Will Science and Technology Solve China’s Unemployment Problem?
Will Science and Technology Solve China’s Unemployment Problem?

Will Science and Technology Solve
China’s Unemployment Problem?*

Fredrik Sjöholm
Research Institute of Industrial
Economics and Örebro
University
P.O. Box 55665
SE-102 15 Stockholm, Sweden
fredrik.sjoholm@ifn.se

Nannan Lundin
E3G and Stockholm
Environment Institute
SE-106 91 Stockholm, Sweden
nanan.lundin@sei.se

Abstract
China needs a substantial growth of modern-sector employment
to absorb its huge supply of underemployed people and new la-
bor market entrants. The present crisis with its massive layoffs of
workers makes the issue even more pressing. Although the gov-
ernment has announced large public investments to deal with the
business cycle downturn, less attention has been paid to the
structural aspects of Chinese underemployment. One exception is
the recent emphasis of technology development. However, sci-
ence and technology (S&T) can have both positive and negative
effects on employment. Using information from a large sample of
manufacturing firms in China between 1996 and 2004, we analyze
how S&T affects employment. Our results suggest that S&T does
not promote employment growth.

1. Introduction

China has weathered the current global economic crisis
well: large public investments and stimulus packages have
enabled growth to remain relatively high. However, the
recent massive layoffs of workers, primarily in the man-
ufacturing sector, are of concern because these layoffs
hit a labor market that was in distress even before the
crisis hit.

* Xiaojing Guan, He Ping, and Jinchang Qian from the National
Bureau of Statistics of China have been most helpful in provid-
ing us with the data. We thank Changwen Zhao, Bhanupong
Nidhiprabha, Wei Zhang, and participants at the Asian Eco-
nomic Panel meeting in Tokyo 2009 for valuable comments and
suggestions on how to improve upon an earlier draft of this pa-
per. Fredrik Sjöholm gratefully acknowledges ªnancial support
from the Torsten and Ragnar Söderberg Foundation.

Asian Economic Papers 9:2

© 2010 The Earth Institute at Columbia University and the Massachusetts

Institute of Technology

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

China suffers from a chronic inability to create jobs for an underemployed and
growing workforce, and not enough attention has been paid to the structural aspects
of Chinese underemployment (Hu 2004). One exception is the emphasis on technol-
ogy development. Over the last few years, the Chinese government has promoted
science and technology (S&T) heavily, stressing technological change in general and
indigenous technological change in particular (Chinese Ministry of Science and
Technology 2006; Sjöholm and Lundin 2010). More recently, in August 2009 Science
and Technology Minister Wan Gang argued again, “The most effective way to with-
stand the impact of the global economic meltdown is to accelerate technological in-
novation, the new economic growth engine.”1 Chinese ªrms have responded to the
ofªcial rhetoric; today, China is one of the world’s largest investors in science and
technology (OECD 2005).

Policymakers expect increased efforts in science and technology to improve the
competitiveness and the growth rate of the Chinese economy. Less discussed is the
effect of science and technology on employment, and this neglect is unfortunate be-
cause of the serious lack of jobs in the formal sector.

Exactly how science and technology will affect employment is unclear. On the one
hand, it could enhance competitiveness and thereby increase demand for labor; on
the other, it could lead to skill- or capital-intensive production and thereby reduce
demand for labor. The dominant mechanism is still open to discussion.

Our analysis of the relationship between S&T and employment draws on a data set
that covers all large- and medium-sized enterprises in the Chinese manufacturing
industry between 1998 and 2004. One methodological concern is that we can only
observe employment in surviving ªrms and survival might be affected by S&T. The
results on how S&T affect employment could therefore be biased. Yet even after us-
ing the Heckman two-step estimation procedure to control for the higher survival
rate of ªrms engaged in S&T, the data show no positive effect on growth in employ-
ment. Our conclusion is that technology development does not seem to solve one of
China’s most important policy issues: insufªcient employment opportunities.

2. S&T and employment—a conceptual framework and previous studies

Progress on the S&T front has both positive and negative impacts on employment.
The positive impact is mainly caused by the effect of S&T on ªrms’ survival and
growth. More speciªcally, ªrms conduct S&T to improve existing production pro-

1 “Government Pledges Strong Support for Innovation-based SMEs.” China Daily, 1 Septem-

ber 2009.

2

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

cesses and products, or to develop new ones. New products and processes will re-
sult in productivity gains through improved efªciency in production (lower costs)
or through higher prices on output (new products). Improved productivity beneªts
the ªrm in terms of higher competitiveness and thereby increases its chances of sur-
viving in the market and expanding its activities.

There are also theories suggesting that some technological change might be negative
for employment. More precisely, the literature on skill-biased technological change
suggests that technology and labor (or some types of labor) might be substitutes
rather than complements. This means that improved technology might, for instance,
make the ªrm use more capital but less labor, or more skilled labor but less un-
skilled labor (e.g., Thoenig and Verdier 2003; Ekholm and Midelfart 2005).

Turning to the empirical literature, the positive relationship between S&T and pro-
ductivity is well documented and need not be elaborated on further.2 There is also
ample evidence of a positive effect of productivity on ªrms’ growth and survival.
For instance, Okamoto and Sjöholm (2005) examine productivity growth in Indone-
sia and ªnd a strong effect on aggregate productivity from increases in market
shares by plants with relatively high productivity growth. Accordingly, Levinsohn
and Petrin (1999) ªnd a similar mechanism in Chile with growth of market shares
for ªrms with high productivity.3 Survival is also closely related to productivity:
ªrms exiting the market tend to have relatively low levels of productivity.4 It should
be noted that ªrm growth is not automatically associated with growth in employ-
ment. Moreover, high productivity can be caused, of course, by factors other than
S&T.

Most empirical studies on technology and employment examine changes in the de-
mand for skilled and unskilled labor, typically in developed countries. There seems
to be substantial evidence of skilled-biased technological change, irrespective of dif-
ferences in methodologies and countries (Berman, Bound, and Machin 1998; Hol-
lander and ter Weel 2002; Kang and Hong 2002; Bauer and Bender 2004; Ochsen and
Welsch 2005; Xiang 2005). Whether skill-biased technological change will reduce to-
tal employment depends on two factors. First, the change in relative prices (wages
for skilled and unskilled labor) will have an impact on the changes in the number of
employees. If, for instance, the relative price of unskilled labor falls, this will miti-

2 See, for example, Wieser (2005) for a recent survey of the literature on R&D and ªrm pro-

ductivity.

3 See also Olley and Pakes (1996) and Foster, Haltiwanger, and Krizan (1998) for similar

ªndings in developed economies.

4 See, for instance, various chapters in the book by Roberts and Tybout (1996).

3

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

gate the negative effect on employment of unskilled labor. Second, changes in the
relative demand for different types of workers decrease the total number of employ-
ees only if the loss of unskilled workers is larger than the increase in skilled work-
ers.

The studies here mentioned are concerned with issues that are only related to the fo-
cus of our paper. We intend to examine the effect of S&T, rather than that of produc-
tivity, on total employment, rather than on the composition of employment. Al-
though, to the best of our knowledge, no such study has been conducted previously
on developing countries, there are a few studies on developed countries. For in-
stance, Van Reenen (1997) examines the effect of innovations on employment in a
panel of 598 British ªrms. The results show a positive effect of innovations on em-
ployment, which is robust to changes in speciªcations. Moreover, Smolny (1998) ex-
amines the effect of process and product innovations on a panel of 2,405 German
ªrms. Once more, there is evidence of a strong positive effect of innovation on em-
ployment.5

3. The Chinese context

The global economic crisis, starting in 2008, has lowered the demand for Chinese
goods, and consequently, Chinese exports. As a result, the demand for workers has
fallen and large numbers of migrant workers are reported to have lost their jobs. Al-
though the exact ªgures are yet unclear, reports suggest the ªgure to be somewhere
between 20 and 23 million, out of a population that in 2008 included between 130
and 140 million migrant workers (see Ye and Batson 2009 for a discussion).

