ArticlePDF Available

A New Indicator of Technological Capabilities for Developed and Developing Countries (ArCo)

February 2004
SSRN Electronic Journal 32(4):629-654

Source
RePEc

Authors:

Italian National Research Council

This paper devises a new indicator (ArCo) of technological capabilities that aims at accounting for developed and developing countries. Building on similar attempts as those devised by UN Agencies, including the UNDP Human Development Report's Technology Achievement Index (TAI) and UNIDO's Industrial Performance Scoreboard, this index takes into account a number of other variables associated with technological change. Three main components are considered: the creation of technology, the technological infrastructures and the development of human skills. Eight sub-categories have also been included. ArCo also allows for comparisons between countries over time. A preliminary attempt to correlate ArCo to GDP is also presented.

A composite index of technological capabilities across countries (ArCo), 1990-2000

…

(continued)

…

Correlation coefficients across the various indexes of technological capabilities (all countries) a

…

Correlation coefficients across the category indexes of technological capabilities

…

Coefficients of variation of the various indexes of technological capabilities

…

Figures - uploaded by Daniele Archibugi

Content may be subject to copyright.

Content uploaded by Daniele Archibugi

Content may be subject to copyright.

A New Indicator of Technological Capabilities

for Developed and Developing Countries (ArCo)

DANIELE ARCHIBUGI

London School of Economics and Political Science, UK

Italian National Research Council, CNR, Rome, Italy

and

ALBERTO COCO

Bank of Italy, Rome, Italy

Universit

eCatholique de Louvain la Neuve, Belgium

Summary. — This paper devises a new indicator (ArCo) of technological capabilities that aims at

accounting for developed and developing countries. Building on similar attempts as those devised

by UN Agencies, including the UNDP Human Development Report’s Technology Achievement

Index (TAI) and UNIDO’s Industrial Performance Scoreboard, this index takes into account a

number of other variables associated with technological change. Three main components are

considered: the creation of technology, the technological infrastructures and the development of

human skills. Eight subcategories have also been included. ArCo also allows for comparisons

between countries over time. A preliminary attempt to correlate ArCo to GDP is also presented.

Key words — technology creation, infrastructures, human skills, development index

1. INTRODUCTION: SCOPE, RELEVANCE

AND ASSUMPTIONS

Technological capabilities have always been

a fundamental component of economic growth

and welfare. One of their key characteristics is

that they are far from being uniformly distri-

buted across countries, regions and ﬁrms.

Knowledge production is largely concentrated

in a few highly industrialized countries. The

access to new and old knowledge, in spite of

international trade, communications, foreign

direct investment, public policies promoting

scientiﬁc cooperation and many other channels

of technology transfer, is a long way away

from being geographically homogenous. A few

countries constantly upgrade their knowledge-

base while the majority of them lag behind and

have many diﬃculties absorbing capabilities

that are already considered obsolete in other

parts of the world.

The determinants of the generation, trans-

mission and diﬀusion of technological inno-

vation have been studied both from the

theoretical and empirical viewpoint in a large

body of literature (Pietrobelli, 2000). But the

www.elsevier.com/locate/worlddev

World Development Vol. 32, No. 4, pp. 629–654, 2004

Printed in Great Britain

0305-750X/$ - see front matter

doi:10.1016/j.worlddev.2003.10.008

Preliminary versions of this paper were presented at

the Workshop on Economic Impact of Innovation and

Globalization, Turin, 13 June 2002, at the Master in

Science, Technology and Society, University of Buenos

Aires Quilmes, October, 8 2002, at the Doctoral Pro-

gram on Economics and Management of Technological

Change, University of Madrid Complutense, January

28, 2003, at the ECA Knowledge Economy Unit, the

World Bank, Washington, DC, May 8, 2003. We also

wish to thank Kim Bizzarri, Liliana Herrera Enriquez,

Francesco Lissoni, Richard Nelson, Mario Pianta, Carlo

Pietrobelli, Giuseppe Zampaglione and two anonymous

referees for detailed comments to previous versions. We

also wish to thank a large number of national statistical

experts who provided information, statistics, and com-

ments to complete and/or corroborate the ArCo data-

base. Financial support from MIUR Project 2001, on

Technological innovation and economic performance,

No. 2001133591, University of Urbino, is gratefully

acknowledged. Final revision accepted: 7 October 2003.

629

current understanding of the devices of tech-

nology creation and transfer is still inadequate,

in part due to the lack of detailed indicators of

technological change. This paper presents a

new index of technological capabilities, ArCo,

for a large number of countries. It builds on

many lessons learned of the nature of techno-

logical change and on other previous attempts

to measure it, including the latest Technology

Achievement Index (TAI) presented by the

UN’s Human Development Report (UNDP,

2001) and the UN Industrial Performance

Scoreboard (UNIDO, 2003).

Among the lessons learned on the measure-

ment of technological capabilities, we wish to

recall the following:

––The technological capabilities of a country

are composed of a variety of sources of

knowledge and of innovation. A comprehen-

sive measure should be able to account for

the activities that are codiﬁed as well as for

those that are tacit (Lundvall, 1992). Some

of the capabilities are disembodied, such as

new ideas and inventions. Others are embod-

ied in equipment, machinery and infrastruc-

tures, while still others are embodied in

human skills (Evangelista, 1999; Pianta,

1995; Smith, 1997).

––Technological capabilities are composed

of clusters of innovations associated with

diﬀerent waves of industrial development

(Freeman & Louta, 2001).

––The integration of new technology sys-

tems requires the mastering of previous tech-

nologies, allowing economic agents to build

competencies in a cumulative manner (Bell

& Pavitt, 1997; Pavitt, 1988a). Often new

systems make previous ones obsolete (Juma

& Konde, 2002). As Schumpeter remarked,

‘‘add as many mail-coaches as you please,

you will never get a railroad by doing so.’’

––The various sources of technological cap-

ability are more likely to be complementary

rather than interchangeable. First rate infra-

structures devoid of a suﬃciently qualiﬁed

labor force will be useless and vice versa (Ab-

ramovitz, 1989, Maddison, 1991). Moreover,

successful integration among the various

waves of innovations has the eﬀect of multi-

plying its economic and social impact (Anto-

nelli, 1999; Amable & Petit, 2001).

––The creation and improvement of techno-

logical capabilities involve a crucial element

of technological ‘‘eﬀort.’’ Access to ad-

vanced technology is a necessary condition,

but it needs to be accompanied by substan-

tial and purposeful investments for it to be

absorbed, adopted and learned (Pietrobelli,

1994; Lall, 2001a).

––Since the diﬀerences across countries’

technological capabilities are colossal, a

measure to account for them meaningfully

should consider the components that are

speciﬁc to both developed and developing

countries (Lall, 2001a).

Our work has been inspired by a variety of

attempts to generate measures of technological

capabilities. Even when we departed from

previous statistical exercises, we beneﬁted

from their methodology. In particular, we

wish to mention, besides the already cited

Technology Achievement Index (UNDP,

2001) and the Industrial Development Score-

board (Lall & Albaladejo, 2001; UNIDO,

2003), also the Technology Index of the World

Economic Forum’s Global Competitiveness

Report (WEF, 2002), and the critical analysis

by Lall (2001b). Throughout the paper, we

specify when we have followed these approa-

ches and when, and why, we have opted for

alternative paths.

