Recent empirical evidence has demonstrated a positive link between an industry’s proximity to the technological frontier and its R&D effort.1 The idea is that R&D intensity increases close to the technological frontier because firms’ survival depends on their ability to innovate. These results have important policy implications, as they reflect the virtuous cycle between research efforts, innovation and economic growth. It provides an additional justification for a more acute support to R&D by policy-makers and underlines the R&D component of the so-called ‘Lisbon Agenda’, which was to achieve a higher level of R&D.

Two sub-targets for R&D spending were clearly defined in 2002:

  • EU R&D intensity (R&D expenditure divided by GDP) was to increase from about 1.8% in the late 1990s to about 3% by 2010; and
  • two-thirds of this spending was to be funded by the business sector, the rest being funded by governments.
A failed ‘R&D’ agenda…

Figure 1 illustrates the failure of Europe regarding its relative research efforts or a dramatic stability of Europe’s total R&D activities over the past 25 years. The US and Japan exceed 2.5%, whereas Europe seems to be well-rooted around 1.8%.

Figure 1. R&D intensity of selected EU and non-EU countries, 2006

Source: OECD, MSTI, 2007 and adapted from van Pottelsberghe (2008).

The intensity of R&D spending across EU member states varies considerably, with some countries having reached relatively high levels, especially Finland and Sweden, which several years ago leapfrogged the 3% threshold. Denmark, Austria and Germany are around 2.5 %, whereas France is just above 2%. However, the vast majority of countries have an R&D intensity of well below 2% of GDP, with a median of 1.2%. Seven American states have an R&D intensity that is higher than 4%, against none for the EU.

... partly due to reduced government funding

No EU member state has fulfilled its commitment, as no country devotes one percent of its GDP to funding public (higher education, laboratories) or business (through subsidies and procurement) research activities. The Lisbon ‘failure’ may actually be partly explained by the drop in government-funded R&D observed in several large European countries. Interestingly, a drop also occurred in the US and Japan over the same period, but a more than proportional increase in business-funded R&D compensated, unlike in Europe.

... and to the industrial structure

In addition to government spending, industrial specialisation is critical to research activities. A country specialised in the finance industry (e.g., Luxemburg) would not need a high level of R&D expenditure in order to ensure growth – at least as commonly measured. Similarly, a country specialised in the tourism, fashion, services or food industries would have a lower R&D intensity than a country specialised in the pharmaceutical, engineering or biotech industries. Traditional country benchmarking of R&D intensity therefore warrants a substantial degree of caution. The role of specialisation is an important issue, as some countries generally praised for their above-average R&D intensity may actually not be performing particularly well given their specialisation in R&D-intensive industries.

In a new policy brief (Mathieu and van Pottelsberghe 2008), a co-author and I try to evaluate the extent to which industrial specialisation may affect assessments of national R&D performance. We use a panel data set of industry-specific R&D intensity for about 20 industrial sectors in 10 countries over the period 1991-2002. The results lead to three observations:

  • Technological specialisation explains the variation in R&D intensity much better than any other country characteristics.
  • Omitting industrial specialisation may lead to a highly skewed ranking of countries.
  • When industrial specialisation is taken into account, only Sweden and the US still outperform other countries. Neither Japan nor Finland has an R&D intensity that is particularly high in relation to what their industrial structures would suggest.

In a nutshell, business R&D intensity is endogenous, not exogenous. Governments should therefore go beyond traditional incentive policies such as direct R&D subsidies or R&D tax credits. To set a business-funded R&D target at the country level is thus either wishful thinking or an implicit industrial policy – a way to alter the country’s industrial structure. At the EU level, technological specialisation has not evolved much towards R&D-intensive industries, which explains the lack of ‘visible progress’ over the past few years. This technological specialisation factor is taken by the Commission to explain both European R&D ‘inertia’ (business R&D intensity has been very stable over the past twenty years) and the EU’s lagging behind US business R&D intensity. However, our results suggest that when the technological specialisation of countries is taken into account, Sweden and, to a lesser extent, the United States display above-average R&D intensity. Something other than technological specialisation also seems to drive R&D intensity in these “exceptions”.

An integrated market for innovation?

Larger markets presumably result in a higher expected return to R&D, and hence higher expenses. The idea that there is a positive relationship between the size of a country and its propensity to invest in R&D is empirically and theoretically supported by Guellec (1999) and Desmet and Parente (2006). This market-size hypothesis may explain why the US has an above-average R&D intensity (or larger than its industrial structure would suggest). The US benefits from a huge and homogeneous market, with one main language and one regulation. In Europe, sending a product from Amsterdam for sale in Brussels is still considered an ‘export’, whereas in the US a product made in New York and sold in Los Angeles is labelled ‘distribution’. Besides these proverbial examples, a large body of evidence has been published on the lack of European integration. Emblematic of this lack is the way the innovation system works in Europe. The European patent system, and hence the European market for technology, is highly fragmented. Once a patent has been granted by the European Patent Office (EPO), it must be validated, translated, monitored and enforced in all relevant national patent offices. This explains why the simulations performed by van Pottelsberghe and François (2006), and presented as well in Guellec and van Pottelsberghe (2007) show that a patent examined by the EPO and then enforced in 13 European countries costs about 11 times more than a patent in the US, and 14 times more than a patent granted in Japan (cf. Table 1).

