The determinants of industrial firms` export behavior in China

Determinants of Exporting Behavior of Chinese Industrial Firms: An Empirical
Test on Heterogeneous Theory1
Xu Kangninga Qiu Bina Gao Boa Sun Shaoqinb
[email protected]
[email protected]
[email protected] [email protected]
Abstract
China has become the largest exporter in the world and tremendously large quantities
of explanations have been made by using traditional theories in the area of
international division of labor. Since Melitz (2003) developed a dynamic industry
model with heterogeneous firms and explained that high-productivity firms self-select
themselves into export markets, which implies a new causal link between firm
productivity and exporting behavior. Therefore, the behavior of Chinese industrial
firms and its determinants need to be retested in the scenario of various
heterogeneities among Chinese firms. This paper analyses the determinants of China’s
industrial exports by using firm-level data 2001 through 2007. In the meantime, this
paper also tests various fixed effects that impact exporting behavior of Chinese firms
to strengthen the robustness of the study. This paper continues to explore the impact
of processing trade which dominates China’s exports. We find that processing trade
distorted the relation between productivity and export behavior of firms. Furthermore,
if we exclude the role of processing trade, Melitz model is still applicable in China.
Keywords:
China, Export Behavior, Productivity, Heterogeneous Theory, Probit
Model
1
Supported by China National Social Science Projects (10BJY081, 10CJY052, 09AZD047) and New Century
Excellent Talents Program in University by Ministry of Education of China (NCET-09-029).
a School of Economics and Management, Southeast University, Nanjing, 210096, China
b School of Economics and Management, Nanjing University of Information Science & Technology, Nanjing,
210044, China
1
1. Introduction
International trade and gains from it have always been the research focus in the field of trade
theories and international economics. It is believed by traditional trade theories that differences in
factor endowment and productivity between two countries lead to their own comparative
advantages over the other, and therefore, inter-industry trade between them. Productivity
improvement derived from specialization and trading results in the aggregate welfare growth of
trading participants. With intra-industry trade and intermediate product trade growingly prevailing
worldwide, new trade theories are proposed from the perspective of scale economies and
consumers’ preference diversity. These theories consider scale economies and product diversity as
the two main sources of trade gains. Recently, studies in the area of trade are increasingly exposed
to micro-econometric analyses where firms are assumed to be heterogeneous. These New-New
trade theories based on firm heterogeneity (with the representative theoretical study by Melitz,
2003) present researchers a brand new perspective with their main view that productivity is a key
determinant of a firm’s export behavior, which means that a more productive firm chooses to
export while a less productive firm produces only for domestic market. International trade will
gradually force the least productive firms to exit market entirely. In that scenario, more resources
will be allocated to more productive firms. Such intra-industry reallocations of resources among
firms within an industry will eventually give rise to aggregate industry productivity growth.
Prior to Melitz’s (2003) model, quantity of studies have been conducted to test correlation
between export behavior and productivity performance. The majority of them have reached a
widely accepted conclusion that exporters are generally more productive than non-exporters. To
interpreter this conclusion, both domestic and foreign researchers identify two effects. One is
learning-by-doing effect. It’s easier for exporters to obtain technology spillovers from
international buyers than non-exporters and to better off their productivity performance. The other
one is self-selection effect, which supposes that a firm must be productive enough to bear sunk
costs2 incurred in exporting and meanwhile be profitable as well, thus, productivity is seen as a
cause rather than a consequence of exporting, so exporters are not supposed to benefit from its
exporting in terms of productivity improvement It’s an interesting question to ask whether such a
widely accepted view as well as its interpretations are equally applicable to China, which is such a
giant exporter in the world.
Since the implementation of Reform and Opening-up to the outside world in the end of 1970s,
China’s foreign trade has witnessed a persistent and impressive annual growth. Its exports has
2
In Melitz’ model (2003), not only iceberg trade costs (e.g. transportation costs and tariffs) exist, but also fixed
entry costs, for example, market research, product modification, compliance to higher standards,
establishment of sale channels in foreign market, and so on are present as well. Of course, sometimes goverments
will share a portion of fixed entry costs. Prior to Melitz’s model, Roberts and Tybout(1997), Clerides, Lach and
Tybout(1998),Das, Roberts and Tybout (2007), Bernard and Jensen(2004), Bernard and Wagner(2001) have
contributed to studies of trade costs similar to that assumed in Melitz’ model.
2
increased annually at the average rate of 18.5% since 1990s, and its merchandise trade share in the
whole world has grown at a rate far above the world average rate, from 3.5% in 1999 to 9.3% in
2008 (Zhang Liqing,Sun Junxin, 2010). However, it’s worth noting that China’s trade structure is
quite different from that of developed countries. In fact, a large fraction of China’s exports are in
the category of processing trade which is attributed to global production networks (GPNs) as well
as multinational enterprises (MNEs) who are exactly the key driver of the formation of global
production networks (GPNs). These processing trade-oriented firms in China only perform on the
operations of processing and assembly which are low value-added slices in a commodity’s value