The Chinese government has responded to the crisis by launching a major stimulus
package of about 4 trillion yuan in November 2008 and by additional packages,
such as a health insurance reform of about 850 billion yuan. A large part of the stim-
ulus focuses on infrastructure projects. This will presumably offset some increase in
unemployment caused by the decline in exports. The leadership remains concerned
that the efforts will not be sufªcient, and there have been frequent claims during
2009 that at least 9 million new urban jobs are needed urgently.

It should be noted that the crisis hits a labor market that is already in large dif-
ªculties: there was a serious lack of jobs in the formal market already before the cri-
sis (Hu 2004; Démurger et al. 2006). The structural problems are not seen in ofªcial

5 There are also other studies on technology change and employment in industrialized coun-
tries conducted at a more aggregated level. Most studies ªnd a positive effect of technology
change on employment. See Pianta (2006) for a survey of the literature.

4

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

statistics: registered urban unemployment increased between 1995 and 2007 but
only from 2.9 to 4 percent of the labor force (National Bureau of Statistics 2006,
2008). However, the ofªcial ªgures only include urban residents between ages 16
and 50 (16 and 45 for women) who register as unemployed. Urban residents who do
not register are not included, nor are rural residents and migrant workers. Knight
and Xue (2006) use adjusted ofªcial data and household data to estimate a more ac-
curate urban unemployment rate. Their estimates suggest that urban unemploy-
ment amounted to above 11 percent in 2001, as compared to the ofªcial ªgure of
3.6 percent. Lee (2000) cites different sources and comes up with a similar ªgure: ur-
ban unemployment in 1996 is estimated at about 13 percent. Finally, Giles, Park, and
Zhang (2005) use survey data for ªve large cities and ªnd the urban unemployment
rate to be about 14 percent in 2002.

The situation in the rural areas is likely to be even more troublesome but with size-
able underemployment rather than unemployment. For instance, almost one third
of the rural labor force are claimed to be “surplus agricultural workers”: workers
that can leave agriculture with little negative impact on output (Lee 2000; Knight
and Xue 2006). A large pool of underemployed workers depresses wages, as evi-
denced by low and declining shares of wages in value-added. For instance, the
World Bank (2007) shows that the wage shares in value-added has declined from 53
percent in 1998 to 41.4 percent in 2005. As a comparison, the corresponding share
was 57 percent in the United States in 2005.

Related to this issue is the large pool of Chinese workers in the informal sector. For
instance, around 65 percent of China’s internal migrants are without hukou (house-
hold registration) and are therefore excluded from the formal job markets (Cai,
Wang, and Du 2005). A ªnal sign of a deteriorating labor market is the large decline
in the labor force participation rate from over 80 percent in 1996 to 71 percent in
2005 (Vodopivec and Tong 2008).

The need for employment growth is stressed by the continued growth of the labor
force, which is predicted to grow at least until 2015 (Chow et al. 1999, p. 483; Cai
and Wang 2006), and there is an expected 24 million new entrants to the labor force
in 2009 alone.

Which Chinese ªrms will then be likely to provide the new jobs? There is strong evi-
dence that ªrm ownership is important for employment (Karlsson et al. 2009). For
instance, one main reason for the insufªcient growth in modern sector employment
in China is that the private sector, including foreign-owned multinationals and joint
ventures, has difªculties in absorbing the same number of workers that are laid off
from state-owned enterprises (SOEs). Employment in SOEs went from a peak of

5

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

145 million in 1995 to about 75 million in 2005 (Vodopivec and Tong 2008). Around
80–90 percent of these laid-off workers have moved to (small) private companies or
are engaged in self-employment, in particular in the informal sector (Giles, Park,
and Cai 2006; Vodopivec and Tong 2008). Hence, private domestic and foreign-
owned ªrms are relatively more likely to generate jobs than are SOEs.

Of other ªrm characteristics that affect employment, ªrm size might be an impor-
tant factor. In a study of the manufacturing sector in Shanghai, Chow et al. (1999)
ªnd small ªrms to be relatively able to generate jobs over the period 1989 to 1992.
This situation is likely to be present today and in other parts of China, considering
that the share of manufacturing employees in small ªrms has increased from 38.6
percent in 2000 to 49.5 percent in 2004.6

Referring to our issue of the impact of technology on employment, there are hardly
any previous studies that can be consulted. It has been shown that large ªrms
(many employees) conduct more S&T than small ªrms (few employees) (Sjöholm
and Lundin 2010) but we cannot draw any conclusions from this stylized fact re-
garding the causality between S&T and employment growth. In other words, it
might be that large ªrms are more willing to invest in S&T and thus, it is not a
causal effect from S&T to employment growth.

4. Data and descriptive statistics

4.1 Data
Our data are on large- and medium-sized enterprises in the Chinese manufacturing
sector over the period 1998–2004 and has been complied by the National Bureau of
Statistics of China.7 The classiªcation of large- and medium-sized ªrms is based on a
combined ªrm-size indicator, where employment, turnover, and ªxed assets are
taken into account.8

The included variables are from two different sources. The ªrst source is balance
sheets of ªrms from the Chinese industrial statistics; the other is S&T statistics.
Merging these two data sets and using unique ªrm identiªcation codes, we obtain a
data set with two categories of variables: (1) ªrm-level economic variables, such as
employment, wages, sales, value-added, proªt, exports, ªxed assets, time of estab-

6 The authors’ own calculation, based on information complied by National Bureau of Statis-

tics of China.

7 See also Xiao (2005, pp. 65–66) for a discussion of the data.

8 See Table A1 in the Appendix for the detailed classiªcation.

6

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

lishment, and ownership, and (2) technology-related variables including S&T and
R&D expenditures, human resource inputs such as S&T personnel and R&D person-
nel, and purchase of foreign technology.

4.2 Industry and ownership classiªcations
The industry classiªcation is similar to the International Standard Industrial
Classiªcation, Rev. 3 classiªcation. When output data, such as value-added and
sales, are deºated into real values, the deºators are based on either the three-digit or
the four-digit producer price deºators, depending on availability.

Furthermore, following the OECD classiªcation, we divide the data set into high-
tech and non-high-tech industries (Hatzichronoglou 1997; OECD 2005). The high-
tech industries include aircraft and spacecraft; pharmaceuticals; ofªce, accounting,
and computing machinery; radio, television, and communications equipment; and
medical, precision, and optical instruments. It should be stressed that products and
processes in ªrms in a high-tech industry do not necessarily have high-technology
content. This is particularly true for non-OECD countries such as China, and is due
to differences in the industrial structure as compared to OECD countries (e.g., the
dominance of labor-intensive processes in manufacturing).

Finally, for a comparison across various ownership groups, we follow the owner-
ship classiªcation applied by Jefferson et al. (2003) and Hu, Jefferson, and Jinchang
(2005) in their previous analyses of S&T activities in Chinese large- and medium-
sized enterprises (LMEs).9

4.3 Other data issues
S&T and R&D expenditures are two key measures on technology development used
in our study. According to the commonly used international classiªcation from the
OECD, these two concepts are deªned as follows (OECD 2002).

S&T: systematic activities, which are closely concerned with the generation, ad-
vancement, dissemination, and application of science and technology. These in-
clude such activities as Research and Experimental Development (R&D), Science
and Technical Education and Training (STET), and Scientiªc and Technological
Services (STS).

R&D: comprise creative work undertaken on a systematic basis to increase the
stock of knowledge, including knowledge of man, culture, and society and
the use of this stock of knowledge to devise new applications. The term R&D

9 See Appendix A2 for the detailed classiªcation.

7

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

covers three activities: basic research, applied research, and experimental devel-
opment.

In the current indicator system in China, the deªnition of R&D is in line with the
Frascati Manual. International classiªcations of S&T indicators are less straightfor-
ward and the Chinese classiªcation is no exception. The deªnition of S&T followed
the UNESCO manual when the Chinese S&T statistics system was ªrst introduced
in the mid-1980s. In the last two decades, the deªnition of S&T has changed more
toward the Frascati Manual recommendation. S&T in the Chinese indicator system
includes R&D, technology acquisition (licenses) and renovation, and miscellaneous
expenditures on preparation for the production of new products and applications of
R&D results. Hence, S&T include several activities not included in R&D. We will
therefore primarily use S&T in our analysis because we want to analyze how tech-
nology development in a broad sense affects employment. R&D expenditures will
be used as a robustness check in parts of the analysis.

