Twin Innovation Systems, Intermediate Technology and Economic Development: History, and Prospect for China

This series is intended to promote discussion and to provide information about work
in progress. The views expressed are those of the authors and should not be quoted
without permission.











DISCUSSION PAPER SERIES







Twin Innovation Systems, Intermediate
Technology and Economic Development: History,
and Prospect for China

Andrew Tylecote University of Sheffield Management School


Discussion Paper No 2005.14
June 2005




Address for correspondence:
Professor A B Tylecote
University of Sheffield Management School
9 Mappin Street
Sheffield, S1 4DT
Tel: + 44 (0) 114 222 3415
Fax: + 44 (0) 114 222 3348
a.tylecote@sheffield.ac.uk



Copies of discussion papers can be obtained by contacting the address below
Mandy Robertson
University of Sheffield Management School
9 Mappin Street
Sheffield, S1 4DT
Tel: + 44 (0) 114 222 3380
Email: M.Robertson@sheffield.ac.uk

---

Abstract

This paper argues that less developed countries (LDCs) need twin national systems of
innovation: systems with one, ‘upper’ level or sub-system to engage with advanced
technology and develop industries which use it; and (cooperating with the upper level)
a ‘lower’ level to help to improve the economy’s existing, backward technology. The
focus is on the lower level. The process there should involve the development and use
of intermediate technologies. These are much better suited to the LDC’s factor
endowment, and maximise the opportunities for learning by doing. Most LDCs lack
such a lower level, and lose much from this. Japan and Taiwan (it is shown)
developed twin NSIs and gained accordingly, until success caused the levels to merge.
Mainland China failed to build on its 1950s beginnings. The paper ends with a brief
discussion of what a dynamic lower level of the NSI would look like for mainland
China.

Keywords: national system of innovation; intermediate technology; lower level NSI;
mainland China’s technological development.


1. Introduction

Developing countries have a fairly clear-cut division of their economies into two
parts, modern and backward:
• The modern part uses more-or-less advanced technology derived from
developed countries.
• The backward part uses more-or-less traditional technology.
To some extent this division will go by sector: there are inevitably high-technology
sectors like aircraft which must be modern to exist at all, and sectors providing
traditional staple foods and traditional services to the domestic market which tend
mostly to use traditional technology. A sector may however in a given country have
both advanced and traditional technology either in distinct sub-sectors (up-market and
down-market) or working together – as where transformation processes are advanced
and transfer processes traditional (Amsalem, 1982, for Brazil).
It is with regard to the firms and sectors using advanced technology (which we shall
call the ‘modern economy’ hereinafter) that one can recognise an NSI much like that
of a typical developed country. The country’s institutions of research and higher
education are likely to be mainly focused on their needs. So is the banking system. So
are the policy-making parts of government, and state-owned utilities (who provide
support in the form of contracts). What is of course unlike a typical developed
country is that there is little really innovative activity: most of the effort is devoted to
finding, accessing, mastering, and to a modest degree adapting technology already
developed elsewhere. But since there is a great deal of all that taking place in every
developed country, the resemblance is still quite close.

The concentration of the NSI on the ‘modern economy’ is partly because it genuinely
needs help in order to survive, particularly once exposed to international competition.
Partly it is because it is believed that progress is about using advanced technology and
that nothing less will do. As for the firms and sectors using traditional technology

2

---

(hereinafter the ‘backward economy’), in most developing countries it is hard to
recognise anything which deserves the name of an innovation system, or even a
technological change system. What changes there mostly trickles in from abroad or
from the rest of the economy: there is nothing systematic about it. The gap thus tends
to grow wider, and progress is likely to take place by the wholesale importation of
advanced technology into a farm or enterprise, or by the growth of ‘modern’
enterprises at the expense of the rest. The relative size of the two economies thus
reflects the level of development.