chains, with nearly all their final outputs to be sold to international market. Then what about the
productivity of China’s exporters under such unique trade regimes? Will they still have higher
productivity relative to non-exporters? Is productivity a key determinant of a firm’s export
behavior? In the frame work of firm heterogeneity model, this paper investigates the determinants
of a firm’s export behavior. In particular, we analyze the productivity’s effect on a firm’s export
behavior, with the impacts of other factors controlled, such as a firm’s scale, ownership type,
region, capital intensity and especially the influences of processing trade on firms’ export
behavior.
The remainder of this paper is as follows: Section 2 reviews relevant literature on the impacts of
productivity on a firm’s export behavior; Section 3 presents a descriptive analysis of the
characteristics of exporters and non-exporters; all empirical results are discussed in Section 4 in
terms of the determinants of a firm’s export behavior in the frame work of firm heterogeneity
model while Section 5 provides some concluding discussions.
2. Literature Review
Prior to Melitz’s work (2003) in which he constructs a dynamic industry model with
heterogeneous firms, a great number of empirical studies of correlation between productivity and a
firm’s export behavior have been well documented. A key finding drawn from them is that
exporters are generally more productive than non-exporters. Both theoretical and empirical studies
are briefly reviewed as follows.
Models of Melitz (2003) and Bernard et al.(2003)are central to theoretical studies of correlation
between productivity and a firm’s export behavior. Those afterwards, for example, Helpman et
al.(2004), Bernard et al.(2004), Falvey et al.(2004) and Melitz, Ottaviano(2005)extend the former
two studies from different perspectives.
Melitz (2003) establishes a dynamic industry model with heterogeneous firms. It is an augment of
Krugman (1980)’s model based on assumptions of monopolistic competition and scale economies,
and he turns the so-called representative firm which most of traditional trade theories rely on into
3
heterogeneous firms. On assumptions of imperfect competition, Bernard(2003)also sets a trade
model where firm heterogeneity is incorporated into the Richard model to generate firm-specific
comparative advantage. Both of Melitz (2003) and Bernard(2003)draw the same conclusion that
only more productive firms choose to export.
Bernard & Jensen (1995) pioneer the empirical analysis of the links between exporting and
productivity. Using the firm-level data, they examine the differences between exporters and
non-exporters in the US, finding that exporters are generally more productive than non-exporters,
but their productivity growth exhibits no obvious differences after exporting, which proves that
exporters are more productive before their entries into export market. Following Bernard & Jensen
(1995), using the firm-level data on firms in Mexico, Morocco and Colombia, Clerides, Lach &
Tybout (1998) find that due to the sunk costs involved in exporting, not all firms will self-select to
export expect those who are productive enough to bear them and simultaneously be able to gain
profits from exporting will choose to export. Besides, studies on Germen (Arnold & Hussinger,
2005), Spain (Delgado et al. 2002), Korea (Aw et al. 2000), UK(Girma et al. 2004 ), Golombia
(Isgut, 2001)and Canada ( Lileeva & Trefler, 2010) all provide evidences of significant impact of
productivity on a firm’s exporting decision.
The above studies are mostly about developed countries while analyses on developing countries
are very few. Using a panel of manufacturing firms in nine African countries, Biesebroeck (2005)
find that high productivity will induce firms to export. In recent two years, Chinese scholars make
some empirical studies on the basis of firm-level data. For instance, using data on domestic firms
in Jiangsu Province (Zhang Jie, et al, 2008), data on firms in Zhejiang Province (Yi jingtao, 2009),
data from China’ Industrial Census in 2005(Tang Yihong, Lin Faqin, 2009 ) , data on
manufacturing firms above designated size 3 in China(Zhang Liqing, Sun Junxin, 2010),
industrial firms in China (Li Chunding, 2009,2010), they all verify that high productivity serves
as a key determinant of export behavior.
Current studies all consider productivity as a key determinant of a firm’s export behavior, which
provides us a perspective from which to examine a firm’s export behavior. What is completed in
this paper adds to the existing literature. Aiming at testing whether firm heterogeneity model
applies to China, this paper examines how productivity affects a firm’s export behavior and
whether processing trade will distort this effect. Meanwhile, we investigate other factors’ impacts
on export behavior, such as a firm’s R&D input, new product development, scale, ownership type,
region.
3
Firms of designated size are firms with annual revenues 5 million RMB or more.
4
3. Observations of Exporters’ Characteristics
3.1 Sample and Data Set
Data used in this paper are extracted from China Industrial Firms Dadabase in China, which
covers basic information of those firms especially on a variety of performance indicators over a
time period of 1999 to 2008, including firm code, firm’s name, province code (the province where
the firm is located), industry type, registration type, gross industrial output value, export value,
number of employed persons, industrial intermediate input, R&D input, new product share, annual
average of net value of fixed assets, wages and so on. With years in which data are incomplete
removed, the sample period in this paper is finally defined as 2001-2003 and 2005-2007.
To obtain the ideal samples used in our analysis, the following treatments have been done. Firstly,
for each year, firms with missing or unreasonable value (value is zero or negative) on any
indicators are deleted, but some indicators are allowed to be zero, for example, export value, R&D
input, output value of new products, etc. Simultaneously, firms that are experiencing halted