Another important deªnition is that of ªrm survival. Using the ªrm identiªcation
code, we deªne ªrm survival as when the ªrm’s identiªcation code remains in the
data set and likewise, the “death” of the ªrm is deªned as when the ªrm code dis-
appears from the data set. However, it is difªcult to distinguish between natural
market exit (bankruptcy) and other reasons for ªrms to disappear from the data set.
More speciªcally, the identiªcation code of a ªrm can disappear for the following
reasons: natural exit; ownership change (e.g., due to privatization or merger and ac-
quisition) or industry switch; and decline of ªrm size to below the threshold when
ªrms become reclassiªed as small ªrms and are excluded from the LME survey.

The existence of different causes for a ªrm to disappear from the data may blur the
ªrm survival analysis. However, our main reason for analyzing survival is to correct
for a possible bias in the job-creation analysis. The difference in reasons causing
ªrms to disappear from the data is presumably of minor importance for this issue.

Finally, the coverage of LMEs was enlarged in the 2004 Economic Census of China,
as compared to surveys in previous years. Furthermore, in the 2004 census, S&T sta-
tistics were reported at the ªrm level. Previous surveys reported S&T at the level of
enterprise groups and all ªrms belonging to a group were added together and re-
corded as one observation. As a result, observations of the total number of ªrms and
the number of ªrms with S&T both increased in 2004.

4.4 Descriptive statistics
Table 1 shows the numbers of ªrms and employees between 1998 and 2004 by S&T
status in Chinese industrial ªrms. The number of ªrms has increased over the pe-

8

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

riod, from 23,105 in 1998 to 27,712 in 2004, and the main part of the increase is in the
second period when the number of ªrms increased by almost 24 percent.10 It is inter-
esting to note that growth has been comparably high in the number of ªrms without
S&T. For instance, the total number of ªrms without S&T increased by about 4 per-
cent during the ªrst period, as compared to a decline of about 10 percent for ªrms
with S&T. The development in the second period is even more striking with a large
increase in ªrms without S&T (40.3 percent) and a small increase in the number of
ªrms with S&T (4.2 percent).11

The development of employment shows a pattern similar to growth in ªrms: em-
ployment declined by almost 20 percent between 1998 and 2001 with a relatively
large decline for ªrms with S&T. Furthermore, employment increased by about 29
percent between 2001 and 2004, once more with a substantial growth in employ-
ment in ªrms without S&T (84 percent) and a small growth in employment in ªrms
with S&T (4 percent).

The relatively large increase in employment in ªrms without S&T should not come
as a surprise at the aggregate level because China has a comparative advantage in
labor-intensive sectors but not in technology-intensive sectors. What we want to ex-
amine is if in a given sector, ªrms with S&T have grown more or less than ªrms
without S&T. Looking at different sectors, it is particularly interesting to note that
even in high-tech industries, ªrms and employment have increased substantially
but with most of the increase taking place in ªrms without S&T. This might suggest
that most activities in high-tech industries are of relatively low skill-intensity.

Table 1 also includes the ªve largest industries (in terms of value-added) at the two-
digit level in 1998. Industry-level ªgures reveal the same story, where employment
and the number of ªrms without S&T tend to increase more (decrease less) than the
corresponding changes in ªrms with S&T. The sectors in Table 1 are rather broad. It
is, of course, possible that ªrms with and without S&T are located in different sub-
sectors, explaining the differences in growth in employment. To control for this pos-
sibility, we calculated employment growth at a four-digit level, which is the most
disaggregated level available. Employment growth tends, again, to be highest in
ªrms without S&T but the difference is less signiªcant than the previous ªgures, es-

10 Once more, some of the increase between 2001 and 2004 is, according to ofªcials at the Na-

tional Bureau of Statistics, caused by an improved coverage of the census and not only by an
increase in the real number of ªrms.

11 Here, once more, some of the changes might be due to the construction of the data rather
than being real changes. All ªrms that belonged to large enterprise groups with S&T were
reporting positive S&T before 2004. In the 2004 census, S&T were reported at the level of the
ªrm and not at the level of the enterprise group.

9

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

)

%

(

t
n
e
m
y
o
l
p
m
e

.

0
9
2

.

7
3
8

0
4

.

.

7
4
6

8

.

5
2

7

.

7
0
2

7

.

2
1

)

%

(

5

.

4
0
2

.

0
0
1
(cid:2)

4

.

9

6

.

1
5

.

7
0
(cid:2)

.

7
4
(cid:2)

9
7

.

.

2
0
1
(cid:2)

s

m

r
i
f

9
.
3
2

3

.

0
4

2

.

4

8

.

0
3

.

9
6
6

.

8
0
1

.

2
9
3
1

.

5
1
2
2

.

0
3
4

.

2
3
2

.

7
0
3

3

.

8
1

.

3
5
(cid:2)

.

3
6
(cid:2)

.

3
4
(cid:2)

1

.

0
2

.

9
9
3

1

.

5
6

.

8
6
1
(cid:2)

.

7
2
2

.

0
9
9

.

3
0
(cid:2)

.

4
4
6

.

9
0
(cid:2)

7

.

9

8

.

3
2
1

4

.

6
1
3

t
n
e
m
y
o
l
p
m
E

7
3
9
,
1
2
1
,
5
3

2
6
4
,
4
7
6
,
5
1

5
7
4
,
7
4
4
,
9
1

8
5
5
,
7
8
8
,
3

4
9
1
,
2
5
5
,
1

4
6
3
,
5
3
3
,
2

2
7
5
,
2
1
6

7
4
9
,
9
3
1
,
2

5
7
3
,
7
2
5
,
1

0
3
1
,
2
9
5

9
1
5
,
6
1
2
,
2

9
8
3
,
4
2
6
,
1

6
3
9
,
2
4
7
,
1

1
1
1
,
0
0
6

5
2
8
,
2
4
1
,
1

1
2
5
,
7
0
8
,
2

0
4
9
,
7
3
7
,
1

1
8
5
,
9
6
0
,
1

0
9
9
,
5
2
5

3
5
7
,
7
9
1

7
3
2
,
8
2
3

s

m

r
i
f

2
1
7
,
7
2

4
8
0
,
7
1

8
2
6
,
0
1

9
1
1
,
3

7
1
4
,
1

2
0
7
,
1

8
2
9

2
7
6

6
5
2

9
9
6

9
6
9

8
6
6
,
1

4
6
6
,
1

9
1
8

5
4
8

0
5
4
,
2

9
9
7
,
1

1
5
6

7
6
3

4
5
2

3
1
1

)

%

(

t
n
e
m
y
o
l
p
m
e

5
.
9
1
(cid:2)

0
.
3
1
(cid:2)

1
.
2
2
(cid:2)

)

%

(

1
.
1
(cid:2)

1
.
9
(cid:2)

8
.
6
4
(cid:2)

9
.
7
1
(cid:2)

8
.
1
3
(cid:2)

9
.
5
1
(cid:2)

5
.
1
(cid:2)

9
.
3
1
(cid:2)

4
.
6
1
(cid:2)

s

m

r
i
f

2
.
3
(cid:2)

9
.
3
(cid:2)

4
.
0
1
(cid:2)

2
.
6
1
(cid:2)

9
.
8
4
(cid:2)

6
.
3
(cid:2)

8
.
9
(cid:2)

3
.
6
(cid:2)

5
.
3
1
(cid:2)

8
.
6
(cid:2)

1
.
2
2
(cid:2)

3
.
1
(cid:2)

7
.
2
2
(cid:2)

3
.
4
1
(cid:2)

8
.
5
2
(cid:2)

9
.
9
2
(cid:2)

1
.
6
3
(cid:2)

9
.
3
2
(cid:2)

8
.
0
3
(cid:2)

0
.
8
4
(cid:2)

4
.
0
4
(cid:2)

8
.
4
(cid:2)

8
.
2
(cid:2)

3
.
1
1
(cid:2)

7
.
3
2
(cid:2)

4
.
4
2
(cid:2)

3
.
2
2
(cid:2)

8
.
5
(cid:2)

0
.
3
1
(cid:2)

