Cisco Innovation Methodology

A new ranking of the world’s most innovative countries: Notes on methodology An Economist Intelligence Unit reportSponsored by Cisco

© Economist Intelligence Unit Limited 2009 A new ranking of the world’s most innovative countries:

Notes on methodology

1

Preface

In April 2009, the Economist Intelligence Unit published A new ranking of the world’s most innovative

countries. The report, sponsored by Cisco, analyses the updated innovation index, published originally in 2007, which ranks 82 countries based on their innovativeness. Notes on methodology is a companion document that explains the factors and process the Economist Intelligence Unit used to create the index.

April 2009

© Economist Intelligence Unit Limited 2009A new ranking of the world’s most innovative countries:Notes on methodology

2

Measuring innovation, especially at the level of the aggregate economy, has always been one of the most perplexing problems facing empirical economics. One approach that has often been adopted is to aggregate a large number of indicators refl ecting various aspects of science, education and technology and related factors into a single composite index. However, it is not clear what exactly is being measured on this basis. The indices combine, rather arbitrarily, indicators of both innovation inputs and outputs, as well as some indicators that appear to be only tenuously related to innovation activity. Such methodological muddle can lead to faulty conclusions.

Research and development (R&D) spending has been widely used as a measure of innovation performance. However, R&D is a measure of the inputs that go into the innovation process rather than of innovation output or success. A measure of innovation performance should focus on the extent of commercially successful applications rather than on the amount of effort going into developing them. In principle, patenting activity gets closer to this. Although the use of patent data has a number of problems, this is the single best available measure for innovation outputs. This conclusion is bolstered by an examination of alternative indicators of innovation performance that correlate closely with patent activity (see below).

In preparing the Economist Intelligence Unit’s innovation index, presented in A new ranking

of the world’s most innovative countries, we distinguish between measures of innovation outputs or performance, on the one hand, and of innovation inputs or drivers on the other. We construct three indices of innovation for 82 countries, drawing on the Economist Intelligence Unit’s business environment rankings (BER) model.

Innovation performance index. The fi rst index measures innovation output or performance, and is based on international patents data. We construct our measure as the sum granted to applicants (by residence) from the 82 economies by three major government patent offi ces—the European Patent Offi ce (EPO), the Japanese Patent Offi ce (JPO), and the US Patent and Trademark Offi ce (USPTO). The data are averaged over 2004-07 (2002-05 in the original version of our index) and expressed in terms of patents per million population for each country. USPTO and EPO data are for 2004-07; JPO data are for 2004-06, since 2007 data are not available.




3

Innovation enablers indices. There are also two composite indices, constructed on the basis of BER scores, that measure innovation inputs, or innovation enablers. The fi rst covers direct innovation inputs and the second the innovation environment or the broad economic, social and political backdrop that stimulates or hinders innovation activity. Indicators for the two enabler indices were chosen on the basis of factors identifi ed in the general literature.

We also constructed an index that aggregates the direct innovation inputs index and the innovation environment index. The weights used—.75 for direct inputs and .25 for the environment index—correspond to weights based on the estimated coeffi cients in the regression equation described below, which relates innovation performance to innovation inputs.

The patents data are averages for 2004-07, whereas the innovation enablers are indices based on 2004-08 averages (in the original version of our index the patents data were for 2002-05 and the enablers for 2002-06). All indices are expressed on a 1 to 10 scale using the following conversion formula:

9 * (indicator value for country - minimum value for the indicator)/(maximum value for the indicator

– minimum value for the indicator) + 1

Patents dataAlthough it is the single best available measure of innovation output, patent numbers are an imperfect proxy for overall innovation activity. Firms often choose to keep innovations that are commercially sensitive a secret; the propensity to patent may also vary according to the costs of patenting; and many patents may never be implemented commercially. Patents may even obstruct innovation on occasion if they slow the diffusion of knowledge or pose prohibitive barriers to market entry.

Inventions, moreover, do not all have the same value. The value distribution of patents is skewed: a few patents have a high value, whereas many have lower values. However, since there are no generally recognised, easily applicable methods for measuring the value of patents, researchers merely count the number of patents meeting various criteria.

International comparisons are also affected by differences in procedures and standards across patenting offi ces. For example, in Japan, a different patent application had until recently to be submitted for each claim; in other countries multiple claims can be made in each application. This helps in part to explain the much larger level of patenting applications in Japan.