production, construction and bankruptcy are removed too. Secondly, to make firms in our sample
comparable over the sample period, we only select firms with persistent performance from 2001 to
2007. Thirdly, indicators including industry type, region, etc are deemed as invariant over the
sample period, and information on such indicators are sourced from data in 2007. Fourthly, current
prices are all converted to constant prices using price deflators.
Eventually, our sample consists of 61043 firms which yield 366258 observations in 6 years. As
shown in Table 1, there are 27705 light industry firms, accounting for 45.39% of the total firms in
our sample. The number of small-sized firms is the largest, with its share in the total sample firms
74.5%; the medium-sized firms have the second largest share while the large-sized firms’ share is
the least (2.68%). Foreign funded firms (including firms with funds from Hong Kong, Macao and
Taiwan) accounts for 7.37% of the overall firms. Table 2 gives details on the percentage of
exporting firms in the total firms by sector. As we can see, the percentage of exporting firms in
sector of Manufacture of Non-metallic Mineral Products is 8.22%, which is the largest among all
sectors. Sectors next to it are Manufacture of Textile, Manufacture of Raw Chemical Materials and
Chemical Products, Manufacture of General Purpose Machinery and Manufacture of Electrical
Machinery and Equipment, with the corresponding percentages of 7.82%, 7.37%, 7.29% and
6.18%, respectively. For sectors of Manufacture of Metal Products, Manufacture of Transport
Equipment, Manufacture of Textile Wearing Apparel, footwear and Caps, Manufacture of Plastics
and Manufacture of Foods, their percentages of exporting firms are all higher than 4%. With
regards to regional distribution of firms, as shown in Table 3 that nearly three quarters of firms are
located in eastern provinces, especially firms in Zhejiang, Guangzhou and Jiangsu all account for
more than 10% of the total firms in our sample.
5
3.2 Observations of Export Behavior
As shown in Table 1, firms of different types differ in export tendency. Firms in light industries are
more likely to export relative to firms in heavy industries, ranging from 17 percentage to 20
percentage of exporters. Also, judging by firm size, the proportion of exporting firms in
large-sized firms is the greatest while that in small-sized firms is the least, which indicates that a
large-sized firm is more capable of bearing sunk costs incurred in exporting and thus is more
likely to export. From the perspective of ownership, foreign funded firms and firms with funds
from Hong Kong, Macao and Taiwan enjoy a greater proportion of exporting firms than domestic
firms. The percentage of exporting firms in the type of foreign firms is about 70% while that in the
type of domestic firms is from 20% to 25%.
Table 1
Percentage of Exporting Firms by Firm Type
percentage of exporting firms within a firm type (%)
number
of firms
2001
2002
2003
2005
2006
2007
light industry firms
27705
43.39
45.32
46.01
46.71
46.36
44.70
heavy industry firms
33338
25.03
25.92
26.61
29.24
29.16
27.61
large-sized firms
45263
28.96
30.24
30.53
31.89
31.42
29.56
medium-sized firms
14145
44.23
45.98
47.78
50.74
51.32
50.44
small-sized firms
1635
61.16
61.35
63.73
65.81
66.48
65.81
domestic firms
44685
21.47
22.82
23.37
25.44
25.14
23.17
8324
65.59
66.79
67.80
68.22
68.34
67.65
68.86
70.19
70.21
69.78
foreign firms with
funds from Hong kong,
Macao and Taiwan
foreign funded firms
8034
66.13
67.69
Source: China Industrial Firms Database, authors’ calculation.
Table 2 describes the proportion of exporting firms by province. It’s easily seen that a larger
proportion of exporters is associated with eastern region than central and western regions. In
eastern provinces, more than 50% of firms in Guangdong Province choose to export, and similarly,
a relatively large proportion of firms in Jiangsu and Zhejiang are exporters as well. The
information in Table 2 is in line with China’s trade structure in terms of regional distribution of
exporting firms.
Table 2
number
of firms
Percentage of Exporting Firms by Province
percentage of exporting firms out of the total firms within a province
2001
2002
2003
2005
2006
2007
Beijing
1217
19.64
24.16
25.14
31.31
32.13
32.13
Tianjing
1142
39.05
42.73
42.82
41.59
41.59
41.51
Hebei
2541
16.49
17.43
17.36
17.32
17.36
17.08
Liaoning
1954
34.14
35.31
35.82
33.11
32.40
32.96
Shanghai
4310
40.39
41.09
42.09
41.95
41.65
41.18
Jiangsu
7863
36.73
37.26
36.53
34.78
34.97
34.78
Zhejiang
9444
45.11
48.73
50.40
51.81
52.00
51.85
6
Fujian
3291
8.51
8.87
9.57
9.75
11.35
10.99
Shandong
4822
31.56
32.21
32.77
32.08
32.02
31.02
Guangdong
8780
53.74
55.50
56.56
57.20
57.28
56.79
Hainan
115
11.30
10.43
11.30
10.43
10.43
10.43
45479
37.80
39.46
40.14
40.20
40.36
40.03
Shanxi
824
8.62
9.10
9.22
10.56
10.19
10.19
Jilin
553
13.02
12.66
13.92
15.37
16.46
14.65
Heilongjiang
564
8.51
8.87
9.57
9.75
11.35
10.99
Anhui
1194
22.78
23.37
23.20
22.53
24.04
22.36
Jiangxi
685
12.26
14.01
16.20
17.08
15.77
18.83
Henan
1926
7.53
8.46
8.93
60.64
50.88
9.50
Hubei
1325
12.68
13.96
12.23
13.28
12.83
13.81
Hunan
1465
15.63
16.25
17.27
16.31
16.93
15.84
sum(central region)
8536
12.76
13.54
13.85
25.73
23.81
14.30
Chongqing
844
16.11
17.42
18.60
19.91
20.62
20.50
Sichuan
1982
9.54
11.15
12.31
12.56
13.02
12.92
Guizhou
584
9.93
9.25
9.76
8.90
9.59
8.22
Yunnan
708
12.99
12.29
13.56
12.71
12.99
12.29
Xizang
13
0.00
0.00
0.00
0.00
0.00
0.00
Shanxi
720
13.75
13.47
13.89
13.61
14.44
14.31
Gansu
408
3.43
4.17
4.17
4.90
5.39
4.41
Qinghai
76
6.58
7.89
9.21
6.58
3.95
5.26
Ningxia
105
13.33
16.19
14.29
15.24
15.24
13.33
Xinjiang
334
6.29
7.19
9.28
8.98
10.48
9.58
Neimenggu
583
9.95
12.01
12.86
11.66
11.15
10.12
Guangxi
671
17.73
17.88
18.78
20.72
19.37
20.12
sum(western region)
7028
11.45
12.24
13.16
13.30
13.59
13.22
sum(eastern region)
Source: China Industrial Firms Database, authors’ calculation.
Table 3 presents the proportion of exporting firms by sector. Generally, the proportions of
exporting firms in manufacturing sectors are relatively high but differ substantially from sector to
sector, ranging from 10% to 80%.The sector of Manufacture of Articles For Culture, Education