0
.
2
(cid:2)

t
n
e
m
y
o
l
p
m
E

6
1
6
,
1
2
2
,
7
2

2
2
9
,
0
3
5
,
8

4
9
6
,
0
9
6
,
8
1

4
8
2
,
0
6
3
,
2

9
2
5
,
4
0
5

5
5
7
,
5
5
8
,
1

2
9
9
,
7
9
8
,
1

4
5
1
,
1
0
2

8
3
8
,
6
9
6
,
1

8
2
5
,
0
9
3

8
4
6
,
6
2
0
,
2

0
2
1
,
6
3
6
,
1

0
0
7
,
9
2
8
,
1

8
8
3
,
6
5
5

2
1
3
,
3
7
2
,
1

2
2
5
,
8
3
3
,
2

9
5
7
,
2
5
0
,
1

3
6
7
,
5
8
2
,
1

4
9
5
,
8
2
4

5
8
3
,
9
9

9
0
2
,
9
2
3

s

m

r
i
f

5
7
3
,
2
2

4
7
1
,
2
1

1
0
2
,
0
1

5
8
3
,
2

9
4
8

6
3
5
,
1

8
8
3

9
0
2

9
7
1

5
3
5

9
1
8

4
5
3
,
1

7
5
7
,
1

4
7
8

3
8
8

1
5
7
,
1

4
9
0
,
1

7
5
6

1
6

4
6
1

3
0
1

t
n
e
m
y
o
l
p
m
E

8
8
4
,
9
9
7
,
3
3

5
3
9
,
0
0
8
,
9

3
5
5
,
8
9
9
,
3
2

0
7
2
,
6
8
3
,
2

8
8
6
,
3
4
3

2
8
5
,
2
4
0
,
2

3
6
4
,
1
1
3
,
2

0
6
9
,
4
9
2

3
0
5
,
6
1
0
,
2

6
9
4
,
6
9
3

4
2
4
,
4
5
3
,
2

8
2
9
,
7
5
9
,
1

6
2
5
,
5
6
3
,
2

9
2
1
,
9
4
6

7
9
3
,
6
1
7
,
1

9
3
1
,
6
3
3
,
3

9
1
3
,
7
4
6
,
1

0
2
8
,
8
8
6
,
1

4
3
1
,
7
6

9
5
6
,
9
1
6

5
2
5
,
2
5
5

s

m

r
i
f

5
0
1
,
3
2

0
2
7
,
1
1

5
8
3
,
1
1

2
5
0
,
2

0
7
5

2
8
4
,
1

0
3
4

3
2
2

7
0
2

8
3
4

0
3
8

8
6
2
,
1

5
4
8
,
1

0
5
8

5
9
9

4
9
2
,
2

8
4
4
,
1

6
4
8

4
5

5
5
1

1
0
1

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

s
l
a
t
e
m

s
u
o
r
r
e
F

y
g
o
l
o
n
h
c
e
t

s
e
i
r
t
s
u
d
n
i

–
h
g
i
H

s

m

r
i
f

l
l

A

s
l
a
c
i
m
e
h
c

c
i
s
a
B

t
n
e
m
p
u
q
e

i

t
r
o
p
s
n
a
r
T

s
e
l
i
t
x
e
T

m
u
e
l
o
r
t
e
P

s
t
c
u
d
o
r
p

n
i
h
t
w
o
r
G

n
i
h
t
w
o
r
G

4
0
0
2
–
1
0
0
2

f
o

.

o
N

4
0
0
2

n
i
h
t
w
o
r
G

n
i
h
t
w
o
r
G

1
0
0
2
–
8
9
9
1

f
o

.

o
N

1
0
0
2

f
o

.

o
N

8
9
9
1

y
r
t
s
u
d
n
i
e
s
e
n
i
h
C
e
h
t
n
i

s
u
t
a
t
s
T
&
S
y
b
t
n
e
m
y
o
l
p
m
e
d
n
a

s

m

r
i
f

f
o
r
e
b
m
u
N

.
1

e
l
b
a
T

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

.
8
9
9
1
n
i

)
d
e
d
d
a
–
e
u
l
a
v
(

e
z
i
s

r
i
e
h
t
n
o

d
e
s
a
b
n
e
s
o
h
c
n
e
e
b

e
v
a
h

s
r
o
t
c
e
S

:
e
t
o
N

.
a
n
i
h
C

f
o

s
c
i
t
s
i
t
a
t
S
f
o
u
a
e
r
u
B

l
a
n
o
i
t
a
N
e
h
t

y
b

d
e
d
i
v
o
r
p
a
t
a
D

:
e
c
r
u
o
S

10

Asian Economic Papers

Will Science and Technology Solve China’s Unemployment Problem?

pecially in the second period. More speciªcally, employment growth was higher in
ªrms without S&T than in ªrms with S&T in 100 of the 141 available sectors in the
ªrst period, and in 75 sectors in the second period (not shown).

Table 1 suggests that employment has increased more in ªrms without S&T than in
ªrms with S&T, but the causality between S&T and growth in employment is un-
clear. An alternative approach to the issue of S&T and employment is to compare
employment growth within ªrms with and without S&T. This is done in Table 2
where, for instance, we compare growth in employment between 1998 and 2001 in
ªrms that conducted S&T and ªrms that did not conduct S&T in 1998. Hence, unlike
Table 1, the sample only includes those ªrms that are present over the period 1998–
2001 and/or 2001–2004.

Table 2 shows that employment has declined in the ªrms included; the number of
employees decreased by about 17.3 percent between 1998 and 2001 and by about
3.2 percent between 2001 and 2004. The performance was similar in ªrms with and
without S&T in the ªrst period, but growth in employment has been positive in
ªrms without S&T and negative in ªrms with S&T in the second period.

It is worth noting that ªrms in high-tech industries have seen a lower than average
decline in employment in the ªrst period and a positive employment growth in the
second period. This could be an indication of an increased importance of high tech-
nology in the Chinese economy. However, it should also be emphasized that, even
within high-tech industries, employment growth has been substantially higher in
ªrms without S&T.

The pattern of a comparably strong employment growth in ªrms without S&T is
also seen in other sectors: employment growth is higher in ªrms with S&T than in
ªrms without S&T in only one industry in 1998–2001 (ferrous metals) and one in-
dustry in 2001–2004 (petroleum products). Hence, there does not seem to be any
positive effect of S&T on employment growth, given the descriptive ªgures in
Table 2.

As previously discussed, employment has declined rapidly in Chinese SOEs. This is
likely to be one cause for the negative growth in employment seen in Table 3. It is
also possible that the development in SOEs shades the role of S&T in employment.
Therefore, we divide our sample of ªrms by ownership in Table 3.

Table 3 shows that, not surprisingly, the number of employees has declined rapidly
in SOEs: around 20 percent between 1998 and 2001, and 12 percent between 2001

11

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

t
n
e
m
y
o
l
p
m
e

n
i
h
t
w
o
r
G

t
n
e
m
y
o
l
p
m
E

t
n
e
m
y
o
l
p
m
E

h
t
o
b
n
I

t
n
e
m
y
o
l
p
m
e

s

m

r
i
f

f
o

.

o
N

n
i
h
t
w
o
r
G

t
n
e
m
y
o
l
p
m
E

t
n
e
m
y
o
l
p
m
E

s

m

r
i
f

f
o

.

o
N

4
0
0
2
d
n
a
1
0
0
2
h
t
o
b
g
n
i
t
s
i
x
e
s

m

r
i
F

1
0
0
2
d
n
a
8
9
9
1
h
t
o
b
g
n
i
t
s
i
x
e
s

m

r
i
F

r
a
e
y
d
n
a

,
r
o
t
c
e
s

,

T
&
S
y
b
t
n
e
m
y
o
l
p
m
E

.
2

e
l
b
a
T

)

%

(
4
0
0
2
–
1
0
0
2

.

2
3
(cid:2)

.

7
0
1
(cid:2)

.

8
7
(cid:2)

.

0
7
(cid:2)

.

8
0
(cid:2)

.

4
8
3
(cid:2)

.

.

.

8
0
1
(cid:2)

3
8
1
(cid:2)

6
2
1
(cid:2)

.

8
8
(cid:2)

.

4
3
1
(cid:2)

.

9
6
1
(cid:2)

.

4
6
1
(cid:2)

.

6
1
1
(cid:2)

.

8
7
1
(cid:2)

.