This paper will argue, however, that such is a poor way to achieve rapid technological
development. The lower-technology parts of the economy need to be integrated
within the country’s NSI even though (indeed largely because) it may be by squeezing
them that most of the revenues required for nurturing the ‘modern economy’. Further,
the ‘modern economy’ can grow much more quickly if there is available to it a large
reservoir of human capital, entrepreneurs and whole businesses which already have
most of the skills and capabilities required within it – developed while part of the
‘backward economy’. Initially, however, the sub-systems of innovation of the two
economies must be quite distinct. The ‘modern economy’ may replace some
mechanised processes by more labour-intensive ones – for example, components may
be moved from one machine to the next by hand rather than by a moving belt, as
Amsalem (1982) found was common in Brazil. It may save again by buying rather
dated, second-hand equipment; it may save yet again by working that equipment for
two shifts rather than one, or three rather than two (cf. Morris-Suzuki, 1994, p.87 on
Japanese cotton-spinning in the late 19th century). But the basic processes of
transformation (in a manufacturing industry; mutatis mutandis for others) will be the
same as in the developed world, however unsuited they are to the developing country
– however much they stretch its technological capability and strain its – the country’s
- resources of money and skill. So the upper level of the NSI - must mirror, and be
closely attached to, the NSIs of the developed world.

In the lower-technology parts, the ‘backward economy’, it is necessarily different.
The more must be spent on the modern economy and the upper level of the NSI, the
less there is for the rest. They must live on what is left to them after being squeezed
by and for the modern economy. They must start from where they are: making
traditional products and services with traditional methods and equipment. And yet,
in spite of all these handicaps, they need not stay there: they may move up as rapidly
as the modern economy expands. This paper argues that history gives at least two
excellent East Asian examples of how this may be done, Japan (1870-1950) and
Taiwan (1950–1980) (though without much detail for Taiwan in the sources available
to me). Both started with a distinct lower-level NSI which gradually merged into the
upper-level NSI. It is then shown that after promising beginnings in the 1950s,
China’s lower-level NSI deteriorated. The concluding section argues that a dynamic
lower-level NSI should and could be developed in mainland China.

2. The theoretical case for ‘incrementalism’.

It can easily be shown that the type of innovation (or established technology) most
suitable for the needs of less developed country (LDC) users is likely to be quite
different from that designed with developed countries in mind (Tylecote and Galvao,
2001.) This follows from the fact that (in the terms of mainstream economics) there is

3

---

a large difference in relative factor prices: LDCs have cheaper low-skilled labour and
more expensive capital. There is also a difference in technological capability which
affects the suitability of innovation, partly by making LDCs dependent on imported
(and therefore expensive) capital goods, components, licenses and know-how when
they adopt sophisticated developed country (DC) technology. Technological
capability can be treated as a factor of production, but of a very special kind. The
difference from the others is that the more it is used, the more it grows, rather than
being used up. This is true at a world level for innovation: new technological
advances start from what was known before, and usually (except sometimes for
radical innovations) those who make them are among those who are thoroughly
familiar with the initial state of the art. It is also true firm by firm for diffusion: it is
almost always much more difficult for a firm to master a new technology if it is not
familiar with its predecessor.

Within this theoretical framework, the fundamental problem of each LDC is that there
are, for any product it might wish to make, usually two sets of technology available
(in some sense): the set it currently has and knows, and the set developed in DCs.
The former will be very limited in product quality and specification, and low in labour
productivity. The latter will have been designed to suit the DCs’ factor endowment,
requiring therefore much physical and human capital and little low-skilled labour. It
will also require a high level of technological capability. If the LDC wishes to adopt
the technology, it will be obliged to use an unsuitably high ratio of physical and
human capital to low-skilled labour, increasing its scarcity of the former and its
surplus of the latter.

Clearly if one were starting with a blank sheet of paper, and considering what was
best for the LDCs, one would come up with a range of technologies, all of them
• intermediate between ‘traditional’ and ‘DC’ in their relative use of the
conventional three factors,
• relatively small in scale, and
• incorporating or starting from many capabilities which the LDC already
possessed.
That is, one would come up with appropriate technology:

Appropriate technology (AT) is now recognised as the generic term for a wide range
of technologies characterised by any one of several of the following features: low
investment cost per work-place, low capital investment per unit of output,
organizational simplicity, high adaptability to a particular social or cultural
environment, sparing use of natural resources, low cost of final product or high
potential for employment (Jequier & Blanc, 1985, p. 9).

Unfortunately the development of each technology has a cost. The optimal policy for
an LDC, with scarce resources available for innovation (and diffusion) would appear
to be, then,
1. ‘Upgrade’ some traditional technologies, blending them where possible with
advanced technologies which represent a particular leap in productivity or
which are particularly easy to assimilate. (On ‘technology blending’ see for
example Bhalla, 1996, ch.3.)