Table 1. European patent costs (enforced in 13 countries) relative to the US and Japan

  Cumulated fees and translation costs2 Total cost for 20 years3 (**)
US 11 9
Japan 14 7

Source: Adapted from van Pottelsberghe and François (2006). These figures represent the simulated costs of a European patent divided by the simulated cost of an average patent in the US and in Japan.

The policy implication is straightforward. The failure to create an EU patent places a heavy burden on the shoulders of European innovators and entrepreneurs at the very beginning of the innovation process - a clear comparative disadvantage for Europe with respect to the US and Japan.

More and better academic research?

Market size does not explain the performance of Sweden. Its R&D intensity is probably linked to its spending on academic research, the highest (as a percentage of GDP) in the whole OECD area, as illustrated in Figure 2. This strong emphasis on academic research is also a stimulus for business R&D: universities generate new ideas that are then transferred to the private sector. The transformation of these ideas into products or processes requires further applied research activity and development. Not surprisingly, the four countries in Figure 2 with the highest academic R&D intensities are also the four countries with the highest business R&D intensities. Provided effective technology transfer systems are put in place, academic research is probably the most effective source of new ideas, which in turn induce further research in the business sector. Guellec and van Pottelsberghe (2004) provide evidence suggesting that the social return to academic research is indeed higher than the social return to business R&D. In this respect, the European Research Council (ERC), which provides merit-based fundamental research grants, is a recent positive example of what the EU can achieve.

Figure 2. Business-funded R&D and R&D carried out in institutions of higher education as a percentage of GDP, 2006 or closest

Source: OECD, MSTI, 2007. and adapted from van Pottelsberghe (2008).

Not only does academic research feed ideas to the market, but it also attracts more funding from the business sector and promotes the setting up of scientific clusters. For instance, Abramovsky et al (2007) show that, in the UK, universities with a high scientific output attract significantly more local and foreign research laboratories to their neighbourhood. This question is key because gaining a technological edge is the main driving force behind foreign business R&D investment, be it in the US, in Europe, or elsewhere. In fact, large firms nowadays increasingly invest in emerging markets, which provide a high-quality labour force at much lower cost than in Europe. If Europe wants to reverse that trend and increase its R&D intensity, it needs to heed the lessons from the United States and Sweden.


Abramovsky L., R. Harrison and H. Simpson, 2007, University Research and the Location of Business R&D, The Economic Journal, 117 (519), pp. 114-141.

Acemoglu D., Aghion P., and F. Zilibotti (2006) Distance to Frontier, Selection, and Economic Growth, Journal of the Economic European Association, 4(1), 37-74.

Desmet K. and S. L. Parente (2006), Bigger is better: market size, demand elasticity and resistance to technology adoption, CEPR Discussion Paper 5825, September.

Guellec D. (1999), A la recherche du tant perdu, Revue Française d’Économie, 14(1), pp. 117-169.

Guellec D. and B. van Pottelsberghe de la Potterie (2004), From R&D to productivity growth: do the institutional settings and the sources of funds of R&D matter? Oxford Bulletin of Economics and Statistics, 66(3), pp. 353-376.

Guellec, D. and B. van Pottelsberghe de la Potterie (2007), The economics of the European patent system, Oxford University Press, Oxford, 250 p.Acemoglu, Aghion and Zilibotti (2006).

Mathieu A. and B. van Pottelsberghe de la Potterie, 2008, A note on the drivers of R&D intensity, CEPR Discussion Paper 6684.

van Pottelsberghe de la Potterie B., 2008, Europe’s R&D: missing the wrong targets?, Bruegel Policy Brief, Issue 2008/03, February.

van Pottelsberghe de la Potterie B. and D. François, 2006, The cost factor in patent systems, CEPR Discussion Paper 5944.


1 Acemoglu, Aghion and Zilibotti (2006).

2 The costs include the expenses (fees and translation costs) for a patent examined by the European Patent Office (EPO) and validated in 13 European countries after granting.

3 The total cost for 20 years also includes the renewal fees for 20 years in 13 European countries. These costs are related to the absolute cost of an average patent. The recently ratified London Protocol will reduce translation costs somewhat.

210 Reads