5
1
(cid:2)

.

6
9
(cid:2)

.

9
7
(cid:2)

.

.

.

3
3
2
(cid:2)

2
1
3
(cid:2)

6
1
2
(cid:2)

,

3
0
2
0
2
6
,
4

,

2
4
9
7
0
3
,
6
1

9
3
7

,

7
8
6
,
1
1

4
0
0
2
n
I

5
4
8
5
4
4

,

,

2
3
3
5
3
7
,
1

7
8
4

,

9
8
2
,
1

8
3
6

,

6
9

,

2
6
0
6
5
2
,
1

,

4
2
4
9
5
1
,
1

6
7
9

,

9
0
2

,

7
5
0
7
1
2
,
1

,

1
8
0
7
0
0
,
1

,

7
8
3
3
2
9

7
5
6

,

2
1
2

0
3
7
0
1
7

,

,

2
7
8
5
1
2
,
1

6
8
0

,

4
9
4

,

6
8
7
1
2
7

9
1
3
4
4

,

5
4
6

,

6
7
2

6
2
3

,

2
3
2

8
3
1
,
3
7
1
,
4

9
1
0
,
9
4
8
,
6
1

1
8
8
,
5
7
6
,
2
1

1
1
2
,
2
2
3

4
2
9
,
1
2
6
,
1

3
1
7
,
9
9
2
,
1

5
9
6
,
1
8

5
6
7
,
7
0
4
,
1

0
7
0
,
6
2
3
,
1

4
3
0
,
3
9
1

5
2
9
,
4
0
4
,
1

1
9
8
,
1
1
2
,
1

4
1
6
,
0
4
2

6
0
4
,
4
6
8

0
2
0
,
5
0
1
,
1

1
7
4
,
4
3
2
,
1

7
9
7
,
0
5
4

4
7
6
,
3
8
7

3
1
4
,
4
6

3
7
8
,
0
6
3

0
6
4
,
6
9
2

7
8
8
,
8

2
1
7
,
3

5
7
1
,
5

4
2
8

3
1
3

7
3
1
,
1

5
6

1
8
1

6
1
1

3
7
6

8
8
1

5
8
4

1
7
6

5
2
2

6
4
4

4
3
6

1
1
3

3
2
3

1
0
1

5
2

6
7

1
0
0
2
n
I

4
0
0
2
,
1
0
0
2

)

%

(
1
0
0
2
–
8
9
9
1

3
.
7
1
(cid:2)

8
.
5
1
(cid:2)

8
.
7
1
(cid:2)

5
.
3
(cid:2)

0
.
2
1
(cid:2)

3
.
4
1
(cid:2)

7
.
4
1
(cid:2)

4
.
4
2
(cid:2)

7
.
3
1
(cid:2)

9
.
6
1
(cid:2)

9
.
0
2
(cid:2)

2
.
6
1
(cid:2)

2
.
1
2
(cid:2)

2
.
9
1
(cid:2)

8
.
1
2
(cid:2)

0
.
7
1
(cid:2)

7
.
6
1
(cid:2)

2
.
7
1
(cid:2)

9
.
3
(cid:2)

3
.
2
4
(cid:2)

5
.
5
4
(cid:2)

8
5
9
,
8
7
7
,
4

6
0
6
,
5
2
1
,
9
1

8
4
6
,
6
4
3
,
4
1

1
0
0
2
n
I

4
1
6
,
0
4
2

1
9
2
,
0
1
6
,
1

7
7
6
,
9
6
3
,
1

5
9
4
,
9
0
1

1
7
5
,
3
0
4
,
1

6
7
0
,
4
9
2
,
1

1
5
9
,
9
0
2

4
8
2
,
7
0
6
,
1

3
3
3
,
7
9
3
,
1

9
8
6
,
8
0
3

0
4
8
,
5
5
9

9
2
5
,
4
6
2
,
1

6
0
0
,
8
4
4
,
1

2
8
9
,
3
2
6

4
2
0
,
4
2
8

4
5
2
,
3
3

5
3
0
,
8
5
2

1
8
7
,
4
2
2

5
2
2
,
3
3
1
,
3
2

9
7
0
,
4
7
6
,
5

6
4
1
,
9
5
4
,
7
1

8
9
9
1
n
I

7
0
5
,
2
3
2

2
8
7
,
0
3
8
,
1

5
7
2
,
8
9
5
,
1

6
9
7
,
4
4
1

2
9
8
,
4
4
6
,
1

6
9
0
,
0
0
5
,
1

5
4
4
,
5
6
2

8
9
8
,
3
3
9
,
1

3
5
4
,
8
6
6
,
1

1
5
1
,
2
8
3

9
1
8
,
4
0
6
,
1

8
6
6
,
2
2
2
,
1

1
6
7
,
3
4
7
,
1

7
7
0
,
9
4
7

4
8
6
,
4
9
9

6
9
5
,
4
3

0
0
4
,
7
4
4

4
0
8
,
2
1
4

1
0
0
2
,
8
9
9
1

h
t
o
b
n
I

9
2
1
,
6

9
4
5
,
7

4
3
3

8
9
3
,
1

4
6
0
,
1

8
7
6
,
3
1

6
9

3
3
2

7
3
1

8
7
8

6
5
2

2
2
6

8
5
4

0
6
6

8
1
1
,
1

9
6
0
,
1

2
1
6

7
5
4

0
0
1

8
2

2
7

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

h
c
e
t
h
g
i
H

t
n
e
m
p
u
q
e

i

t
r
o
p
s
n
a
r
T

s
l
a
c
i
m
e
h
c

c
i
s
a
B

s
e
l
i
t
x
e
T

s
u
o
r
r
e
F

s
l
a
t
e
m

m
u
e
l
o
r
t
e
P

s
t
c
u
d
o
r
p

l
l

A

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

.
a
n
i
h
C

f
o

s
c
i
t
s
i
t
a
t
S
f
o
u
a
e
r
u
B

l
a
n
o
i
t
a
N
e
h
t

y
b

d
e
d
i
v
o
r
p
a
t
a
D

:
e
c
r
u
o
S

12

Asian Economic Papers

Will Science and Technology Solve China’s Unemployment Problem?

t
n
e
m
y
o
l
p
m
e

n
i
h
t
w
o
r
G

t
n
e
m
y
o
l
p
m
E

t
n
e
m
y
o
l
p
m
E

h
t
o
b
n
I

t
n
e
m
y
o
l
p
m
e

s

m

r
i
f

f
o

.

o
N

n
i
h
t
w
o
r
G

t
n
e
m
y
o
l
p
m
E

t
n
e
m
y
o
l
p
m
E

s

m

r
i
f

f
o

.

o
N

4
0
0
2
d
n
a
1
0
0
2
h
t
o
b
g
n
i
t
s
i
x
e
s

m

r
i
F

1
0
0
2
d
n
a
8
9
9
1
h
t
o
b
g
n
i
t
s
i
x
e
s

m

r
i
F

r
a
e
y
d
n
a

,

p
i
h
s
r
e
n
w
o

,

T
&
S
y
b
t
n
e
m
y
o
l
p
m
e

e
g
a
r
e
v
A

.
3

e
l
b
a
T

)

%

(
4
0
0
2
–
1
0
0
2

.

4
5
(cid:2)

.

0
2
1
(cid:2)

.

5
5
1
(cid:2)

.

4
2
(cid:2)

.

7
4
(cid:2)

.

8
0
(cid:2)

.

.

.

1
5
2
(cid:2)

6
9
2
(cid:2)

0
0
2
(cid:2)

.

.

.

6
7
1
(cid:2)

9
6
2
(cid:2)

9
0
1
(cid:2)

.

.

.

3
8
3
(cid:2)

7
0
4
(cid:2)

6
1
3
(cid:2)

.

5
1
(cid:2)

.

5
5
(cid:2)

.

4
0
(cid:2)

.

.

.

0
2
2
(cid:2)

2
5
2
(cid:2)

1
9
1
(cid:2)

.

.

.