It should be noted that statistics of techno-

logical activities for the restricted group of the

30 most developed countries could be much

more sophisticated in terms of coverage and

signiﬁcance. For this group of leading coun-

tries, many more indicators are available (and

the quality of the data is much more satisfac-

tory than for other countries). If we were to

limit our analysis to this restricted number of

countries, we would have used diﬀerent indi-

cators and methodology (for a discussion of

the various attempts to measure scientiﬁc and

technological capabilities of advanced coun-

tries see Archibugi & Pianta, 1992; Patel &

Pavitt, 1995). It is hardly surprising that data

for the selected number of countries that con-

centrate the bulk of inventive and innovative

activities are much richer. The attempt here is

to provide measures for a much larger group of

countries which, as a whole, have a much more

limited level of technological capabilities.

Monitoring the existing capabilities will permit,

to identify of the nature and intensity of the

technology gap and the appropriate strategies

to bridge it.

This analysis is based upon a number of

assumptions. First, we assume that a compar-

ative analysis across countries is meaningful

(Sirilli, 1997). In spite of the enormous diﬀer-

ence across countries (how can one describe in

a single number the technology gap between

WORLD DEVELOPMENT630

Switzerland and Somalia?), countries can be

compared. But we also assume that a battery of

indicators could provide a more comprehensive

picture of the diﬀerences than a single indicator

would. The statistics produced achieve greater

signiﬁcance when considering homogeneous

groups of countries and allow comparisons

between countries geographically, culturally

and economically close to each other, (such as,

for example, Switzerland and Germany,

Somalia and Ethiopia. For a discussion, see

Pietrobelli, 1994).

Second, we assume that a country-level

analysis still proves useful despite the enormous

diﬀerences found within countries. Synthetic

indicators for countries as large as China or

India inevitably overestimate the technological

capabilities of certain areas and underestimate

the capabilities of others. This also applies

to countries with much higher technological

capabilities such as, for example, the United

States and Japan. Moreover, recent research on

technological agglomerations (Cantwell &

Iammarino, 2003) showed that technological

activities tend to cluster in a few hubs even in

the most technologically advanced countries.

Still, the notion of national systems of inno-

vation (see Andersen, Lundvall, & Sorrn-

Friese, 2002; Edquist, 1997; Freeman, 1997;

Lundvall, 1992; Nelson, 1993) indicates that it

makes sense to analyze the technological

capabilities of territorial states, since these

provide one of the main institutional settings

for know-how generation and diﬀusion The

same analysis has already been successfully

applied to developing countries (see Cassiolato

& Lastres, 1999, and Sutz, 1997, for Latin

America; Hobday, 1995, for Asia; Lall &

Pietrobelli, 2002 for Africa).

Third, although we measured technological

capabilities with a variety of indicators, we

made an attempt to provide a synthetic

indicator. Other exercises made an eﬀort to

estimate countries’ technological capabilities

by aggregating data at the ﬁrm level. Unfor-

tunately, this approach has not yet been able

to generate data for larger groups of coun-

tries. Our measure is typically a macro-eco-

nomic one and, at the country level, it is

composed of a selected number of indicators.

In spite of the limitations of a synthetic

indicator, we share with the UNDP, UNIDO

and WEF the belief that the various compo-

nents singled out could be added up in order

to provide a more comprehensive measure of

technological activities.

2. CHANGES COMPARED TO PREVIOUS

ANALYSES

We built upon the TAI attempt developed by

UNDP (Desai, Fukuda-Parr, Johansson, &

Sagasti, 2001; UNDP, 2001), and the Indus-

trial Development Scoreboard developed by

UNIDO (Lall & Albaladejo, 2001; UNIDO,

2003). The TAI takes into account many indi-

cators, by classifying them in four categories:

the creation of technology, the diﬀusion of new

technology, the diﬀusion of old technology, and

human skills. We considered this a more eﬀec-

tive starting point than the index suggested

by the WEF (2002). The UNIDO Industrial

Development Scoreboard divides a battery of

indicators into two broad groups: the ﬁrst

deals with competitive industrial performance

(including manufacturing value-added per

capita, manufactured exports per capita, share

of medium- and high-tech industries in manu-

facturing value-added and share of medium-

and high-tech in manufactured exports); the

second concerns industrial capabilities (includ-

ing foreign direct investment per capita, foreign

royalty payments per capita, tertiary techni-

cal enrolments, enterprise ﬁnanced R&D per

capita, and the infrastructure as measured by

telephone main lines). The main modiﬁcations

we introduced to these two indexes are the

following.

(a) Enlarge the number of countries examined

In order to enlarge the number of countries

examined, without losing data and source

coherence, we focused on indicators whose

coverage was more satisfactory. We took into

account both the availability of data and the

dimension of population: we neglected coun-

tries with less than 500,000 inhabitants, except

for those countries (Luxembourg, Malta,

Cyprus and Suriname) for which we retained

suﬃcient data. For those countries for which

data proved analytically insuﬃcient (as for

most African countries), missing values were

estimated on the basis of national sources,

interviews with country experts, and perfor-

mance in comparatively similar countries and

indicators. In extreme cases, minimum values

were taken for groups of comparable countries

(often equivalent to zero, due to the conditions

of extreme poverty of some of the countries

analyzed). Our pool is comprised of 162 coun-

tries in total.

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 631

(b) Allowing comparisons over time

In addition to crosscountry comparisons, we

attempted time-series comparisons. The pur-

pose of the TAI was not to compare countries

at diﬀerent time points but to perform cross-

country comparisons at particular time points.

Standardized indicators from 0 to 1 were built

according to the following formula:

Observed value Minimum observed value

Maximum observed value Minimum observed value:

In TAI, all observed values referred to the same

time period. Since maximum and minimum

observed values are subject to change over

time, time comparisons are impossible. In

addition, the Industrial Development Score-

board presents a time-series comparison for

1985–98.

In order to allow for time-series compari-

sons, a maximum and a minimum value were

ﬁxed for ArCo, so that both would result

identical for both the time points considered (a

current period which oscillates from 1997 to

2000 and a past period from 1987 to 1990).

Given that during the two time points consid-

ered the majority of countries under observa-

tion experienced progress of some kind, the

minimum observed value was taken from the

past period, while the maximum observed value

was taken from the most current one. Conse-

quently, homogeneous indicators for all time

periods were devised with the certainty that no

country would express a passed minimum value

higher than the more recent one. In other

words, no index in the past could ever over-

come the value of 1. The formula for this new

indicator can be summarized as:

Ix¼Obspresent Minpast

Maxpresent Minpast

Since the literacy rate indicator is known to

oscillate between the values of 0% and 100%,

these were taken automatically as the minimum

and maximum goalposts (therefore eliminating

the need for minimum and maximum observed

values for this indicator).

3. THE ARCO TECHNOLOGY INDEX

Three main dimensions of technological

capabilities were considered:

––the creation of technology;

––the technological infrastructures;

––the development of human skills.