4

---

2. Take some DC ‘cast-offs’: technology now superseded, embodied in equipment
which is cheaply available second-hand (and whose use can be taught by
experts who may also now be ‘obsolete’ in the DCs).
3. Adopt DC technology without much adaptation, where it represents a particular
leap in productivity or is particularly easy to assimilate.
4. Buy in from the DCs products which seem necessary and not available through
routes 1, 2 or 3.

3. Two histories of successful incrementalism.

3.1. Japani.
The development of the upper-level NSI in Japan is well known. After the Meiji
Revolution of 1868 the Japanese government was committed to rapid technological
development. The central government in Tokyo spent heavily on bringing foreign
experts to Japan and on establishing universities. During the 1870s it also set up
(initially-) state-owned firms in key high-technology industries, selling most of them
cheaply during the 1880s to the private entrepreneurs who were the progenitors of the
zaibatsu conglomerates. Another massive state investment was made in the first
decade of the 20th century in the Yawata steelworks, opened in 1901 after going five
times over budget, and profitable only from 1910 (Morris-Suzuki, p.80). Such direct
state intervention was largely due to the unequal treaties forced on it by the United
States in 1858, which expired only in 1911 and until then prevented it from imposing
any tariffs over 5% (Chang, 2.2.7). After that it used tariffs in a conventional way to
protect its nascent modern sectors. ‘Preferential public purchasing’, particularly for
the military, was used from early on: thus the key move in the development of what
became Toshiba was a naval order for torpedoes in 1892 (Morris-Suzuki, p.79). Its
aggressive policies first in Korea, then in NE and other parts of China can be seen
largely as a means of expanding its exploited backward sectors – and of seizing the
mineral resources required as inputs for the modern sectors.

Meanwhile the outline of a lower-level NSI can be discerned. Note, to begin with,
that what can be called the base level of technological competence – among artisans
in particular - was relatively high in Japan in relation to the ‘modern’ level of the time
(Morris-Suzuki, 1994: 86). Moreover the level of literacy was exceptionally high – in
fact similar to that of the advanced countries of the time. Primary and secondary
education was rapidly improved. These were the responsibilities of local government,
and it is also noteworthy that for historical reasons local government had substantial
responsibility, autonomy and capability. It duly played an active role in raising the
technological competences of artisans and farmers (Morris-Suzuki, 1994: 89ff.). In
1872, the mayor of Kyoto, for example, even sent a small group (a silk merchant and
two weavers) to France to acquire knowledge of French woven textiles and buy
weaving machinery (Morris-Suzuki, 1994: 92). This was a cheap version of the
ambitious central government expeditions to get the highest Western technologies: the
Kyoto expedition brought back rather simple equipment and knowledge which was
quickly incorporated into local craftsmen’s workshops. The craftsmen and merchants
themselves formed trade associations in order to spread knowledge of improved
technologies which could be adopted quickly and cheaply by their members. (Morris-
Suzuki, 1994: 94-5). Many of these improvements were developed by Japanese
inventors, mostly craftsmen themselves.


5

---

From 1899-1905, 73% of the patents granted went to Japanese inventors,
most of them for new developments in traditional areas such as better weaving looms
and improved farm tools…. This grassroots innovation, however, was not just a
matter of individual entrepreneurship…. Between central government [which was
focusing on the acquisition of high technology for the internal core] and private
citizen stood an immensely complex layer of intermediate social institutions – local
government, trade associations etc – which were vital to the spread of technological
knowledge in the Meiji Era and beyond. The most important contribution of these
grassroots institutions was their role as channels for the transmission of new ideas and
as instruments in overcoming the innate human fear of the unfamiliar….. Local trade
associations took this process one step further….. serving as a means of sharing the
risks of innovation amongst many producers…. Both prefectural technology centres
and trade associations tested and modified new techniques so that they could easily be
fitted into existing production systems. (Morris-Suzuki, 1994: 96-7)

One such new technique was the bicycle, which became the basis of a classic example
of innovative technology blending (the term is from Bhalla, 1996, the example is from
Morris-Suzuki). A Japanese artisan in the late 1860s had the idea of combining a
bicycle with a traditional hand-pulled carriage. The rikisha (rickshaw) quickly became
the main means of passenger transport in East Asian cities – with a capital cost per
passenger mile probably lower than the traditional rickshaw, and of course much
higher speeds. Again, this spread in its early stages was assisted by local government.
(Morris-Suzuki, 1994: 97).