One patent measure that reduces some of these problems, and that has been widely used in international comparisons, is that of the so-called triadic patent families. Triadic patents are those that have been applied for at the EPO, the JPO, and granted by the USPTO to protect the same invention. The triadic patents are counted on the basis of the earliest priority year—the year in which a patent was fi rst


4

Triadic patents per million population 2005 1995 2005 1995Japan 117.21 75.31 Ireland 14.95 7.93Switzerland 107.56 103.82 Italy 12.33 10.65Sweden 81.01 83.66 Slovenia 6.36 3.52Germany 76.38 58.53 Taiwan 5.01 0.78Netherlands 66.94 47.98 Spain 4.55 2.02Israel 60.28 27.96 Hungary 4.06 2.42Korea 58.40 6.99 Czech Republic 1.54 0.27US 53.11 45.00 Portugal 1.07 0.32Finland 53.04 60.77 South Africa 0.63 0.60Luxembourg 50.48 31.71 Russia 0.44 0.34Denmark 42.18 35.12 Turkey 0.36 0.03Austria 39.70 27.54 Poland 0.34 0.11France 39.35 32.84 Brazil 0.31 0.09Belgium 34.44 36.32 China 0.27 0.02UK 27.41 26.46 Argentina 0.23 0.17Norway 25.59 19.76 Mexico 0.16 0.13Singapore 24.31 6.50 India 0.12 0.01Canada 24.04 12.54 Australia 18.74 12.36 OECD average 42.97 31.95New Zealand 15.32 5.71 EU27 37.52 31.18

Source: OECD

applied for at any patent offi ce.However, although triadic patents are in some ways easier to compare across countries, they cover only

a small subset of total patents. They are also biased towards high-technology fi elds and thus may present a skewed picture of total innovation performance. We thus construct another patent measure as the sum of patents applied for by, or granted to, a country’s applicants by regional centres—that is the USPTO, the EPO and the JPO. This measure differs from the triadic patents measure in that patents do not have to have been fi led in all three offi ces to be counted.

It has been argued that there is an upward bias in JPO patent data, which will bias upward data for patents granted to Japanese innovators in particular. However, Japan would be only marginally behind the US even if we took only USPTO data (as opposed to combing EPO, USPTO and JPO data); Japan comes top on triadic patents and also on our alternative composite innovation performance indicator. We have nevertheless corrected for a possible “Japan patents effect” by assuming that the maximum value that the index can take, with a top score of 10, is based on 800 patents per million population. This is below the actual value for Japan of 1,275 to allow for the fact that until recently Japan required a different patent application for each claim.

Our patents-based measure correlates well with other proxies for innovation performance. In our



5

original study we looked at data on citations from scientifi c and technical journals; UNIDO estimates for 2000 of the share of medium- and high-technology products in a country’s manufacturing output and its manufacturing exports; and the results of a survey question from the World Economic Forum’s Global Competitiveness Report that asked respondents to rate the extent to which companies were adept at, or able to absorb, new technology. Up-to-date data on article citations and medium-and high-technology output and exports was not available. We thus added some other indicators. Our new set of alternative indicators of innovation output consist of: ■ high-technology manufacturing output per head in constant 2000 US$. The data is for 2006 from

Science and Engineering Indicators 2008 published by the US National Science Foundation.■ high-technology services output per head in constant 2000 US$. The data is for 2006 from Science and

Engineering Indicators 2008 published by the US National Science Foundation.■ royalty and licence fee receipts as a percentage of GDP. Data are for 2006 from the World Bank’s World

Development Indicators 2008.■ a survey question in the World Economic Forum’s Global Competitiveness Report that asked

respondents to rate the extent to which companies in 134 countries were adept at, or able to absorb, new technology (this indicator was also used in our previous report), as well as an additional survey question that asked companies to evaluate the extent to which countries relied on domestic, as opposed to imported, innovation.

We have constructed a composite measure of innovation performance that consists of these indicators as well as our patents measure of performance. The composite index gives very similar results to the “patents only” measure. For example, Japan remains fi rst-ranked and Switzerland second. The US, Sweden and Finland are also in the top 5. Thus we can examine the “patents only” index with considerable confi dence that it truly refl ects a country’s innovation performance.

What factors lead to innovation on a national level? There are two types of factors: direct and indirect drivers of innovation.

Direct innovation inputsThe innovation inputs index is based on an unweighted average of the following BER indicators:■ R&D as a percentage of GDP■ quality of the local research infrastructure■ education of the workforce■ technical skills of the workforce■ quality of IT and communications infrastructure ■ broadband penetration


6

All indicators are rated on 1-5 scale based on original raw values where relevant and as defi ned in the methodology at the end. The composite index is converted to a 1-10 scale.

Innovation environmentInnovation performance is not only determined by direct inputs. It is also infl uenced by a wide range of policies and attributes of the business environment. The policies include both those specifi cally designed to strengthen innovation outcomes and those that are guided by other concerns, but may still have an impact on innovation. The environment for innovation is likely to be especially sensitive to the degree of protection of intellectual property rights; the extent of tax-based incentives for innovation; competition policy; product and labour market regulations; the development of fi nancial markets; and the openness of the economy to trade and investment.