The choice was based on the assumption that

the three components play a comparative role

in the making of a country’s technological

capabilities. Thus, the overall Technology

Index (ArCo) has been built upon the equal

weighting of the three mentioned categories

(each of which is indexed).

The ArCo index

formula can therefore be sketched as:

ArCoTI ¼X

i¼1

kiIi;

where Iirepresents the three indexes (technol-

ogy creation, actual technology infrastructures

and actual human skills) for each country and

kiare the constants of 1/3.

The index of each category is calculated by

the same procedure used for the overall index,

that is, through the simple mean of certain

subindicators. In total we considered eight

basic indicators: two for the ﬁrst category and

three for the second and the third. The eight

subindexes are the following:

(a1) patents;

(a2) scientiﬁc articles;

(b1) Internet penetration;

(b2) telephone penetration;

(b3) electricity consumption;

(c1) tertiary science and engineering enrol-

ment;

(c2) mean years of schooling;

(c3) literacy rate.

The following is a detailed explanation of

each indicator:

(a) Creation of technology

(i) (a1) Patents

Patents are one measure of accounting for

the technological innovations generated for

commercial purposes. They represent a form

of codiﬁed knowledge generated by proﬁt-

seeking ﬁrms and organizations. Among the

various patent sources (for surveys on patents

as internationally comparable indicators, see

Archibugi, 1992; Pavitt, 1988b), we considered

patents granted in the United States. Since the

latter is the largest and technologically more

developed market of the world, it is reasonable

to assume that important inventions and

innovations are legally protected in the US

market. The TAI considers those patents that

are taken out by individuals in their home

country. Such data were not used here since

countries exhibit signiﬁcant legal diﬀerences––

for example, the very high number of patented

WORLD DEVELOPMENT632

inventions registered by Japanese and Korean

inventors at their national patent oﬃces is also

associated with the legal practice that requires

inventors to ﬁle an application for each claim.

The patent index is based on utility patents

(that is, invention patents) registered at the US

Patent and Trademark Oﬃce (USPTO, 2002).

Patents taken out in the United States by the

inventor’s country of residence were consid-

ered. The USPTO receives a greater number of

foreign patent applications than any other

patent oﬃce. Despite the fact that many

inventions are never patented, especially in

developing countries, patents represent never-

theless a good proxy for commercially exploit-

able and proprietary technological inventions.

The propensity of US inventors to register

inventions in their own national patent oﬃce is

higher than that of foreign inventors. To elim-

inate the bias toward US domestic patents, we

replaced the eﬀective number of domestic pat-

ents with our own estimation. The latter is

based on a comparison between the Japanese

and the US patents registered at the European

Patent Oﬃce (EPO), which represents a foreign

institution both for Japanese and American

inventors. We used the following estimation:

Estimated US domestic patents

¼ðJAPUSA USAEPOÞ=JAPEPO ;

where JAPUSA is the eﬀective number of patents

granted to Japanese inventors in the United

States, and USAEPO and JAPEPO are the eﬀective

number of patents granted to US and Japanese

inventors at the European Patent Oﬃce. Pro-

portions for patents granted in Japan to Euro-

pean inventors were also estimated and

appeared not to exhibit any major diﬀerences.

The number of patents for each country was

normalized by dividing it for the country’s

respective population (the number of patents

was expressed for a million people). In order to

account for the eﬀects that yearly ﬂuctuations

might have on the results obtained from small-

and medium-sized countries, a four-year mov-

ing average for 1987–90 and 1997–2000 was

considered.

The goal posts were set as the maximum and

the minimum observed value for 1997–2000

(230 for the maximum value––corresponding to

Japanese patents for a million people––and

zero for the minimum value) and the stan-

dardized patent activity index was constructed

by application of the general formula, with

values oscillating between zero and one. As

explained above, in order to allow for com-

parisons to be made across time as much as

across geographical borders, the same goal-

posts were kept for the previous years, so that a

comparable index for 1987–90 could be calcu-

lated while allowing us to evaluate each coun-

try’s growth rate during the two points in time.

(ii) (a2) Scientiﬁc articles

Scientiﬁc literature is another important

source of codiﬁed knowledge. It represents the

knowledge generated in the public sector, and

most notably in universities and other publicly

funded research centres, although researchers

working in the business sector also publish a

signiﬁcant share of scientiﬁc articles.

There is no single source of information

concerning all the scientiﬁc literature published

in the world. We were forced to rely on the

available, if limited, sources. Among them, the

most comprehensive and validated is the Sci-

ence Citation Index generated by the Institute

for Scientiﬁc Information. The index reports

information concerning the scientiﬁc and tech-

nical articles published in a sample of about

8,000 journals selected among the most presti-

gious in the world. The ﬁelds covered are:

physics, biology, chemistry, mathematics, clin-

ical medicine, biomedical research, engineering

and technology, and earth and space sciences.

It is often argued that the journals in this

sample are biased toward English-speaking

countries. Although there is some evidence

supporting this claim, it might be more accu-

rate to state that journals reﬂect the most visi-

ble part of the scientiﬁc literature, while they

ignore other important components in both

developed and developing countries––though

we believe the data source do not discriminate

heavily against developing countries. It is

certainly signiﬁcant that late industrializing

countries have begun to be active in both pat-

enting and scientiﬁc publications (see Amsden

& Mourshed, 1997).

Data were taken from the US National

Science Foundation’s most recent publications

(NSF, 2000, 2002) and the World Bank’s

database.

Article counting was based on

fractional assignments: for example, an article

written by two authors in two diﬀerent coun-

tries was counted as one-half article to each

country.

Switzerland scored the highest

number of articles for 1997–99 with 977 annual

articles per million people, while the minimum

goal post was zero for many countries with no

published scientiﬁc articles.

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 633

Data on R&D would have nicely comple-

mented the measure of national technological

creation, especially since they document devel-

oping countries’ learning eﬀort for acquiring

scientiﬁc and technological expertise. This

source however, was not employed due to a

lack of available data for all countries (see

UNESCO, 2002; World Bank, 2003, Table

5.12). UNIDO (2003) reported these data for

87 countries only, and for 16 of them the values

prove negligible. Moreover, some developing

countries tend to include some activities in

R&D statistics that do not ﬁt the standard

OECD Frascati Manual deﬁnitions (OECD,

2002). The advantage of using patents and

scientiﬁc articles consists of both sets of data

being validated by external sources as much as

by national ones (the US Patent Oﬃce in the

ﬁrst case, and the academic journals monitored

by the Institute for Scientiﬁc Information in

the second). This guarantees that individual

observations are collected according to stan-

dard criteria. A rank correlation was calculated

between the hierarchy of countries according to

US patents per million population and the

enterprise ﬁnanced R&D per capita (employed

in UNIDO, 2003). The result for the 61 coun-

tries with available data proved very high, with

a value of 0.92 (Archibugi & Coco, 2004),

demonstrating that a combination of patents

and scientiﬁc articles provide a robust measure

of national technological eﬀorts also compris-

ing R&D inputs.

(b) Technological infrastructures

We considered three diﬀerent indicators of

technological infrastructures: Internet, tele-

phony and electricity. They correspond to three

major industrial revolutions of the 20th century

(Freeman & Louta, 2001). They are basic

infrastructures for economic and social life.