7
5
1
(cid:2)

2
0
2
(cid:2)

5
4
2
(cid:2)

4
0
0
2
n
I

,

8
1
0
9
5
0
,
8

,

0
0
3
7
2
6
,
1

,

8
1
7
1
3
4
,
6

2
7
8
1
8
7

,

7
0
9
5
1
3

,

5
6
9
5
6
4

,

1
0
0
2
n
I

5
9
9
,
5
5
1
,
9

4
2
6
,
4
4
5
,
1

1
7
3
,
1
1
6
,
7

4
8
9
,
0
0
8

3
1
4
,
1
3
3

1
7
5
,
9
6
4

,

6
8
9
2
9
2
,
1

8
3
7
,
3
3
0
,
1

8
3
2
0
1
7

,

8
4
7
2
8
5

,

7
6
6
1
5
9

,

6
4
2
1
3
4

,

1
2
4
0
2
5

,

0
8
7
5
6
6

,

8
2
0
8
9
4

,

2
5
7
,
7
6
1

9
6
1
1
7
8

,

,

3
7
6
6
0
2
,
4

,

4
0
5
5
3
3
,
3

9
3
4
6
2
3

,

3
2
7
9
5
1

,

6
1
7
6
6
1

,

7
0
5
3
2

,

2
9
5
,
6

5
1
9
6
1

,

4
2
0
,
8
4
5

4
1
7
,
5
8
4

7
4
2
,
9
0
8

3
7
7
,
9
3
3

4
7
4
,
9
6
4

9
8
2
,
1
8
4

0
6
8
,
3
5
3

9
2
4
,
7
2
1

6
4
3
,
2
2
9

0
5
3
,
2
7
2
,
4

4
0
0
,
0
5
3
,
3

6
3
5
,
7
6
2

4
1
6
,
7
2
1

2
2
9
,
9
3
1

4
8
4
,
5

0
8
8
,
7
2

6
9
3
,
2
2

4
0
0
2
,
1
0
0
2

8
0
2
,
3

9
1
1
,
1

9
8
0
,
2

2
4
6

5
0
3

7
3
3

7
3
9

5
2
5

2
1
4

4
3
8

3
1
4

1
2
4

0
2
4

5
2
3

5
9

1
7
4
,
2

0
3
8

1
4
6
,
1

8
3
3

3
8
1

5
5
1

7
3

2
1

5
2

)

%

(
1
0
0
2
–
8
9
9
1

1
.
0
2
(cid:2)

2
.
1
2
(cid:2)

8
.
9
1
(cid:2)

5
.
1
1
(cid:2)

2
.
5
1
(cid:2)

7
.
8
(cid:2)

5
.
5
(cid:2)

8
.
4
(cid:2)

0
.
4
1
(cid:2)

6
.
0
(cid:2)

6
.
9
(cid:2)

2
.
2
1
(cid:2)

4
.
2
2
(cid:2)

4
.
3
2
(cid:2)

5
.
8
(cid:2)

8
.
2
1
(cid:2)

1
.
0
2
(cid:2)

2
.
0
1
(cid:2)

7
.
3
(cid:2)

1
.
4
(cid:2)

3
.
3
(cid:2)

3
.
4
2
(cid:2)

8
.
6
2
(cid:2)

2
.
3
2
(cid:2)

8
9
9
,
8
4
7
,
2

7
9
5
,
2
0
8
,
3
1

9
9
5
,
3
5
0
,
1
1

6
5
0
,
7
4
4
,
1

1
0
0
2
n
I

0
5
2
,
5
8
5

6
0
8
,
1
6
8

2
4
2
,
6
2
7

9
3
8
,
5
6
3

3
0
4
,
0
6
3

5
7
3
,
8
1
7

9
3
5
,
6
3
3

6
3
8
,
1
8
3

4
0
5
,
4

1
2
4
,
6
8
1

7
1
9
,
1
8
1

1
9
8
,
8
1
1
,
2

9
3
8
,
8
0
5

2
5
0
,
0
1
6
,
1

1
4
2
,
9
4

8
2
1
,
8
2

3
1
1
,
1
2

3
8
7
,
6
7

8
4
4
,
3
2

5
3
3
,
3
5

7
4
3
,
3
7
2
,
7
1

7
2
1
,
9
8
4
,
3

0
2
2
,
4
8
7
,
3
1

0
7
2
,
4
3
6
,
1

8
9
9
1
n
I

3
2
1
,
0
9
6

7
4
1
,
4
4
9

6
0
1
,
8
6
7

4
6
1
,
9
4
3

2
4
9
,
8
1
4

6
4
5
,
2
2
7

1
5
9
,
9
9
2

5
9
5
,
2
2
4

3
2
9
,
4

6
2
3
,
2
5
1

3
0
4
,
7
4
1

7
4
0
,
0
3
4
,
2

2
7
9
,
6
3
6

5
7
0
,
3
9
7
,
1

9
5
1
,
1
5

3
2
3
,
9
2

6
3
8
,
1
2

6
1
0
,
2
3

8
0
4
,
9
6

4
2
4
,
1
0
1

1
0
0
2
,
8
9
9
1

h
t
o
b
n
I

8
4
6
,
7

2
5
0
,
3

6
9
5
,
4

9
3
9
,
1

3
8
9

6
5
9

0
3
9

3
6
5

7
6
3

3
9
5

6
3
4

5
3
2

7
2
2

8

9
2
0
,
1

1
1
7
,
1

9
1
6

2
9
0
,
1

8
7

9
4

9
2

3
4

5
6

8
0
1

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

0
(cid:3)
T
S

0
(cid:4)
T
S

l
l

A

e
v
i
t
c
e
l
l
o
C

E
O
S

u
a
c
a
M

;

n
a
w
i
a
T

;

g
n
o
K
g
n
o
H

.
t
n
e
v
t
n
i
o
J

.
t
n
e
v
t
n
i
o
J

g
n
i
d
l
o
h
e
r
a
h
S

n
g
i
e
r
o
F

e
t
a
v
i
r
P

r
e
h
t
O

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

.
a
n
i
h
C

f
o

s
c
i
t
s
i
t
a
t
S
f
o
u
a
e
r
u
B

l
a
n
o
i
t
a
N
e
h
t

y
b

d
e
d
i
v
o
r
p
a
t
a
D

:
e
c
r
u
o
S

13

Asian Economic Papers

Will Science and Technology Solve China’s Unemployment Problem?

and 2004. Employment has also declined in both periods in collective, shareholding,
and other domestic ªrms. The result for private domestic ªrms is mixed with a
small decline in the ªrst period ((cid:2)3.7 percent) and with an increase in the second
period (22 percent).

Firms with foreign ownership are divided in three groups: joint ventures with ªrms
from Hong Kong, Macau, and Taiwan; joint ventures with ªrms from other coun-
tries; and wholly foreign-owned ªrms. Joint ventures with greater China have had
a growth in employment in both periods, whereas the other type of joint ventures
had a stagnant job growth in the ªrst period and a positive job growth in the
second period. Wholly foreign-owned ªrms have shown the highest growth in
employment, about 22 percent in the ªrst period and about 38 percent in the second
period.

Returning to the relationship between S&T and job growth, our previously ex-
pressed suspicion that a negative relation is caused by the development in SOEs is
only partly supported by the data. Job growth has been poorer in SOEs with S&T
than in SOEs without S&T. However, the same development is also found in all
three groups with foreign ownership where employment has grown faster in ªrms
without S&T. In fact, all types of foreign ªrms with S&T had a negative employment
growth in the ªrst period.

Firms with S&T have a higher employment growth than ªrms without S&T in two
ownership groups, collectives and shareholdings, whereas the results for private
ªrms are inconclusive with a seemingly positive effect in the ªrst period, but a nega-
tive effect in the second period.

These results show that S&T does not have a positive impact on employment. If
anything, the results suggest that ªrms without S&T have increased their employ-
ment faster.

Survival is another mechanism through which S&T might affect employment. In
other words, there might be a positive relation between S&T and the survival of
ªrms, something that is overlooked in Tables 3 and 4 where, obviously, only surviv-
ing ªrms are included. Table 4 includes ªgures on how large a proportion of all
ªrms that were present in, for instance, 1998, survived until 2001. The survival rate
is divided among ªrms with and without S&T. The ªgures show that roughly
59 percent of all ªrms that existed in 1998 survived until 2001. The survival rate de-
creases substantially in the second period, where it amounts to about 40 percent.