and Sport Activities has the largest proportion of exporting firms, 80% of its firms choose to
export while only 10% of firms in sector of Processing of Petroleum, Coking, Processing of
Nuclear Fue are exporters which makes it the sector with the least percentage of exporting firms.
70% of firms in sectors of Manufacture of Textile Wearing Apparel, Footwear and Caps and
Manufacture of Leather, Fur, Feather and Related Product, and about 65% of firms in sector of
Manufacture of Communication Equipment, Computers and Other Electronic Equipment are
exporters. The above figures indicate that for labor-intensive sectors and high-tech sectors aiming
at taking advantage of labor abundance in China, the proportion of exporting firms is relatively
high. Regarding time dimension, the proportions of exporting firms for all sectors all grow over
time, and an increasing number of firms have gradually enter export market.
7
Table 3
Percentage of Exporting Firms by Sector
percentage of exporting firms out of the total firms
Number
industry
within a sector(%)
of firms
2001
2002
2003
2005
2006
2007
(6)Mining and Washing of Coal
928
3.23
2.80
2.59
7.54
6.36
2.59
(7)Extraction of Petroleum and Natural Gas
42
9.52
11.90
11.90
14.29
11.90
14.29
(8)Mining and Processing of Ferrous Metal Ores
199
0.50
2.01
1.51
2.01
1.01
1.01
Metal Ores
357
7.00
6.16
5.60
8.40
8.40
5.32
(10)Mining and Processing of Nonmetal Ores
383
17.49
18.54
16.71
16.19
15.67
12.79
(9)Mining and Processing of Non-Ferrous
3
33.33
33.33
33.33
33.33
33.33
33.33
(12)Processing of Food from Agricultural Products
2523
25.21
26.36
26.48
27.78
27.43
23.62
(13)Manufacture of Foods
1285
29.34
31.67
32.76
36.03
36.11
33.23
(14)Manufacture of Beverages
956
15.48
15.90
17.89
19.35
19.25
16.32
(15)Manufacture of Tobacco
68
22.06
20.59
22.06
22.06
20.59
20.59
4771
49.21
51.44
51.94
51.62
50.16
47.96
2650
69.96
73.40
72.87
71.47
71.43
70.60
1393
70.85
72.72
73.58
73.30
72.36
71.79
(19)Processing of Timber, Manufacture of Wood,
670
33.88
38.66
37.76
42.39
43.43
42.39
(20)Bamboo, Rattan, Palm and Straw Products
606
48.35
51.65
53.80
56.77
57.10
55.61
(21)Manufacture of Furniture
1783
16.71
17.55
17.16
18.28
17.84
15.82
(22)Manufacture of Paper and Paper Products
1168
12.16
11.90
12.24
16.27
17.55
16.27
(23)Printing, Reproduction of Recording Media
856
80.84
84.93
85.16
84.35
83.76
83.53
316
11.39
11.39
12.66
15.19
12.97
10.13
4499
27.09
28.54
28.58
30.92
29.96
28.25
Chemical Products
1525
27.15
28.52
29.77
31.67
30.16
29.64
(27)Manufacture of Medicines
267
21.72
23.97
25.09
28.09
26.97
25.47
(28)Manufacture of Chemical Fibers
764
38.74
39.79
41.10
45.29
44.90
42.28
(29)Manufacture of Rubber
2565
37.82
40.43
41.17
42.14
43.27
42.07
(30)Manufacture of Plastics
5020
17.09
18.15
18.61
22.53
22.63
19.20
Mineral Products
1060
16.42
17.26
17.08
20.47
22.17
20.66
(32)Smelting and Pressing of Ferrous Metals
1073
20.97
23.30
26.10
28.05
27.31
24.70
(33)Smelting and Pressing of Non-ferrous Metals
2769
40.41
41.46
42.25
43.08
43.05
42.25
(34)Manufacture of Metal Products
4447
31.57
32.27
33.10
35.53
35.19
34.47
(35)Manufacture of General Purpose Machinery
2124
28.48
29.14
30.18
35.36
34.60
33.05
(36)Manufacture of Special Purpose Machinery
2793
26.92
28.64
29.97
32.83
34.51
33.66
(37)Manufacture of Transport Equipment
3771
37.26
38.48
40.20
40.84
41.21
41.50
(39)Manufacture of Electrical Machinery
2018
63.83
64.37
65.61
66.20
65.96
66.25
(11)Mining of Other Ores
(16)Manufacture of Textile
(17)Manufacture of Textile Wearing Apparel,
Footwear and Caps
(18)Manufacture of Leather, Fur, Feather and
Related Products
(24)Manufacture of Articles For Culture,
Education and Sport Activities
(25)Processing of Petroleum, Coking, Processing
of Nuclear Fuel
(26)Manufacture of Raw Chemical Materials and
(31)Manufacture of Non-metallic
8
and Equipment
(40)Manufacture of Communication Equipment,
981
48.83
51.27
55.05
55.05
54.13
53.62
1161
76.74
76.31
77.61
75.97
77.52
74.76
47
12.77
12.77
12.77
8.51
8.51
8.51
2127
0.75
0.56
0.66
3.43
2.59
0.38
And Heat Power
110
1.82
0.91
0.91
1.82
1.82
0.00
(45)Production and Supply of Gas
965
0.31
0.21
0.21
1.55
1.35
0.21
0.31
0.21
0.21
1.55
1.35
0.21
Computers and Other Electronic Equipment
(41)Manufacture of Measuring Instruments and
Machinery for Cultural Activity and Office Work
(42)Manufacture of Artwork and
Other Manufacturing
(43)Recycling and Disposal of Waste
(44)Production and Supply of Electric Power
965
Source: China Industrial Firms Database, authors’ calculation.
(46)Production and Supply of Water
3.3 Characteristics Comparison between Exporters and Non-exporters
The above figures show that not all firms export. Even within the same sector, firms of different
sizes and with different ownerships differ in the percentage of exporting firms. In the following
part, we make a comparison between exporters and non-exporters by sector in terms of R&D input,
new product development, wages, capital-labor ratio, labor productivity and TFP, in an attempt to
explore the factors determining a firm’s export behavior.
Table 4 shows the characteristics comparison between exporters and non-exporters, where total
value of wages divided by annual average number of employed persons is used as a proxy for
wage per capita (in thousand yuan per capita), and annual average of net value of fixed assets
divided by total number of employment denotes capital-labor ratio (in thousand yuan per capita,
values in the table are in natural logarithm), and value added of industry in each sector divided by
annual average number of employed persons measures labor productivity (in thousand yuan per
capita, values in the table are in natural logarithm). As data on value added of industry in some
sectors are not available in some years, alternatively, it can be calculated by the production
approach as follows: value added of industry = gross industrial output - industrial intermediate
input + value-added tax. Data on gross industrial output, industrial intermediate input and
value-added tax are all available in the database. TFP is calculated by the SOLOW residual
procedure where the production function is Cobber-Douglas function as follows:

1
Y i  Ai K i Li
Where Y i is value added of industry in each sector, and K i denotes fixed capital input
and labor input, respectively. Now, both sides of equation above are set in natural logarithm:
ln Y i / Li  ln A   ln K i / Li
Then we run regressions on this equation to obtain the value of  . Next, the TFP of each firm can
be gained by the following equation:
TFPi  ln Y i / Li   ln K i / Li
9
It’s noteworthy that data in 2004 are missing, so the calculation of TFP is made on a two-stage
basis, i.e. from 2001 to 2003 and from 2005 to 2007.
In terms of R&D, exporters have a higher R&D intensity than non-exporters in 23 manufacturing
sectors out of 28 total sectors (sector of Manufacture of Artwork and Other Manufacturing not
included), While for sectors of Manufacture of Textile Wearing Apparel, Footwear and Caps,
Manufacture of Leather, Fur, Feather and Related Products, Manufacture of Plastics, Manufacture
of Communication Equipment, Computers and Other Electronic Equipment and Manufacture of
Measuring Instruments and Machinery for Cultural Activity and Office Work, R&D intensity of
non-exporters is slightly higher than that of exporters.
On a whole, for exporters, the percentage of firms with new products is generally higher than that
of non-exporters, which suggests that as a firm aiming at international market, it has to develop
new products to satisfy as differentiated needs as possible. In fact, to reinforce its competitiveness
in international market, a firm also has to enhance the degree of product differentiations.
In terms of per capita wage, exporters pay higher wages than non-exporters in 34 sectors out of 39
total sectors, with exceptions in 5 sectors as follows: Mining and Processing of Ferrous Metal
Ores (8), Manufacture of Paper and Paper Products (22), Manufacture of Communication
Equipment, Computers and Other Electronic Equipment (40), Manufacture of Measuring
Instruments and Machinery for Cultural Activity and Office Work (41) and Production and Supply
of Water (46).
Out of 28 manufacturing sectors, there are 9 sectors where capital-labor ratio of exporters is higher
than that of non-exporters. These 9 sectors are: Manufacture of Foods (13), Manufacture of Textile
Wearing Apparel, Footwear and Cap (17), Manufacture of Leather, Fur, Feather and Related
Products (18), Manufacture of Leather, Fur, Feather and Related Products (19), Manufacture of
Furniture (21), Manufacture of Articles For Culture, Education and Sport Activities (24),
Manufacture of Non-metallic Mineral Products (31) and Manufacture of Electrical Machinery and
Equipment (39). We note that most of them are labor-intensive, and that sector of Manufacture of
Electrical Machinery and Equipment (39) is greatly featured by processing trade.
Out of 39 sectors, there are 16 sectors where labor productivity of exporters is less than that of
non-exporters, and there 22 sectors where TFP (including the 16 sectors drawn when we measure
productivity using labor productivity) of exporters is less than that of non-exporters. Further
analysis reveals that these sectors are either labor intensive or processing trade intensive. What is
revealed here seems to be against the predictions by firm heterogeneity model that exporters are
generally more productive than non-exporters.
10
Table 4
industry
Comparison of Characteristics between Exporters and Non-exporters from 2001 to 2007
R&D
new product
share
share
Wages
per capita
capital-labor
labor
ratio
productivity
TFP
code
N-EX
EX
N-EX
EX
N-EX
EX
N-EX
EX
N-EX
EX
N-EX
EX