Although they are not necessarily connected to

industrial capabilities, production knowledge is

strongly associated to their availability and

diﬀusion.

(i) (b1) Internet penetration

The Internet is a vital infrastructure not only

for business purposes, but also for access to

knowledge. Internet users access a worldwide

network. They diﬀer from Internet hosts, which

are computers with active Internet Protocol

(IP) addresses connected to the Internet. The

data on users, when available, are preferable to

those on hosts for two reasons: ﬁrst, they give a

more precise idea about the diﬀusion of Inter-

net among the population; second, some hosts

do not have a country code identiﬁcation and

in statistics are assumed to be located within

the United States, therefore causing a bias. The

source here used was the World Bank (see also

World Bank, 2003, Table 5.11), which extracted

the data from ITU (2001) (the same data are

employed in UNDP, 2001).

In order to compare the penetration of the

Internet among the diﬀerent countries we divi-

ded the number of users by population. The

maximum goal post is 540 per 1,000 people,

value belonging to Iceland, while the minimum

is zero, observed both in the recent and in the

past period for some very poor countries. The

internet is a new technology that has quickly

become the keystone of the Information and

Communication Technology, but it was not yet

commercially available in 1989–90. For this

reason, we postponed the past period to 1994

so that data referred to a time interval of ﬁve

instead of 10 years.

(ii) (b2) Telephone penetration

Telephony, besides its civilian component, is

also a fundamental infrastructure for business

purposes, and it allows tracing populations

with human skills and acquiring technical

information. Telephone mainlines are tele-

phone lines connecting a customer’s equipment

to the public switched telephone network. They

are another fundamental infrastructure for

economic and social life. Data are presented

per 1,000 people for the entire country (for

more information, see World Bank, 2003,

Table 5.10) both by World Bank database and

UNDP (2001), which both collected the data

from ITU (2001). To main lines, we added

mobile phones per 1,000 people, since they

represent the natural evolution of telecommu-

nication. An equal weight was assigned to older

and newer telephonic component since they

share the same function despite incorporating

diﬀerent degrees of technology.

As telephony represents a deﬁnitively

acquired form of technology for a large number

of countries (the developed ones), we expressed

the sums between ﬁxed and mobile lines in

natural logarithms. This ensures that, as the

level of telephony increases (therefore as we

move toward the more developed countries),

the diﬀerence between the new and the old

(lower) value expressed in logarithms decreases,

consequently reducing the gap among coun-

tries, for the exception of those countries with

WORLD DEVELOPMENT634

very low initial values. In other words, the use

of log creates a threshold above which the

technological capacity of a country is no longer

enriched by the use of telephones.

Furthermore, since many countries can said

to have reached the desired level of telephony

penetration, the chosen goal value for the cal-

culus of the index was not taken as the maxi-

mum observed value, but the OECD average

(960 telephones for 1,000 people). This not only

increases the index for all countries, but also

allows to eliminate useless diﬀerences among all

those countries whose telephony share is supe-

rior to the mean one (they all get the value one).

Therefore, as the minimum observed value is

zero (transformed to one due to the use of

logarithms), the formula becomes:

Ln ðobserved valueÞ

Ln ðOECD averageÞ:

(iii) (b3) Electricity consumption

Electric power consumption (kilowatt per

hour per capita) measures the production of

power plants and combined heat and power

plants, less distribution losses, and own use by

heat and power plants (for more information,

see World Bank, 2003, Table 5.10). This indi-

cator accounts for the oldest technological

infrastructure. Electricity consumption is also a

proxy measure for the use of machinery and

equipment, since most of it is generated by

electric power. Although we are aware that this

is likely to be larger for capital-intensive

industries than for services, we believe that the

use of logs provides values that respond to the

real use of machinery and equipment. Other

valuable measures of industrial capacity devel-

oped, for example, by Lall and his colleagues

(see Lall & Albaladejo, 2001; UNIDO, 2003)

are available for a smaller number of countries

only.

The observations on the telephony index over

the use of logarithms and the adoption of the

OECD average as the maximum goalpost,

apply a fortiori for the electricity consumption

index. The OECD average corresponded to

8,384 kwh per capita, whilst Ethiopia (1989–90)

produced the minimum value of 17 kwh per

capita. For those other low-income countries

whose data were not available a minimum

estimate was calculated.

Data on high technology production and

trade were not included. Although various

sources provide this kind of data (UNDP, 2001;

UNIDO, 2003; World Bank, 2003), some

problems emerge. Concerning high-tech pro-

duction, data for many countries are missing.

Moreover, available data are not always reli-

able, especially concerning production, since

they are derived from national sources, which

often apply diﬀerent criteria for deﬁning high-

tech sectors. Concerning high-tech trade, high

exports can simply imply high imports (as in

the case of Singapore and Hong Kong).

Moreover trade, including high-tech, is strongly

associated to the size of a country’s economy:

large countries have a lower propensity to trade

than small ones do, and vice versa. It was not

possible to produce an index able to account

for intraindustry trade and size, however a

comparison of ArCo with high-tech imports

data is attempted in Section 7.

Measures of capital equipment and ma-

chinery were not included either, despite these

representing a key component of embodied

technological capacity vital both for developed

and developing countries (Evangelista, 1999;

Pianta, 1995; Scott, 1989). The closest substi-

tute would be gross ﬁxed capital formation,

which is also available for a large number of

countries in the World Bank data base (World

Bank, 2003, Table 4.9). This measure, however,

was not accounted for either since: (i) it is not

possible to separate the component of gross

capital formation devoted to investment in

capital equipment and machinery from other

forms of investment; and (ii) the indicator is

expressed in monetary values, which would

make it diﬃcult to link ArCo to other currency-

based economic variables.

Technological capabilities are strongly asso-

ciated with human skills. Disembodied knowl-

edge (as measured by patents and scientiﬁc

literature) and technological infrastructures (as

measured by the Internet, telephony and elec-

tricity) have little value unless used by experi-

enced people. To complement our index, we

took into account three diﬀerent measures of

human skills.

(i) (c1) Tertiary science and engineering

enrolment

The indicator considered the share of uni-

versity students enrolled in science and engi-

neering related subjects in the population of

that age group. This indicator provides an

estimate of the science and technology human

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 635

capital, through the creation of a skilled human

base. It is obtained by multiplying two per-

centages, which are gross tertiary enrolment

ratio and percentage of tertiary students in

science and engineering.

The gross tertiary enrolment ratio is the ratio

of total enrolment at the tertiary level, regard-

less of age, to the population of the age group

that oﬃcially corresponds to the level of edu-

cation considered. Tertiary education, whether

or not to an advanced research qualiﬁcation,

normally requires, as a minimum condition for

admission, the successful completion of educa-

tion at the secondary level (for more informa-

tion, see World Bank, 2003, Table 2.12). Data

were gathered from the World Bank data set––

originally produced by UNESCO (2002).

Science and engineering students include

students at the tertiary level in the following

ﬁelds: engineering, natural science, mathemat-

ics and computers, and social and behavioral

science. By multiplying the two percentages, we

obtained the desired indicator. The maximum

value was scored by Finland in 1998 with a

value of 32.6%, while the minimum value

scored was zero for more than one country.