14

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

Table 4. Survival by S&T, sector, and year (%)

No. of firms
in 1998

Remained
in 2001

All firms

High tech

Ferrous metals

Transport equipment

Basic chemicals

Textiles

Petroleum products

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

All
ST(cid:3)0
ST(cid:4)0

23,105
11,720
11,385

2,052
570
1,482

430
223
207

1,268
438
830

1,845
850
995

2,294
1,448
846

155
54
101

13,678
6,129
7,549

1,398
334
1,064

233
96
137

878
256
622

1,118
458
660

1,069
612
457

100
28
72

Source: Data provided by the National Bureau of Statistics of China.

%

59.2
52.3
66.3

68.1
58.6
71.8

54.2
43.0
66.2

69.2
58.4
74.9

60.6
53.9
66.3

46.6
42.3
54.0

64.5
51.9
71.3

No. of firms
in 2001

Remained
in 2004

22,375
12,174
10,201

2,385
849
1,536

388
209
179

1,354
535
819

1,757
874
883

1,751
1,094
657

164
61
103

8,887
3,712
5,175

1,137
313
824

181
65
116

673
188
485

671
225
446

634
311
323

101
25
76

%

39.7
30.5
50.7

47.7
36.9
53.6

46.6
31.1
64.8

49.7
35.1
59.2

38.2
25.7
50.5

36.2
28.4
49.2

61.6
41.0
73.8

The exit rate in the ªrst period is broadly in line with the results for other coun-
tries.12 The second period, however, shows an exit rate that is considerably higher
than what is typically the case in other countries. Once more, our exit rate can be
caused by other factors than the “death” of a ªrm and is therefore not directly com-
parable with ªgures from other studies.

The survival rate differs between industries and seems to be particularly high in pe-
troleum and low in textiles. More importantly, there seems to be a positive relation
between S&T and survival: ªrms with S&T are comparably likely to survive in all
industries and in both periods. One plausible reason is that investment in S&T is
typically a long-term decision that should be appealing only to those ªrms who ex-
pect to remain in business over some time.

To sum up the results, the simple tabulations in the tables seem to suggest that, ªrst,
S&T have no positive effect on job-creation, and second, S&T have a positive effect
on ªrm survival. Hence, although the ªgures suggest that S&T do not create jobs,
they seem to maintain jobs by affecting the survival rate.

12 See, for example, Roberts and Tybout (1996) and Bernard and Sjöholm (2003).

15

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

Table 5. Firm characteristics by S&T and year (firm average 1,000 yuan)

Average employment per firm

Export as a share of sales (%)

Import of technology as a share of sales (%)

Profits as a share of sales (%)

Average wage per employee

Value-added per employee

Fixed assets (capital) per employee

ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0
ST(cid:3)0
ST(cid:4)0

1998

836
2,108

20.3
9.7
0.2
0.7
0.0
3.2
6.9
8.9
93.9
112.7
92.1
93.0

2001

701
1,832

22.0
12.3
0.1
0.6
3.9
6.8
10.2
12.8
176.2
211.6
140.5
148.6

2004

917
1,830

31.1
17.0
0.1
0.4
5.4
7.9
14.3
20.3
288.8
438.8
125.4
201.0

Source: Data provided by the National Bureau of Statistics of China.

The main constraint of this analysis is obvious: job growth and ªrm survival are af-
fected by a host of factors other than those included in the tables. If such characteris-
tics differ between ªrms with and without S&T, there is a risk that our comparison
is biased. Indeed, Table 5 shows there to be large differences between ªrms with and
without S&T in all sectors and in all periods. More speciªcally, ªrms with S&T tend
to be relatively large, capital-intensive ªrms with high proªts, productivity, and
wages, and with a large amount of imports of technologies. Firms with no S&T tend
to have a substantially higher share of exports.

Controlling for various factors that affect employment and allowing all Chinese
ªrms to be included in the data require an econometric approach that we now
employ.

5. Econometric model and results

5.1 Model
We use a Heckman two-step estimator to control for the sample selection problem
caused by attrition (ªrms dropping out from the data set) (Puhani 2000). The
Heckman approach controls for the effect of ªrm survival before we estimate the im-
pact of S&T on employment. In the ªrst step, we estimate a probit model for ªrm
exit as speciªed in equation (1). We experiment with using different sets of controls,
ranging from an S&T status dummy only, to the most comprehensive model, which
includes S&T intensity, ownership, skill- and capital-intensities, and a set of dummy
variables to control for export- and import-status, as well as for year- and industry-
speciªc effects. We use the most comprehensive model to calculate the inverse Mills
ratio.

16

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

(cid:2)
P(Exitit) (cid:3) (cid:5)(Zi,t(cid:2)1)

(cid:3) (cid:6) (cid:7) (cid:8)

Zi,t(cid:2)1
(cid:9)
3Capital_intensityi,t(cid:2)1

stS&T_sharei,t(cid:2)1
(cid:7) (cid:10)(cid:8)

(cid:7) (cid:9)

1Firmsizei,t(cid:2)1
(cid:7) (cid:8)

wOwnershipi

(cid:8)

imImport_dummyi,t(cid:2)1

(cid:7) (cid:10)(cid:8)

tYear_dummy (cid:7) (cid:10)(cid:8)

(cid:7) (cid:9)

2Skill_Sharei,t(cid:2)1
exExport_dummyi,t(cid:2)1
indInd_dummyj

(cid:7)
(cid:7)

(1)

In the second step, the inverse Mills ratio is added to the model of employment
growth as an explanatory variable. The employment growth model is speciªed as:13

(cid:11)Xi,t
(cid:10)(cid:8)

(cid:3) InXit
wOwnershipi
(cid:10)(cid:8)

(cid:3) (cid:6) (cid:7) (cid:10)(cid:8)
(cid:2) InXit(cid:2)1
(cid:7) (cid:10)(cid:8)
tYear_dummy (cid:7) (cid:10)(cid:8)
RReg_dummy (cid:7) (cid:12)Millsit

nS&T_sharei,t n

(cid:7) (cid:9)Firmi,t(cid:2)1
(cid:7)
indInd_dummyj
(cid:7) (cid:13)

it,

(cid:7)

(2)

where i is the index for ªrms, j is the index for industries, and t is the index for year.
The model is estimated by applying OLS and ªxed effect estimators on the full data
set as well as on sub-samples by ownership and by industry sector. The variables in-
cluded in the speciªcation are deªned as:

Xit

(cid:3) Employment

S&T_sharei,t n

(cid:3) The ratio of S&T expenditures to sales, where n is the number

of lags.

Firmi,t(cid:2)1

(cid:3) A vector of lagged ªrm characteristics such as size, labor productiv-

ity, skill intensity, export- and import-shares.

Ownershipi

(cid:3) Ownership dummy variables(cid:3) SOE, collective, joint venture
with ªrms from Taiwan, Hong Kong, and Macau, joint venture with ªrms
from other foreign countries, wholly foreign-owned, and private domestic
ªrms.

Yeart

(cid:3) Year dummy variable.

Industryj

(cid:3) Industry dummy variables at the four-digit level.

Reg_dummy (cid:3) Regional dummy variables at the province level.

Millsit

(cid:3) The inverse of Mills ratio from the probit model estimation in Step 1,
calculated as(cid:3) f(
Z
)
, where (cid:14) is the standard normal probability den-
it
1 (cid:2) F
Z
)
(

it

sity function and (cid:5) is the standard normal cumulative density function.

13 See Table A2 in the Appendix for detailed deªnitions of the control variables at the ªrm-

and industry-level.

17

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

Firm characteristics such as size and labor productivity are expressed in log forms.
We try to avoid an endogeneity problem by using lagged values on S&T and other
independent variables in our estimations. However, we will also use a matching ap-
proach, both as a robustness check and as an alternative attempt to control for the
possibility that S&T is a function of, for instance, job growth.

