bias in favor of English-speaking countries may be observed. As for the technological efficiency, patent applications are related to the productive structure and to the property rights protection patterns, and, as a result, affect R&D technological efficiency. In both cases (i.e., in the scientific and in the technological ones) scale economies tend of positively affect efficiency (similarly to the relationship between productivity and investments). Future regressions may confirm that R&D intensity (for example, R&D investments as a percentage of GDP) affects R&D efficiency. At any rate, the analysis of the reasons behind the low R&D efficiency in Brazil can have a considerable impact in the designing of ST&I policies in the country.

* Legislative advisor at the Brazilian Federal Senate and professor at the Brazilian Institute of Education, Development and Research (IDP).

 Acknowledgments: Eduardo Viotti, Fernanda De Negri, Paulo Meyer Nascimento, Pedro Fernando Nery and the team of the Center for Research on Science, Technology and Society.

[1] FREEMAN, C. A Schumpeterian renaissance? In: HANUSCH, H.; PYKA, A. Elgar Companion to Neo-Schumpeterian Economics. Cheltenham (UK): Edward Elgar Publishing Limited, 2007. p. 130-141.

[2] https://www.sciencemag.org/news/2021/03/biden-proposes-250-billion-investment-research