The central government was not slow to notice what was happening below it. A report
by the senior official Maeda Masana in 1882-1884 argued persuasively for
strengthening the foundations of traditional industries. Entrepreneurs who had moved
incrementally from traditional technology to major success in export industries
became influential – like the Katakura silk firm which had by 1894 gone from a
backyard workshop to the largest silk factory in Japan. In 1900 and 1903 laws were
enacted to strengthen the position of local trade associations (Morris-Suzuki, 1994:
100). In 1903 a law was introduced to encourage and regulate the establishment of
technical colleges. These colleges taught middle-level technicians who went to work
in private firms – a good source of innovation and in particular of technology-
blending. In 1901 the Ministry of Agricultural and Commerce finally began the
implementation of Maeda’s proposal for developing networks of prefectural
laboratories linked to a national laboratory, though it was another four years before
they put money into it. But even the national laboratories, like the Oji sericultural
station, the Fermentation Laboratory and the Tokyo Industrial Research Laboratory
put most of their resources into using modern science to understand traditional
processes and products, so that they could be progressively improved.

The lower level was also able rapidly to improve its power sources, starting from the
traditional water wheels available in areas close to hills (i.e. most of Japan) and
introducing relatively modest improvements following Western practice, until the
upper-level NSI had got to grips with steam technology and thus made it cheaply
available. (In many cases by that time it was possible to go straight on to
hydroelectricity using the same fall of water.)

The dissemination of improved traditional technology through the lower level played
a vital economic role: it allowed the rapid development of large internationally-

6

---

competitive cotton and silk textile industries, which earned most of the foreign
exchange that paid for the modern part’s imports. These were dominated by small and
medium businesses. The lower level of the NSI was expanding the surplus which
could be creamed off for the modern economy.

Once Japan had got back its freedom of action in tariff matters, modern industry
became increasingly profitable and became dominated by the great privately-owned
zaibatsu conglomerates, which had used either connections in government (e.g.
Mitsubishi) or their own resources as merchant trading houses (Mitsui) to get
established. At some point in the 20th century an important force for the convergence
of the upper and lower levels began to appear, in the shape of what became known as
kinyu keiretsu (vertical groups): the big modern-economy companies established more
and more close relationships with small firms to which they could subcontract
relatively simple and labour-intensive processes. Often indeed they would be set up at
the instance of the large firm, which might suggest that one of its skilled workers take
some of its old machinery back to his own village and supply it from there, more
cheaply than it could make it in-house. (That same skilled worker might have learned
most of his skills within the lower level before being recruited.) The beauty of this
system from an economic point of view was that the smaller the firm, the smaller the
village in which it could be located, and the cheaper the labour on which it could
draw: peasant labour when it was not required in the fields (or at school) had a very
low opportunity cost. In effect, the backward economy was paying most of the cost
of these workers, so long as they stayed on the farm: and the modest contribution of
their employment in those sub-contractors helped them to do just that.

The feudal traditions of the Japanese countryside encouraged the arrangement of these
subcontractors in tiers, forming an inverted pyramid, so that Toyota for example
developed a three-tier system in which the third tier supplied the second, which
supplied the first, which supplied Toyota. They also encouraged the forging of close
relationships, which facilitated the concerted upgrading of technology. It is notable
that the Japanese motor industry’s leading role in the development of CNC equipment
in the 1960s and 70s has been explained by the prevalence at the outset of pre-Fordist
methods among their subcontractors, who simply had too short production runs to be
able to afford dedicated Fordist equipment – but for whom CNC was affordable
because it could be re-programmed to deal with a long series of short runs. This
could be called the culmination of the co-operation between the Japanese upper and
lower levels, and the point at which (in the motor industry at least) they merged.

It was however not only the subcontractors of firms like Toyota which emerged from
the lower level of the NSI. Much bigger firms did that. Toyota did, rather gradually,
going from loom manufacture (and invention) to motor vehicles. Others did it in one
generation. Soichiro Honda was a mere bicycle mechanic who got hold of a motor
cycle engine which he managed to fix to a bicycle. That and the rest is (company)
history. Konosuke Matsushita started in business with similar origins by inventing,
then making, a simple adaptor which allowed one or two electrical appliances to be
powered from the same wire that led to a poor Japanese family’s solitary light bulb.
These were highly innovative entrepreneurs whose ultimate ability to run a large high-
technology business was based on practice in running a very small one whose
technology and markets they thoroughly understood. Only the lower level of the NSI
could nurture such people: in the modern economy, success naturally depended much

7

---

more on ability to get on well with the rich and the powerful. To appreciate the
relative merits of the two types of business leader and of firm, contrast the histories of
Honda and Nissan, or of Matsushita and Mitsubishi Electric.