6
0.09
0.50
0.01
0.34
13.84
16.97
3.35
4.08
3.44
4.03
2.30
2.65
7
0.35
0.61
0.09
0.03
29.91
104.56
6.19
6.14
5.66
7.52
3.56
5.43
8
0.09
0.17
0.01
0.21
13.83
12.77
3.69
3.68
4.32
3.75
3.07
2.50
9
0.10
0.09
0.01
0.15
12.13
13.71
3.90
3.82
4.22
4.24
2.90
2.95
10
0.07
0.17
0.01
0.16
12.01
12.87
3.36
3.77
3.75
4.04
2.61
2.76
11
0.36
0.00
0.33
0.00
4.57
15.21
2.69
4.71
3.69
4.33
2.78
2.73
13
0.11
0.15
0.04
0.10
12.44
13.54
4.01
3.96
4.43
4.34
3.07
2.99
14
0.18
0.24
0.07
0.13
13.84
17.61
3.95
4.14
4.08
4.13
2.74
2.72
15
0.17
0.29
0.10
0.23
12.44
21.73
4.25
4.71
4.20
4.69
2.76
3.09
16
0.37
0.69
0.10
0.39
32.99
53.69
5.00
5.66
5.11
6.48
3.42
4.55
17
0.06
0.12
0.04
0.12
11.53
13.66
3.64
3.53
3.81
3.73
2.58
2.53
18
0.08
0.06
0.03
0.05
12.73
13.69
2.79
2.64
3.56
3.35
2.61
2.45
19
0.10
0.09
0.04
0.06
12.51
13.03
3.04
2.60
3.83
3.39
2.80
2.51
20
0.07
0.12
0.04
0.11
10.98
13.11
3.53
3.40
3.92
3.79
2.72
2.64
21
0.11
0.10
0.05
0.09
12.49
15.11
3.49
3.38
3.82
3.72
2.64
2.57
22
0.07
0.11
0.03
0.11
20.64
17.68
3.91
4.09
3.97
4.11
2.64
2.72
23
0.07
0.14
0.03
0.12
15.92
19.60
4.19
4.24
3.92
3.88
2.50
2.44
24
0.10
0.12
0.05
0.08
13.04
13.84
2.96
2.85
3.65
3.36
2.65
2.40
25
0.16
0.41
0.04
0.21
16.38
29.11
4.47
5.39
4.79
5.14
3.27
3.32
26
0.18
0.34
0.07
0.24
14.72
21.98
3.90
4.35
4.30
4.55
2.98
3.08
27
0.46
0.58
0.23
0.47
15.43
21.72
4.25
4.42
4.31
4.45
2.86
2.95
28
0.14
0.32
0.07
0.30
13.65
19.29
4.38
4.94
4.33
4.54
2.84
2.86
29
0.14
0.22
0.06
0.19
11.70
15.11
3.38
3.57
3.78
3.76
2.63
2.55
30
0.10
0.10
0.06
0.09
13.54
16.20
3.83
3.85
4.07
3.84
2.77
2.54
31
0.09
0.22
0.04
0.21
11.52
17.16
3.86
3.65
3.76
3.95
2.45
2.71
32
0.08
0.32
0.05
0.30
13.54
21.62
3.88
4.79
4.37
4.83
3.05
3.20
33
0.10
0.33
0.05
0.25
13.84
18.22
3.85
4.45
4.45
4.55
3.14
3.04
34
0.09
0.14
0.04
0.12
14.03
16.45
3.50
3.57
4.00
3.86
2.81
2.65
35
0.17
0.33
0.11
0.32
13.76
18.34
3.42
3.83
3.90
4.01
2.74
2.71
36
0.25
0.41
0.14
0.39
14.63
20.68
3.56
3.98
3.92
4.08
2.71
2.73
37
0.25
0.42
0.15
0.36
15.13
19.84
3.60
4.10
3.86
4.15
2.64
2.76
39
0.22
0.29
0.13
0.24
15.09
18.51
3.62
3.59
4.21
3.98
2.98
2.77
40
0.33
0.31
0.22
0.24
20.89
20.80
3.58
3.82
4.16
4.03
2.94
2.73
41
0.39
0.33
0.22
0.23
19.67
19.19
3.43
3.44
4.04
3.76
2.88
2.59
42
0.13
0.12
0.07
0.08
13.24
14.12
3.26
2.68
3.72
3.43
2.61
2.52
43
0.04
0.00
0.06
0.00
14.17
26.20
3.41
4.86
4.58
4.90
3.42
3.25
44
0.08
0.36
0.00
0.41
23.55
178.20
5.45
5.48
4.36
4.62
2.51
2.76
45
0.06
0.00
0.01
0.08
22.40
29.11
5.28
6.94
4.16
4.70
2.37
2.35
0.03
0.25
0.00
0.36
16.59
15.06
Source: China Industrial Firms Database, authors’ calculation.
4.99
4.90
3.46
3.54
1.76
1.87
46
11
Up to now, it seems that exporters are not necessarily more productive than non-exporters, then
how exactly does productivity affect a firm’s export behavior? Would the above empirical results
be in accordance with the predictions by firm heterogeneity model if we control the impacts of
factor endowment and processing trade?
4. Empirical Analysis
4.1 Model Specifications
With regards to the studies on correlation between productivity and export behavior, Melitz’s
model is the most influential one. On the basis of Krugman’s model (1980) based on assumptions
of monopolistic competition and scale economies, Melitz incorporates heterogeneous firms. His
model fits quite well the linkage between productivity and export behavior that productivity is a
key determinant of exporting, namely, a more productive firm is more likely to export.
Conditional on a closed economy, a firm’s profits can be written as follows:
   1 r  * 
 * 
f
if   *
      