Recent OECD studies have found empirical support for the following propositions: ■ a strengthening of intellectual property rights leads to an increase in the propensity to patent;■ tight product and labour market regulations can reduce signifi cantly innovation activity;■ the propensity to patent is sensitive to foreign investment restrictions;■ stable macroeconomic conditions and low real interest rates help to encourage innovation activity;■ developed fi nancial systems, especially equity-based ones, are associated with greater innovation

activity.

The importance of the diffusion of knowledge across national borders is often emphasised. Open economies will have a greater exposure to the foreign knowledge stock. A number of studies suggest that foreign knowledge can diffuse more rapidly through the direct transmission of ideas rather than through trade in goods and services that embody them. So the international mobility of skilled researchers and multinational fi rms may be especially important channels for knowledge transfer. However, empirical work has also suggested that the positive effects of diffusion are offset partially by signifi cant negative effects on innovation from trade openness. This suggests that greater openness helps to make new knowledge accessible, but also reduces the ability to undertake domestic innovation. This may be an important factor in explaining the domestic innovation performance of some extremely open European economies.

We thus constructed an index for the indirect inputs, the broad economic, social and political factors that facilitate (or hinder) innovation activity. The innovation environment index is based on the following 12 factors, weighted on the basis of survey results on the relative importance for innovation of the various factors.



7

The source for all the indicators is the BER, with the exception of popular attitudes toward science, which is taken from the World Values Survey, compiled by a worldwide network of social scientists since 1981.

Explaining innovation performanceThe power of our two input indices in explaining innovation performance across countries is very high. A regression for the 82 countries shows that the two indices explain 93% of the inter-country variation in the index based on patents per million inhabitants. The direct inputs index has a more powerful infl uence on international patenting, but the innovation environment also has a signifi cant impact.

Dependent variable: International patents index 2004-07 Coeffi cients t StatConstant 0.15 0.515Innov1 0.237 3.168Innov2 0.735 16.151 Adj R2 0.932 N 82

Innov1: Innovation environment index 2004-08Innov2: Direct innovation inputs index 2004-08

Weights for the innovation environment indexPolitical stability 0.109Macroeconomic stability 0.089Institutional framework 0.107Regulatory environment 0.108Tax regime 0.075Flexibility of labour market 0.072Openness of national economy to foreign investment 0.083Ease of hiring foreign nationals 0.069Openness of national culture to foreign infl uence 0.063Popular attitudes towards scientifi c advancements 0.058Access to investment fi nance 0.056Protection of intellectual property 0.112


8

Innovation effi ciencyA comparison of a country’s rank on its innovation output with its ranking on direct innovation inputs can provide an insight into the effi ciency of innovation. A large discrepancy in the two rankings suggests either a high level of effi ciency (high innovation output relative to inputs) or a high degree of ineffi ciency if the direct inputs rank exceeds signifi cantly a country’s ranking on innovation performance.

Forecasting innovation performanceThe estimated equation above can also be used to predict innovation performance over the next fi ve years. Forecast values of the two innovation input indices, based on the Economist Intelligence Unit’s BER forecast scores, are inserted into the equation to yield forecasts of innovation performance (based on expected patents activity) in 2009-13 (the procedure assumes that estimating errors for the performance equation in 2004-07 also apply in 2009-13).

International patents index in some major economies* Predicted ActualBrazil 5.89 5.71China 5.65 5.39France 9.23 8.88Germany 9.37 9.40India 5.14 5.31Mexico 5.69 5.72Netherlands 9.18 9.16Russia 6.48 6.03Singapore 8.68 8.76South Korea 8.63 8.94Spain 7.77 7.51Sweden 9.51 9.44Switzerland 9.48 9.71UK 8.97 8.72USA 9.42 9.50

*2004-07 average.

Whilst every effort has been taken to verify the accuracy of this information, neither the Economist Intelligence Unit Ltd. nor the sponsor can accept any responsibility or liability for reliance by any person on this report or any of the information, opinions or conclusions set out herein.De

sign

: Mik

eKen

ny@

mac

.com

C

over

imag

e: iS

tock

phot

o.co

m

LONDON26 Red Lion SquareLondon WC1R 4HQUnited KingdomTel: (44.20) 7576 8000Fax: (44.20) 7576 8476E-mail: [email protected]

NEW YORK111 West 57th StreetNew York NY 10019United StatesTel: (1.212) 554 0600Fax: (1.212) 586 1181/2E-mail: [email protected]

HONG KONG6001, Central Plaza18 Harbour RoadWanchai Hong KongTel: (852) 2585 3888Fax: (852) 2802 7638E-mail: [email protected]

Documents

Cisco Innovation Methodology