3.2 Taiwanii
Until 1945 Taiwan had of course been under Japanese rule and to some limited degree
shared in what has been described above. Under the Guo Min government after 1950
it developed a similar, and equally successful, twin innovation system. Again, of
course, the story of the upper level of the NSI is much better known. During the
1950s there was a rather standard policy of industrial tariff protection. (The Guo Min
government, like that of Japan and South Korea, had the luxury of complete freedom
of action in economic policy, due to the US fear of the PRC. This freedom extended
to all sorts of theft of intellectual property; an activity particularly popular in
Taiwanese industry.) This protection was refined in the 1960s to include selective
support for what could be defined as medium-technology export industries: industries
which were able with such support quickly to establish competitive advantage on
international markets. Key high-technology industries were identified by the
government, which set up state-owned firms in those sectors. However a key
characteristic of the rapid move up-market (and ‘up-technology’) that followed during
the 1970s and 80s, was the prominent role of SMEs in Taiwanese exports (Hobday,
1995). These do not appear to have been the recipients of substantial government aid,
and yet they moved steadily and successfully up-market. No doubt they got some
benefit from relatively cheap and high quality inputs provided by the state-owned
high-tech firms, and from government spending on research institutes and
universities; in other words, from a well-functioning upper-level NSI. Yet, more
important are the technological and entrepreneurial capabilities they had before they
started exporting. Where did they – that is the entrepreneurs themselves, and their
employees - get them from? As with Honda and Matsushita, it appears that most of
them originated from the lower-level NSI.

As of 1950 the large majority of the population of Taiwan were peasants, and so that
is where the lower-level NSI started from. The Japanese had created a fairly good
rural infrastructure of dirt roads and other facilities. In the early 50s, the Guo Min
government carried through a far-reaching land reform, which left the peasantry
owners of the smallholdings on which they lived and worked.1 This was followed up
by a programme of improvement to rural infrastructure, particularly focusing on roads
and agricultural extension services. At the same time there was heavy spending on
mass education: the emphasis was first on primary and vocational education, later on
secondary, and only when economic advance had gone a long way, in the late 1970s
and 80s, did the emphasis move up to higher education. (Contrast the longstanding
privileging of higher education in Brazil, and apparently similar policies in China and
India.)

These policies provided the basis for a tremendous flowering of manufacturing SMEs
using progressively higher technology. Consider a typical Taiwanese peasant family
of the 1950s and 60s. Its food and shelter (at the minimum) was provided by a farm
which was too small to need all its labour except very occasionally, and could usually

1 This action was extraordinary in view of their alliance with landowners on the mainland. It took place
under US pressure and guidance, as did the similar land reforms carried out in S.Korea and Japan at
about the same time, with the aim in all of them of creating a loyal anti-Communist peasantry.

8

---

spare several of its younger members for at least a standard working week. These
people could work very hard for very little, and with even meagre wages the family
would still have quite enough to live on and to provide the basis for a rapidly
improving standard of education. On those dirt roads a bicycle or moped could bring
them to work – and a simple tractor and cart could bring the family’s crops to market.
Much as argued above for Japan, only more so, the SMEs for which these people
worked could rapidly raise their productivity and progressively improve their
products. The difference from Japan is that they generally did so without any kind of
dependence on large high-technology domestic firms already established in an upper
level of the NSI. This has as good a claim to the title of ‘economic miracle’ as any.
The key seems to have been the existence of suitable networks in sub-sectors of the
mechanical, electrical and electronic industries which were such as to allow a fine
division of labour, allowing SMEs each to specialise in one operation or the
manufacture of one component – selling largely to American and Japanese firms
which could and would deal with any operation (including global marketing) which
really did require large scale or high technology (Hobday, 1995).

The main difficulty in writing about the Taiwanese lower level of the NSI is that it
worked so well that it eliminated itself in an exceptionally short period. This was
partly due to the quality of the institutions, partly to the favourable situation of
Taiwan as a small island economy with excellent access (both in terms of transport
and free trade) to world markets. It was therefore able to devote a substantial part of
its economy to medium-technology labour-intensive export activities without flooding
any of the markets it was supplying: this generated a comfortable surplus for
reinvestment, while within the industries like electronics in which this was going on,
its firms were able to find relatively cheap, incremental routes up-market (cf. the
much more expensive routes up-market taken by the Korean chaebol).