0
if   *

(1)
Where  is product substitution elasticity, and r represents revenues, and f refers to fixed costs,
and
 * denotes the critical point of production. If and only if    * , a firm will decide to
produce:
   

r   

 f 0
    1 
We can obtain:     max 0,  *   1
   


f

Obviously, when    , profits are positively related to productivity. Conditional on an open
*
economy, profits from exporting is:
 1
 x   
rx  

 1 rd  
 fx 
 fx 


1
 
 * 
 

rd  * 
 fx 4
(2)
Where subscriptions d and x denote domestic factors and exporting, respectively, and
 refers to
is iceberg trade costs, and f x is regular fixed trade costs5. Profits from exporting can be written as

4
For each period of time, a firm pays
f x ,namely,
 1   
t 1
12
t
f x  f ex ,therefore f x   f ex .
follows:
 x    
 1
1
 
 *
 
f  fx
(3)
It is apparent that profits from exporting are positively associated with productivity. Let  x* be
the critical point of production, then only firms with productivity above  x* will decide to export:
prob(ex  1)  prob(ex  0)  f ( , control )
(4)
Referring to the existing literature and the features of China’s exports, our baseline empirical
model is specified as follows:
EX ijt    1 l nLijt   2 l nkijt   DRD
3
ijt   DNPD
4
ijt   TFP
5 ijt
 6 Re gionijt   8Scaleijt   Typle
9
ijt   j   t   ijt
(5)
To examine how scale and ownership (firm type) will affect the influence of productivity on
exporting, we design two interaction terms of productivity by scale or productivity by ownership
(firm type) and put these two interaction terms into equation (6):
EX ijt    1 ln Lijt   2 ln kijt  3 DRDijt   4 DNPDijt  5TFPijt
 6 Re gionijt  8 Scaleijt * TFPijt  9Typeijt *TFPijt   j   t   ijt
(6)
Where EX ijt is the variable reflecting a firm’s export behavior, and i、j、t refer to firm i, industry j
and year t, and  j 、  t
 ijt
capture unobserved effects related to a specific industry i or year t, and
denotes the error term.
4.2 Variables Specifications
Annual average of employed persons and annual average of net value of fixed assets per capita
measure labor input (L) and capital-labor ratio (k). Calculation of TFP was given previously. R&D
input (DRD) and new product development (DNPD) here are two dummies that are set to be 1 if a
firm has R&D input and a new product, or else DRD and DRD will be 0. The location of a firm
(Region) is also measured by a dummy. If a firm is located in eastern region in China, let Region
be 1, or else it will be 0. By size (Scale), firms here are classified into three categories: large-sized
firms, medium-sized firms and small-sized firms. We use a mix of two dummies (Dscale1 and
Dscale2) to represent different types of firm. Specifically, let both Dscale1 and Dscale2 be 0 if a
firm is small in size, Dscale1 be 0 and Dscale2 be 1 if medium in size and Dscale1 be 1 and
Dscale2 be 0 if large in size. Similarly, we divide firms by ownership into three types: domestic
firms, foreign-funded firms and firms with funds from Hong Kong, Macao and Taiwan. A mix of
two dummies (Dtype1and Dtype2) is set to reflect these three types by ownership. Both Dtype1
and Dtype2 are 0 for a domestic firm, and with Dtype1 to be 1 and Dtype2 to be 0 for a firm with
funds from Hong Kong, Macao and Taiwan, and with Dtype1 to be 0 and Dtype2 to be 1 for a
13
foreign funded firm. This paper adopts data from 4 years, namely 2001, 2003, 2005 and 2007.
4.3 Empirical Results
One distinctive feature of data in this paper is that the explained variable is Censored Data which
sometimes could be 0. Under such circumstances, two econometric methods are useful. One
method is Probit model which in fact is a binary discrete model where we let the non-zero
explained variable be 1. Probit model is designed to test explanatory variables’ impacts on the
explained variable’s probability of being 1. Alternatively, we can turn to Tobit model which allows
for the direct investigation of explanatory variables’ impacts on the explained variable’s quantities.
To get more robust estimates, this paper adopts both of these two econometric methods.
Empirical results are summarized in Table 5. In all the regressions, industry fixed effects and year
fixed effects are all controlled. In the process of including dummies of Region, Scale and Type,
the signs of all the estimated coefficients of explanatory variables as well as their statistical
significance keep invariant, except that, however, the estimated coefficient of TFP in regression (7)
is no longer statistically significant. All the other estimated coefficients of explanatory variables
are statistically significant at the 1% level. Now, we sharpen our focus to the interpretations on
estimates of regressions (7) and (8).
The estimated coefficients of DRD and DNPD are both statistically significant and positive. As
proxies which stand for a firm’s innovative capability, a more innovative firm is more capable of
satisfying the diverse needs of consumers, so it’s more adaptable to international market and is
more likely to export and achieve more export value.
After controlling the impacts of firm type, however, ―productivity paradox‖(Li Chunding,
2009,2010) appears. It’s found that productivity growth is not positively associated with exporting,
which seems to contradict the widely accepted view that a higher productivity leads to a decision
to export and export more. There are several possible reasons for so called ―productivity paradox‖.
Firstly, for exporters in developed countries, they usually don’t enter international market until
they have gained a large portion of domestic market, hence they are generally larger in size and
more competitive and productive than non-exporters. While in China, more than half aggregate
trade is processing trade. For exporters involved in processing trade, they rely on labor abundance
to integrate themselves into operations of processing and assembly. It is labor abundance rather