The main advantage of the matching method is the ability to control for
endogeneity. The idea behind the propensity score-matching estimator is that for
every ªrm that performs S&T, we identify an “identical” ªrm that does not perform
any S&T. We then compare job growth in the treated group (performs S&T) and the
control group (does not perform S&T).14 The treatment is deªned by the S&T
dummy variable (S&T_dummyi,t(cid:2)1), namely, whether ªrm i performs S&T activities
or not at time t 1, and employment growth ((cid:11)Xi) is the outcome variable. We use a
set of lagged ªrm characteristics (Firmi,t(cid:2)1), such as ªrm size, labor productivity, ex-
port-share, import-share, capital-intensity, and industry afªliation at the two-digit
level (Industryj) to identify similar ªrms and perform the matching of treated and
control ªrms. The propensity score is estimated as:

p(Firmi,t(cid:2)1, Industryj) (cid:3) Pr{S&T_dummyi,t(cid:2)1

(cid:3) 1|Firmi,t(cid:2)1, Industryj}

(3)

Finally, the average treatment effect on the treated (ATT) is estimated as:

ATT (cid:3) E{E{(cid:11)X1i

(cid:2) (cid:11)X0i|S&T_dummyi,t(cid:2)1

(cid:3) 1, p(Firmi,t(cid:2)1, Industryj)}}.

(4)

5.2 Results
Table 6 shows probit estimations on ªrms’ likelihood to exit from the market and
how this likelihood is affected by ªrm characteristics. A negative coefªcient means
that the likelihood of exit decreases. In addition to controlling for sample selection
bias, we can also make use of this estimation to identify the factors that affect ªrm
exit. As previously discussed, the data are constructed in such a way that we cannot
distinguish death of ªrms from two other forms of exit: a change in ownership or a
decline in size to below the threshold. Bearing this caveat in mind, we notice in the
ªrst column that S&T has a positive and statistically signiªcant impact on survival:
ªrms with any S&T are signiªcantly less likely to exit compared to ªrms without
S&T.

In the previous sections, we have seen that ªrms with and without S&T differ in a
number of aspects, which could also affect the exit rate. We try to control for such

14 We apply the nearest neighbor matching with replacement; see Becker and Ichino (2002) for

more details.

18

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

Table 6. Firm exit (probit estimations; dependent variable: exit

1, survival

0)

S&T dummy

S&T intensity

Size

Ownership
SOE

Ownership
collective

Ownership
JV_KTM

Ownership
JV_Foreign

Ownership
foreign

Ownership
shareholding

Ownership
private

Skill share

Capital intensity

Export dummy

Import
dummy

Year dummy

Industry dummy
(4-digit)

(1)
(cid:2)0.338**
(0.008)

(2)
(cid:2)0.228**
(0.009)

(cid:2)0.254**
(0.003)
(cid:2)0.224**
(0.040)
(cid:2)0.083*
(0.040)
(cid:2)0.318**
(0.041)
(cid:2)0.330**
(0.042)
(cid:2)0.529**
(0.044)
(cid:2)0.248**
(0.040)
(cid:2)0.221**
(0.042)

Yes

Yes

Yes

Yes

(3)
(cid:2)0.211**
(0.009)

(cid:2)0.247**
(0.003)
(cid:2)0.220**
(0.040)
(cid:2)0.079*
(0.040)
(cid:2)0.291**
(0.041)
(cid:2)0.304**
(0.042)
(cid:2)0.489**
(0.045)
(cid:2)0.240**
(0.040)
(cid:2)0.219**
(0.042)
(cid:2)0.011
(0.009)
(cid:2)0.0001**
(0.00004)
(cid:2)0.103**
(0.010)
(cid:2)0.045**
(0.017)

Yes

Yes

(4)

(5)

(6)

(cid:2)0.001
(0.002)

(cid:2)0.018**
(0.005)
(cid:2)0.273**
(0.003)
(cid:2)0.266**
(0.040)
(cid:2)0.078*
(0.040)
(cid:2)0.295**
(0.041)
(cid:2)0.319**
(0.042)
(cid:2)0.478**
(0.045)
(cid:2)0.275**
(0.040)
(cid:2)0.207**
(0.042)

Yes

Yes

Yes

Yes

(cid:2)0.014**
(0.007)
(cid:2)0.259**
(0.003)
(cid:2)0.255**
(0.040)
(cid:2)0.074*
(0.040)
(cid:2)0.269**
(0.041)
(cid:2)0.290**
(0.042)
(cid:2)0.442**
(0.045)
(cid:2)0.261**
(0.040)
(cid:2)0.208**
(0.042)
(cid:2)0.028*
(0.015)
(cid:2)0.0001**
(0.0004)
(cid:2)0.119**
(0.009)
(cid:2)0.155**
(0.016)

Yes

Yes

No. of obs.

170,489

165,964

165,796

165,964

165,964

165,796

Source: Data provided by the National Bureau of Statistics of China.

Note: Robust standard errors are within parentheses. *Statistically significant at the 5 percent level. ** Statistically significant at the

1 percent level.
JV_KTM = joint ventures with firms from Hong Kong, Taiwan, and Macau; JV_Foreign = joint ventures with foreign firms outside of
Hong Kong, Taiwan, and Macau.

characteristics in the following estimations. Column (2) shows that large ªrms are
substantially less likely to exit. Moreover, all the included ownership variables are
statistically signiªcant with negative signs showing that ªrms with any of these
ownerships are less likely to exit than the group of comparison: other domestic
ªrms.15 We can also see that the coefªcients differ between ownership groups with a
large negative coefªcient for foreign ownership and a smaller negative coefªcient
for collective ownership. The inclusion of additional variables decreases the effect of

15 The group other domestic ªrms consists of state-collective jointly operated enterprises, other
jointly operated enterprises, limited liability enterprises, and shareholding limited enter-
prises.

19

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

S&T on survival in column (1); thereby suggesting that some of the previously esti-
mated effect is caused by differences in other characteristics than S&T.

We include a number of new variables in column (3). The results show that ªrms in-
tegrated with the global economy in terms of export or import of technology are rel-
atively less likely to exit. Moreover, a high skill-share or high capital-intensity has
no, or a very limited, impact on survival and the inclusion of these two additional
control variables does not affect the other coefªcients.

The previous estimations show that ªrms with any S&T are less likely to exit than
ªrms without S&T. In columns (4)–(6), we continue to examine if the amount of S&T
affects exit by examining the effect of S&T intensities on ªrm survival. The results
suggest that the higher the S&T intensity, the less likely is the ªrm to exit. The other
coefªcients are similar to previous estimations.16

Next, we turn to our question of main interest: how S&T affects job growth. We ap-
proach the issue by estimating regressions in Table 7 with growth in employment as
the dependent variable and with various independent variables, including the S&T
intensity, which potentially affects job growth. As previously stated, it is important
to control for the possible bias caused by a sample where we only observe growth in
employment in surviving ªrms. The need to control for this aspect seems particu-
larly high in view of the positive effect of S&T on job survival found in Table 6. We
therefore calculate the Mills ratio from column (6) in Table 6 and then include it in
the job-growth regressions.

The time it takes for S&T to affect job growth is uncertain. We therefore start in col-
umn (1) by including ªve lags of S&T. The results show that only lag 1 is statistically
signiªcant with a positive sign. One disadvantage with the inclusion of many lags is
that it substantially reduces the sample. This is seen in column (2) where the sample
increases from 16,834 observations (column (1)) to 130,150 observations when only
one lag is included. The change of sample size presumably explains the change in
the result for S&T, which is not found to affect job growth in estimation 2. We see
that large ªrms have a relatively low job-growth when looking at the other variables
in the OLS estimations in columns (1) and (2). Moreover, there is a positive impact
on job growth of productivity, skills, export, and import of technology. Job growth
also differs between different ownership types.

16 We did also try with the more narrow measure on technology development, R&D. The re-

sults did not change in any major respect.

20

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e
–
p
d

/

l

f
/

/

/

/

/

9
2
1
1
6
8
3
0
4
4
a
s
e
p
_
a
_
0
0
0
1
3
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Will Science and Technology Solve China’s Unemployment Problem?

Table 7. Employment growth regression (dependent variable: employment growth)

Without Mills ratio

With Mills ratio

S&T share
(lagged (cid:2)1)

S&T share
(lagged (cid:2)2)

S&T share
(lagged (cid:2)3)

S&T share
(lagged (cid:2)4)

S&T share
(lagged (cid:2)5)

Year dum

Industry dum

Regional dum

Lagged firm size

Lagged labor productivity

Ownership SOE

Ownership collective

Ownership JV_KTM