4. Mainland China since the establishment of the PRC: the agricultural NSI.

4.1 Context and historical background.
Mainland China now has a very clearly-marked and highly-developed upper-level
NSI (Gu, 1999). The foundations for a strong modern economy were laid in the
period of autarky before 1979, and they have been built on with determination since
then in spite of the opening to the outside: tariffs and other protective measures have
remained, and been accompanied by a full range of nurturing measures including the
development of strong institutions of research and higher education, and ‘policy
loans’ at low interest rates from state banks. The only serious structural defect of the
Chinese upper-level NSI appears to lie in its institutions of finance and corporate
governance, and these are being progressively improved (Tylecote and Cai 2004, Cai
and Tylecote 2004). In a few areas the Chinese modern economy is now
internationally competitive without protection or subsidy – at least in supplying
relatively high-tech goods to other developing countries.

The main deficiencies are in the lower-level NSI. This was not always the case. As
Wu, Tu and Gu (2003) show, after the establishment of the PRC in 1949 the new
Communist Government made a great effort to develop agricultural technology. (We
can take virtually all of agriculture then and since as forming part of the backward
economy, and making up the bulk of it.) By 1958 a comprehensive system of
technology diffusion stations had been set up, with 4549 of them, covering 55% of all

9

---

counties. Just as in Taiwan, this went in parallel with fundamental social and
institutional change. As of 1949 115 million acres – half the cultivated land –
belonged to landlords. A Land Reform Act in 1950 redistributed all of that, and the
materials to farm it, to 300 million landless peasants. The Communist government
took the view that individual peasant agriculture was highly inefficient, and promoted
aggregation, first through ‘farmer collective teams’, in which 40% of peasant families
were involved by 1952. The size of units under collective management was
progressively increased, and at the same time the degree of collective control was
intensified. This process culminated in 1958 with the foundation of the People’s
Communes: by the end of 1958 740,000 collective organisations had been recombined
into 26,000 People’s Communes, involving 99% of farmer families. As is well
known, this last massive aggregative step proved to be a disaster, leading to sharp
falls in agricultural production and widespread famine. Nonetheless it was maintained
(with only slight modification in 1962 to give more autonomy to Production Brigades)
until 1978.

As Mao’s ill-considered policies convulsed the countryside, the agricultural
technology services suffered with the peasantry. One third of the technology diffusion
stations were closed during the 1959-61 famine. There was a similar degree of
disruption during the Cultural Revolution (1966-76), with dismissal and even
persecution of the workforce in addition.

For better or for worse (mostly for worse) what can be called the Maoist lower-level
NSI was built around the People’s Communes. It was with them that the technology
diffusion stations learned to work.2 When the agricultural technology support system
was reorganised and rehabilitated in 1974-6, it continued to be based on the People’s
Commune structures: by the end of 1975 there were 26,872 Commune agricultural
technology diffusion stations, with more than three hundred thousand agricultural
technology diffusion teams (Production Brigade level), and 2.2 million agricultural
technology groups, at the lowest level (Production Team). (Wu et al., 2003, p.7).
Between 1978 and 1984, with the replacement of the collective production system by
the Household Contract-Responsibility system, the technology diffusion structures
lost their rationale, and much of their workforce left (Wu et al., 2003).

4.2 Deng and after
Replacing the old structures with structures appropriate to the new individualised
system of agricultural production does not seem to have been a priority of the central
government, but gradually new mechanisms emerged. Public sector institutions have
been remodelled, with a notable turning point in 1985 with the reform programme for
the science and technology R&D system (Gu, 1999, 2004). Agricultural R&D
institutes and extension services were pushed towards commercial earnings, though
provision of operational fees from public funds was continued. The public services
still take main responsibility for new crop breeding, technology diffusion, and
coordination. There were as of 2003 1,600 independent agricultural R&D institutes in
China, with employment of nearly 147,000, of whom 44600 are scientists and
engineers (thus the average institute has less than ten employees and less than three

2 When Mao’s faith in large-scale urban industrialisation wavered in the late 1950s (in parallel with the
break with the USSR, which provided the model for it) it was mainly to the People’s Communes that
he looked to develop intermediate technology manufactures – including the famous (infamous)
backyard steel furnaces.

10

---