This indicator rests on an implicit assumption,

namely that the quality of education pro-

vided across countries is comparable. On the

contrary, we are aware that the quality of

education, and the successful completion of

education, is subject to great variation across

countries. The capability of developing coun-

tries is probably overestimated in our analysis,

while the capability of developed countries

is probably subject to underestimation. The

completion of courses is not accounted for

since it is assumed that enrolment in science-

and engineering-related subjects contributes to

the technological capability of a country inde-

pendently as to whether courses are completed.

(ii) (c2) Mean years of schooling

They represent the average number of years

of school completed in the population over

14. Although this indicator does not consider

diﬀerences in the quality of schooling, it gives

an indication of the human skill level (the

‘‘stock’’). The sources are the UNDP (2001),

which collected an elaboration by Barro and

Lee (2001),

and World Bank (2003, Table

2.13). The maximum goalpost is 12 and corre-

sponds to United States’ mean years of

schooling, while the minimum value ð0;7Þwas

observed in Mali (zero index was extended to

other poor countries without available data).

Even for this indicator we had to implicitly

assume the level of education to be comparable

across countries.

(iii) (c3) Literacy rate

Literacy rate represents the percentage of

people over 14 who can, with understanding,

read and write a short, simple statement about

their everyday life. Data were collected from

World Bank (2003) and UNDP (2001) (for

more information, see World Bank, 2003, Table

2.14). This indicator allows performing a better

distinction between the less-developed coun-

tries. We considered the literacy rate as a nec-

essary condition for the development of human

ability. In this case the index oscillates between

zero and 100%, which consequently represent

the minimum and the maximum goalpost.

A ﬁnal note about population, which is the

base for the calculus of the pro capita indexes. It

is based on the de facto deﬁnition of population,

which counts all residents regardless of legal

status or citizenship, except for refugees not

permanently settled in the country of asylum,

who are generally considered part of the popu-

lation of their country of origin (for more

information, see World Bank, 2003, Tables 1.1

& 2.1).

An interesting feature of the indicator here

devised is that none of the eight individual

components is based, directly or indirectly, on

monetary values. This means that it could be

matched by indicators expressed in monetary

value without any risk of collinearity. For

instance, it could be compared to indicators

such as international trade (including trade in

high-tech products), value added per employee

(which is often used as a measure for produc-

tivity), gross capital formation (a measure of

investment, including investment in capital

goods), and, of course, GDP and its growth. The

full database can be freely downloaded at http://

www.danielearchibugi.org/pdf/Theory_Measure-

ment_Techn_Change/ArCo_Index.xls.

4. THE RESULTS AT THE COUNTRY

LEVEL

Results do not diﬀer in a revolutionary

manner from other similar studies, but a num-

ber of fresh considerations can be made. First of

all, we tried, as in the TAI case, to group the 162

examined countries in diﬀerent blocks, by clas-

sifying them along with the level of the overall

WORLD DEVELOPMENT636

Table 1. A composite index of technological capabilities across countries (ArCo), 1990–2000

Actual

ranking

Current ArCo

Technology Index

Past ArCo

Technology Index

Past

ranking

Growth rate from the

last decade (%)

1 Sweden 0.867 0.681 2 27.2

2 Finland 0.831 0.614 6 35.2

3 Switzerland 0.799 0.735 1 8.7

4 Israel 0.751 0.669 4 12.2

5 United States 0.747 0.663 5 12.6

6 Canada 0.742 0.678 3 9.4

7 Norway 0.724 0.581 9 24.6

8 Japan 0.721 0.569 12 26.8

9 Denmark 0.704 0.584 8 20.6

10 Australia 0.684 0.561 14 21.9

11 Netherlands 0.683 0.571 10 19.7

12 Germany 0.682 0.593 7 15.0

13 United Kingdom 0.673 0.562 13 19.8

14 Iceland 0.666 0.484 18 37.8

15 Taiwan 0.665 0.436 22 52.6

16 New Zealand 0.645 0.570 11 13.3

17 Belgium 0.642 0.523 15 22.7

18 Austria 0.619 0.502 16 23.4

19 Korea, Rep. 0.607 0.415 31 46.3

20 France 0.604 0.499 17 21.0

21 Singapore 0.573 0.397 37 44.5

22 Hong Kong, China 0.569 0.435 24 30.8

23 Ireland 0.567 0.450 20 26.0

24 Italy 0.526 0.444 21 18.5

25 Spain 0.516 0.410 34 25.8

26 Slovenia 0.507 0.412 33 23.1

27 Greece 0.489 0.416 30 17.5

28 Luxembourg 0.486 0.426 27 13.9

29 Slovak Republic 0.481 0.428 26 12.3

30 Russian Federation 0.480 0.464 19 3.4

31 Czech Republic 0.475 0.432 25 9.9

32 Estonia 0.472 0.413 32 14.4

33 Hungary 0.469 0.402 36 16.8

34 Poland 0.465 0.393 39 18.3

35 Portugal 0.450 0.346 53 30.0

36 Bulgaria 0.449 0.435 23 3.2

37 Cyprus 0.440 0.384 41 14.4

38 Latvia 0.439 0.423 29 3.7

39 Belarus 0.431 0.403 35 6.8

40 Argentina 0.426 0.379 45 12.5

41 Chile 0.424 0.336 57 26.2

42 Ukraine 0.417 0.426 28 )2.2

43 Uruguay 0.417 0.348 52 19.9

44 Croatia 0.414 0.376 46 10.3

45 Bahrain 0.410 0.355 49 15.4

46 Lithuania 0.408 0.380 43 7.4

47 Kuwait 0.405 0.380 44 6.7

48 Moldova 0.395 0.394 38 0.2

49 United Arab Emirates 0.394 0.321 63 23.1

50 Romania 0.393 0.383 42 2.5

51 Panama 0.382 0.337 56 13.3

(continued next page)

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 637

Table 1—(continued)

Actual

ranking

Current ArCo

Technology Index

Past ArCo

Technology Index

Past

ranking

Growth rate from the

last decade (%)

52 Kazakhstan 0.381 0.393 40 )2.8

53 Trinidad and Tobago 0.380 0.348 51 9.3

54 Qatar 0.380 0.353 50 7.6

55 Georgia 0.379 0.371 47 2.3

56 South Africa 0.372 0.334 58 11.1

57 Lebanon 0.370 0.292 72 26.5

58 Malaysia 0.369 0.295 69 25.2

59 Venezuela, RB 0.369 0.328 60 12.4

60 Costa Rica 0.361 0.322 62 12.2

61 Malta 0.361 0.325 61 10.9

62 Yugoslavia, Fed. Rep. 0.358 0.334 59 7.2

63 Mexico 0.358 0.320 64 11.8

64 Tajikistan 0.356 0.369 48 )3.6

65 Turkey 0.347 0.286 75 21.4

66 Jamaica 0.346 0.264 85 30.8

67 Peru 0.345 0.292 74 18.2

68 Thailand 0.342 0.278 80 23.3

69 Jordan 0.341 0.300 67 13.6

70 Azerbaijan 0.337 0.342 54 )1.4

71 Colombia 0.331 0.286 76 15.6

72 Brazil 0.330 0.280 77 17.6

73 Armenia 0.326 0.339 55 )3.6

74 Puerto Rico 0.326 0.293 71 11.4

75 Saudi Arabia 0.326 0.280 78 16.4

76 Paraguay 0.323 0.269 84 20.0

77 Philippines 0.322 0.277 81 16.4

78 Cuba 0.322 0.313 65 2.8

79 Ecuador 0.319 0.294 70 8.3

80 Uzbekistan 0.319 0.313 66 1.9

81 Iran, Islamic Rep. 0.313 0.241 90 29.9

82 Libya 0.312 0.274 83 13.7

83 El Salvador 0.311 0.236 93 31.9

84 Dominican Republic 0.308 0.258 86 19.4

85 China 0.306 0.227 97 34.7

86 Kyrgyz Republic 0.306 0.300 68 1.9

87 Bolivia 0.305 0.254 88 19.8

88 Fiji 0.304 0.278 79 9.1

89 Oman 0.300 0.238 91 26.0

90 Macedonia, FYR 0.300 0.276 82 8.5

91 Turkmenistan 0.289 0.292 73 )1.2

92 Tunisia 0.288 0.227 98 26.8

93 Mauritius 0.285 0.231 95 23.6

94 Syrian Arab Republic 0.282 0.256 87 10.2

95 Sri Lanka 0.280 0.227 96 23.0

96 Zimbabwe 0.279 0.248 89 12.2

97 Algeria 0.277 0.221 100 25.1

98 Guyana 0.271 0.226 99 20.0

99 Egypt, Arab Rep. 0.269 0.219 101 22.6

100 Indonesia 0.265 0.190 108 39.7

101 Suriname 0.264 0.219 102 20.1

102 Honduras 0.258 0.218 103 18.3

103 Botswana 0.255 0.189 109 34.8

104 Albania 0.251 0.231 94 8.5

WORLD DEVELOPMENT638

Table 1—(continued)

Actual

ranking

Current ArCo

Technology Index

Past ArCo

Technology Index

Past

ranking

Growth rate from the

last decade (%)

105 Iraq 0.246 0.238 92 3.4

106 Zambia 0.240 0.213 104 12.3

107 Vietnam 0.239 0.164 118 45.5

108 Nicaragua 0.238 0.202 106 17.8

109 Guatemala 0.234 0.187 110 25.2

110 Gabon 0.231 0.204 105 13.1

111 India 0.225 0.169 116 32.9

112 Swaziland 0.222 0.184 111 20.4

113 Morocco 0.217 0.169 117 28.5

114 Namibia 0.217 0.184 112 17.6

115 Congo, Rep. 0.207 0.195 107 6.4

116 Kenya 0.204 0.177 114 15.1

117 Ghana 0.203 0.163 119 24.3

118 Mongolia 0.197 0.176 115 11.6

119 Cameroon 0.192 0.163 120 18.0

120 Pakistan 0.191 0.158 121 20.9

121 Korea, Dem. Rep. 0.187 0.179 113 4.9

122 Myanmar 0.179 0.135 123 32.2

123 Lesotho 0.178 0.154 122 15.4

124 Tanzania 0.155 0.126 124 23.2

125 Senegal 0.151 0.109 130 38.1

126 Papua New Guinea 0.146 0.119 125 22.4

127 Togo 0.145 0.097 133 48.8

128 Nigeria 0.141 0.114 127 23.6

129 Sudan 0.140 0.096 136 46.3

130 Yemen, Rep. 0.140 0.112 128 24.2

131 C^

ote d’Ivoire 0.136 0.080 141 69.8

132 Malawi 0.134 0.106 131 26.4

133 Uganda 0.133 0.097 134 37.6

134 Haiti 0.129 0.117 126 10.4

135 Congo, Dem. Rep. 0.125 0.110 129 13.6

136 Gambia 0.123 0.070 146 76.1

137 Bangladesh 0.123 0.086 138 43.2

138 Djibouti 0.122 0.099 132 22.3

139 Nepal 0.121 0.070 145 72.9

140 Madagascar 0.116 0.096 135 20.8

141 Benin 0.114 0.078 143 46.3

142 Rwanda 0.113 0.081 140 39.5

143 Mauritania 0.111 0.077 144 43.6

144 Central African

Republic

0.110 0.081 139 36.1

145 Angola 0.107 0.088 137 21.7

146 Bhutan 0.103 0.063 148 65.2

147 Lao PDR 0.098 0.057 151 73.6

148 Mozambique 0.098 0.069 147 41.6

149 Cambodia 0.096 0.047 156 103.3

150 Liberia 0.095 0.079 142 20.5

151 Eritrea 0.093 0.048 154 92.8

152 Guinea 0.079 0.045 158 73.9

153 Burundi 0.078 0.057 152 38.2

154 Guinea-Bissau 0.076 0.061 149 26.2

155 Sierra Leone 0.075 0.060 150 24.4

(continued next page)

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 639

ArCo Technology Index (Table 1). We identi-

ﬁed four groups,

according to the existence of

a signiﬁcant gap among the last country of a

group and the ﬁrst of the subsequent

––leaders (from 1 to 25 ranking);

––potential leaders (from 26 to 50);

––latecomers (from 51 to 111);

––marginalized (from 112 to 162).

(a) Leaders (from 1 to 25 ranking)

The ﬁrst group includes those countries able

to create and sustain technological innovation.

This is the group that concentrates the bulk of

the creation of technology. Seven consider-

ations can be made:

(i) What can be immediately noted is the

excellent performance of Nordic Euro-

pean countries: Sweden ranks ﬁrst, Finland

second, Norway seventh. These countries

hold extraordinary technological infra-

structures, and highly qualiﬁed human

resources. In addition to the static picture,

is noteworthy their trend: all but Den-

mark improved their ranking with respect

to a decade ago, with rates of growth beyond

20%.

(ii) Still more pronounced is the growth

of Newly Industrialized Countries, the so-

called Asian tigers: Taiwan, South Korea,

Hong Kong, Singapore. In a decade, their

index has grown by 52% in Taiwan and 31%

in Hong Kong. A huge growth occurred in

the category of the creation of technol-

ogy (1100% in South Korea and 200% in Sin-

gapore).

(iii) North American countries are more or

less stable in the top positions: the United

States ranks ﬁfth and Canada sixth, losing

a few positions. The United States has a

more prominent position in the creation of

technology than it did in the other two cate-

gories.

(iv) Japan occupies the eighth place (gaining

four positions in a decade), fruit of an excel-

lent performance in technology creation,

very good in technological infrastructures,

and relatively poor in human skills.

(v) Western Europe shows a slowdown: Ger-

many, France, Belgium, Austria, and Italy

fell behind during the decade, not so much

due to a slow growth, as much as due to

better performance by other countries (this

is particularly the case in technological infra-

structures). Switzerland ranked ﬁrst a decade

ago and now ﬁnds itself in third position.

Germany is now 12th, losing ﬁve positions.

The United Kingdom is stable at the 13th

position, while Ireland (23rd) lost two ranks.

Only Spain gained a few positions, resting

on the borderline (25th rank), between the

ﬁrst and the second grouping.

(vi) Australia and New Zealand almost ex-

changed places: the ﬁrst gained rank (from

14th to 10th) while the second lost rank

(from 11th to 16th).

(vii) Finally, Israel ranks fourth, even ahead

of the United States. This apparently sur-

prising result is attributable to the high num-

ber of patents granted in the United States,

accompanied by an excellent achievement

in the formation of human capital.

(b) Potential leaders (from 26 to 50 ranking)

The second group comprises countries that

have, on the one hand, invested in the forma-

tion of human skills and developed standard

technological infrastructures, and on the other

they have achieved little innovation.

Table 1—(continued)

Actual

ranking

Current ArCo

Technology Index

Past ArCo

Technology Index

Past

ranking

Growth rate from the

last decade (%)

156 Chad 0.071 0.050 153 42.6

157 Ethiopia 0.067 0.047 155 41.1

158 Mali 0.066 0.032 159 108.2

159 Afghanistan 0.056 0.046 157 20.5

160 Burkina Faso 0.050 0.028 160 79.2

161 Niger 0.031 0.017 162 84.0

162 Somalia 0.028 0.024 161 13.9

Sources: CSRS (1996a, 1996b), EPO (2002), ITU (2001), NSF (2000, 2002), UNESCO (2002), USPTO (2002) and

World Bank (2003).

WORLD DEVELOPMENT640

(i) The largest number of countries in this

group comes from the former Socialist East-

ern European countries. Predictions here are

particularly risky, especially since the eco-

nomic and social conditions of these coun-

tries have been particularly turbulent. Data

and trends for the ex-Soviet or ex-Yugosla-

vian new states are not entirely reliable. In

spite of turmoil, these countries show a good

performance in human skills. Russia lost

position considerably in the last decade in

all three categories as a consequence of the

transition to a market economy. Bulgaria

and Romania lost meaningful positions

too, while Hungary and Poland have gained

a few positions.

Greece and Portugal, the countries to have

always lagged behind in technological cap-

abilities within the European Union, are

slowly bridging the gap. The latter, with a

growth rate of 30%, climbed from the

53rd up to the 35th rank. Greece gained

a few positions by reaching the 27th posi-

tion.

(ii) Some South American countries have

also gained positions during the decade:

Argentina, Uruguay and especially Chile

had a grow rate of 26%, with Argentina

reaching 40th.

(iii) Within the Arab countries the perfor-

mance of United Arab Emirates is noteable:

thanks to a good availability of infrastruc-

tures it gained 14 positions and almost

reached Kuwait, which remains the leader

of the Arabic countries for technological

progress at the 47th place.

The third group, the largest, is composed of

countries which, in one way or another, try to

stimulate their technology growth parallel to

their development eﬀorts: technological infra-

structure and formation of human skills.

(i) Central and South American countries

deserve a special comment since none of

them, with the exception of Cuba, have

shown a downgrading trend compared to a

decade ago (Panama, Venezuela, Costa

Rica, Mexico, Jamaica, Peru, Colombia,

Brazil, Paraguay and Bolivia). These coun-

tries have developed particularly good tech-

nological infrastructure (growth rates

around 20%), though human skills have

not grown as eﬀectively (not superior to

10%).

(ii) A similar trend can be observed among

Asian countries, where Malaysia and Thai-

land (both with a growth rate beyond 20%)

are in the top positions, followed by the Phil-

ippines (growth rate of 16%). Although

placed at the bottom of this list (100th),

Indonesia shows the highest growth rate

since the previous decade (40%).

(iii) In Asia, China and India deserve a sepa-

rate comment: China has shown an extraor-

dinary growth rate of technological

infrastructures (71%) but has remained al-

most stable in human skills wise. Overall, it

has shown one of the highest growth rates

in the last 10 years (35%, second only to Indo-

nesia), by gaining 12 positions (from 97th to

85th).

(iv) India closes the third grouping by rank-

ing at 111th. This may seem unfair but,

apart from some African countries and Viet-

nam––which do not have reliable data relat-

ing to the past––India is the country that

shows the highest growth rate (33%), driven,

like China, by the development of techno-

logical infrastructure.

(v) In the Middle East, Lebanon climbed to

the 57th position (growth rate of 26%), plac-

ing behind Qatar (54th) and ahead of Jordan

(69th), while Saudi Arabia increased its rank

to the 75th position.

(vi) Finally, a restricted set of African

countries showed signs of catching up,

with South Africa (56th) in the lead and

North African countries Tunisia (92nd),

Algeria (97th) and Egypt (99th) just

behind. These countries show a delay in

the development of technology infrastruc-

tures, but are growing in terms of human

skills.

(d) Marginalized (from 112 to 162 ranking)

The fourth and last group is composed of

marginalized countries, which do not have

large access even to the oldest technologies,

such as electricity and telephony. In this group,

relative position is not particularly meaningful,

due mainly to the lack of available data. Even

high growth rates can simply be due to the very

low values in both periods. These countries are

practically lacking the ﬁrst category––creation

of technology––and have poor technological

infrastructures and human skills. Many African

countries fall within this grouping where the

low technological level is associated to the very

low income levels.

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 641

5. SOME STATISTICS ON THE

INDICATORS

After having commented the results at the

country level, we wish to report some simple

statistics about the indicators. In Table 2 we

calculated the correlation matrix across the

eight indicators presented. As expected, all

correlation coeﬃcients are positive. The values

are diﬀerent, however, indicating that the var-

ious indicators taken into account highlight

diﬀerent aspects of technological capabilities.

As predictable, the correlation is greater

across indicators belonging to the same cate-

gory of technology (creation, infrastructures or

human skills), but with some exceptions. For

example, the correlation between Internet users

and scientiﬁc publications is high. At the same

time the Internet is less correlated with the

traditional infrastructures (telephony and elec-

tricity). The latter are more highly correlated

with literacy rate and years of schooling. So it

would appear that more traditional forms of

technology remain closer to each other: the

indices of technology creation (patents, scien-

tiﬁc articles) have little correlation with literacy

rate, telephony and electricity.

It is also interesting to note the degree to

which each indicator is correlated with the ﬁnal

ArCo Technology Index. Since the ArCo

Technology Index represents the mean of the

eight components, it is natural to expect a high

correlation between them. This is indeed the

case, although patents show the weakest cor-

relation and schooling the strongest.

Diﬀerent results emerge if we consider the

correlation within each group. In particular,

diﬀerences emerged within the group of poten-

tial leaders (Table 3): composed mainly of East

European countries, it shows a negative corre-

lation (although very weak) between indicators

of human skills and those of technological

infrastructure. In this group of countries there

is no correlation between education perfor-

mance on the one hand, and infrastructures and

patenting activity on the other. Moreover, there

is no connection between scientiﬁc articles and

patents, conﬁrming that the sources of codiﬁed

knowledge creation from the business sector

and the academic community are not neces-

sarily complementary.

Table 4 reports the correlation matrix for the

latecomers, which signals a practical indepen-

dence between indicators of human skills and

indicators of creation of technology. The

former also exhibit little correlation with the

technological infrastructures, although a posi-

tive correlation is found between indicators of

creation and indicators of technology infra-

structure. Correlation within the leaders group

or within the marginalized group was not

reported. While for the latter group data can-

not be considered suﬃciently reliable, countries

within the group of leaders have already

reached the maximum level for more than one

indicator. The linear correlation coeﬃcients

would therefore prove less informative.

Table 5 reports the correlation matrix for the

three category indexes. The category of tech-

nology creation is a little less correlated with

the other two as well as with the ﬁnal ArCo

index. The intragroup analysis does not reveal

any new information, though it is interesting to

look at the indicators’ coeﬃcients of variation

(Table 6), which signal diﬀerent levels of

polarization of technological capabilities across

the 162 countries. As expected, the most sig-

niﬁcant dispersion occurs in the case of the

generation of technology, which is very highly

concentrated in a small cluster of countries. In

addition, Internet users and, to a lesser extent,

the scientiﬁc tertiary formation, are concen-

trated in just a small number of countries.

Concerning infrastructures, we note that the

older the technology is, the less its utilization is

polarized. Literacy is the least dispersed indi-

cator.

Historians who have taken into account the

geographical location of inventions over 3,000

years would not be surprised that the genera-

tion of inventions and innovations is strongly

concentrated in certain areas. They have in fact

shown that in the past inventive activity was

concentrated in what now we would call

‘‘hubs’’ such as the Greek cities, the Italian

Renaissance republics and Britain during the

industrial revolution (see Smithsonian Visual

Timeline of Inventions, 1994). Today some-

thing similar is happening in Silicon Valley as

well as in the Balgalore district. What might

appear surprising to an historian is the geo-

graphical diﬀusion of contemporary innovation

compared to its concentration in the past.

A comparison of the variation coeﬃcients

across the two periods allows also to test

whether the 162 countries are somehow con-

verging or diverging in their technological

capabilities. All the indicators show a certain

convergence from the past (that is, a reduction

of the divergence signalled by the coeﬃcients),

especially with regard to Internet (many coun-

tries in the past did not possess it at all, while

WORLD DEVELOPMENT642

Table 2. Correlation coeﬃcients across the various indexes of technological capabilities (all countries)a

Patent

index

Articles

index

Internet

index

Telephony

index

Electricity

index

Tertiary

index

Schooling

index

Literacy

index

ArCo technology

index

Patent index 1.000 0.791 0.692 0.446 0.445 0.537 0.530 0.320 0.705

Articles index 0.791 1.000 0.833 0.571 0.567 0.699 0.665 0.420 0.828

Internet index 0.692 0.833 1.000 0.607 0.594 0.618 0.659 0.431 0.805

Telephony index 0.446 0.571 0.607 1.000 0.843 0.713 0.819 0.818 0.890

Electricity index 0.445 0.567 0.594 0.843 1.000 0.674 0.744 0.712 0.854

Tertiary index 0.537 0.699 0.618 0.713 0.674 1.000 0.707 0.617 0.837

Schooling index 0.530 0.665 0.659 0.819 0.744 0.707 1.000 0.805 0.903

Literacy index 0.320 0.420 0.431 0.818 0.712 0.617 0.805 1.000 0.788

Sources: CSRS (1996a, 1996b); EPO (2002), ITU (2001), NSF (2000, 2002), UNESCO (2002), USPTO (2002) and World Bank (2003).

––Patent index: patents granted at the USPTO by country per million people (annual average for 1997–2000).

––Articles index: scientiﬁc articles by country per million people (annual average for 1997–99).

––Internet index: Internet users by country per million people (1999).

––Telephony index: ﬁxed and mobile telephone lines by country per million people (1999).

––Electricity index: electricity consumption by country per million people (annual average for 1997–98).

––Tertiary index: gross tertiary science and engineering enrolment by country (annual average for 1996–98).

––Schooling index: mean years of schooling by country (2000).

––Literacy index: adult literacy rate by country (2000).

––ArCo technology index: weighted mean of the previous indexes.

A NEW INDICATOR OF TECHNOLOGICAL CAPABILITIES 643

Table 3. Correlation coeﬃcients across the various indexes of technological capabilities for the potential leaders (countries from 26 to 50)a

Patent

index

Articles

index

Internet

index

Telephony

index

Electricity

index

Tertiary

index

Schooling

index

Literacy

index

ArCo technology

index

Patent index 1.000 )0.018 0.519 0.385 0.401 )0.318 )0.258 0.095 0.325

Articles index )0.018 1.000 0.362 0.530 0.342 0.067 0.143 0.191 0.798

Internet index 0.519 0.362 1.000 0.768 0.580 )0.532 )0.194 )0.236 0.447

Telephony index 0.385 0.530 0.768 1.000 0.606 )0.435 )0.188 )0.207 0.531

Electricity index 0.401 0.342 0.580 0.606 1.000 )0.309 )0.475 )0.575 0.325

Tertiary index )0.318 0.067 )0.532 )0.435 )0.309 1.000 )0.106 0.396 0.220

Schooling index )0.258 0.143 )0.194 )0.188 )0.475 )0.106 1.000 0.489 0.199

Literacy index 0.095 0.191 )0.236 )0.207 )0.575 0.396 0.489 1.000 0.438

Sources: CSRS (1996a, 1996b), EPO (2002), ITU (2001), NSF (2000, 2002), UNESCO (2002), USPTO (2002) and World Bank (2003).

––Patent index: patents granted at the USPTO by country per million people (annual average for 1997–2000).

––Articles index: scientiﬁc articles by country per million people (annual average for 1997–99).

––Internet index: Internet users by country per million people (1999).

––Telephony index: ﬁxed and mobile telephone lines by country per million people (1999).

––Electricity index: electricity consumption by country per million people (annual average for 1997–98).

––Tertiary index: gross tertiary science and engineering enrolment by country (annual average for 1996–98).

––Schooling index: mean years of schooling by country (2000).

––Literacy index: adult literacy rate by country (2000).

––ArCo technology index: weighted mean of the previous indexes.

WORLD DEVELOPMENT644

Table 4. Correlation coeﬃcients across the various indexes of technological capabilities for the latecomers (countries from 51 to 111)a

Patent

index

Articles

index

Internet

index

Telephony

index

Electricity

index

Tertiary

index

Schooling

index

Literacy

index

ArCo Technology

Index

Patent index 1.000 0.508 0.631 0.476 0.374 )0.012 0.001 0.161 0.431

Articles index 0.508 1.000 0.437 0.511 0.447 0.159 )0.008 0.094 0.501

Internet index 0.631 0.437 1.000 0.723 0.236 0.035 0.097 0.141 0.500

Telephony index 0.476 0.511 0.723 1.000 0.244 0.311 0.087 0.332 0.686

Electricity index 0.374 0.447 0.236 0.244 1.000 0.169 0.079 0.022 0.627

Tertiary index )0.012 0.159 0.035 0.311 0.169 1.000 0.114 0.189 0.507

Schooling index 0.001 )0.008 0.097 0.087 0.079 0.114 1.000 0.421 0.528

Literacy index 0.161 0.094 0.141 0.332 0.022 0.189 0.421 1.000 0.586