than innovativeness or productivity that these exporters rely on to export. Secondly, as predicted
by firm heterogeneity model, a firm must be productive enough to bear sunk costs incurred in
exporting, so an exporter is generally more productive than a non-exporter. But in China, since
domestic market is severely separated, which results in the consequence that a firm serving
domestic market is sometimes at the expense of rather high rent-seeking costs. Therefore, it truly
14
happens that a non-exporter is more productive than an exporter.
Table 5
lnL
Lnk
DRD
DNPD
TFP
Empirical Analysis of Determinants of Firms’ Export Behavior(Ⅰ)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Probit
Tobit
Probit
Tobit
Probit
Tobit
Probit
Tobit
0.7608
2.3073
0.7775
2.375
0.6297
2.0112
0.5522
1.8648
(70.56)***
(85.49) ***
(73.85) ***
(89) ***
(50.18) ***
(65.21) ***
(45.44) ***
(61.64) ***
0.1665
0.5636
0.1516
0.5447
0.1090
0.4298
0.0244
0.2622
(21.28) ***
(28.79) ***
(19.72) ***
(28.05) ***
(13.72) ***
(21.47) ***
(3.17) ***
(13.3) ***
0.0963
0.2412
0.1012
0.2608
0.0851
0.2239
0.1326
0.3351
(4.96) ***
(5.35) ***
(5.26) ***
(5.8) ***
(4.41) ***
(4.98) ***
(7.01) ***
(7.51) ***
1.3035
2.9611
1.3390
3.0537
1.3242
3.0166
1.3583
3.1673
(55.05) ***
(57.63) ***
(57.17) ***
(59.46) ***
(56.46) ***
(58.79) ***
(59.32) ***
(62.2) ***
0.0763
0.4880
0.0588
0.4571
0.0381
0.4069
-0.0021
0.3305
(8.9) ***
(23.25) ***
(6.92) ***
(21.85) ***
(4.46) ***
(19.36) ***
(-0.24)
(40.15) ***
1.9197
5.9654
1.9545
6.0497
1.3044
4.1759
(55.58) ***
(54.98) ***
(55.4) ***
(55.43) ***
(39.26) ***
(15.89) ***
1.4675
3.9581
1.2855
3.6775
(14.88) ***
(14.65) ***
(13.79) ***
(14.75) ***
0.8502
2.4501
0.5673
1.7203
(21.31) ***
(22.07) ***
(15.7) ***
(16.71) ***
Region
Dscale1
Dscale2
Dtype1
Dtype2
2.8478
7.8765
(70.5) ***
(70.6) ***
2.5974
7.2520
(65.11) ***
(65.18) ***
Industry
YES
YES
YES
YES
YES
YES
YES
YES
Year
YES
YES
YES
YES
YES
YES
YES
YES
observations
244172
244172
244172
244172
244172
244172
244172
244172
left-censored
observations
Rho
157912
0.90503
0.85152
157912
0.89163
0.83957
157912
0.89516
0.84093
157912
0.86858
Note: Values of z statistics in parentheses, ***,** and * donate significance at the 1%,5% and 10% level
respectively.
To examine the influence of productivity on a firm’s export behavior, we further extract firms
whose exports are zero over 2001-2006 but positive in 2007, and then assign them into their
corresponding sectors. If productivity serves as a key determinant of a firm’s export behavior, it is
supposed for each sector consisting of selected firms, its productivity in 2007 should be the
maximum over 2001-2007. As shown in Table 6, Y refers to the possibility that the sector
productivity in 2007 is the maximum over 2001-2007, and N denotes that sector productivity in
2007 is not the highest over 2001-2007. It’s found that for 22 out of 33 sectors, sector productivity
in 2007 is not the maximum over 2001-2007, which also lent evidence to our finding that
15
0.81220
productivity may be not a main determinant of exporting in China.
Table 6
Productivity by Sector in the Sample of Firms Commencing Exporting since 2007
sector code
sector code
sector code
6
N
22
N
33
N
7
N
23
N
34
Y
10
N
24
N
35
Y
13
Y
25
N
36
Y
14
N
26
N
37
N
15
Y
27
N
39
Y
17
N
28
N
40
N
18
Y
29
Y
41
Y
19
N
30
Y
42
N
20
N
31
N
44
N
N
46
Y
21
N
32
Source: China Industrial Firms Database, authors’ calculation.
Next, we see how estimates vary when the dummy of firm type is included. Regressions (1)-(6)
suggest the estimated coefficients of TFP are all positive and statistically significant when factors
of Region and Scale are gradually included. But after the factor of firm type is controlled, they
turn negative and statistically insignificant in Pobit model. The variance of the estimated
coefficient of TFP indicates that the positive impact of productivity on exporting revealed in
regressions (1)-(6) should actually be attributed to the role played by firm type. The fact is that
most of foreign funded firms and firms with funds from Hong Kong, Macao and Taiwan in China
engage in processing trade, which means they are more likely to export than domestic firms.
Simultaneously, they are obviously more productive than domestic ones. Therefore, there is no
wonder that the presence of foreign funded firms and firms with funds from Hong Kong, Macao,
Taiwan generates a positive linkage between productivity and exporting. In Table 7, F-TFP and
D-TFP represent average productivity of foreign funded firms (including firms with funds from
Hong Kong, Macao and Taiwan included) and that of domestic firms. In the column named
―differentiation‖, Y means foreign funded firms are more productive than domestic firms, and if
the opposite case holds, the difference led by domestic firms is given by sector. It’s seen that out
of 39 sectors, there are 29 sectors where foreign funded firms are more productive than domestic
firms, and only a small differentiation is observed in the rest 10 sectors where domestic firms are
more productive than foreign funded firms.
16
Table 7
TFP differentiation between Domestic Firms and Foreign Funded Firms
(Including Foreign Firms with Funds from Hong Kong, Macao and Taiwan)
sector
sector
code
F-TFP
D-TFP
differentiation
code
F-TFP
D-TFP
differentiation
6
2.647
2.413
Y
27
3.356
2.930
Y
7
10.222
3.841
Y
28
3.121
2.944
Y
8
3.825
3.212
Y
29
2.728
2.690
Y
9
3.488
3.036
Y
30
2.723
2.840
4.11
10
2.951
2.735
Y
31
2.854
2.579
Y
11
2.892
2.793
Y
32
3.527
3.177
Y
13
3.250
3.161
Y
33
3.265
3.232
Y
14
3.023
2.790
Y
34
2.852
2.863
0.40
15
3.303
2.834
Y
35
3.125
2.778
Y
16
3.403
3.824
11.01
36
3.127
2.760
Y
17
2.674
2.665
Y
37
3.134
2.698
Y
18
2.587
2.585
Y
39
2.962
3.028
2.18
19
2.533
2.825
10.33
40
2.912
2.915
0.09
20
2.784
2.812
1.00
41
2.875
2.798
Y
21
2.703
2.726
0.84
42
2.542
2.702
5.92
22
2.882
2.745
Y
43
3.631
3.537
Y
23
2.807
2.565
Y
44
3.619
2.601
Y
24
2.457
2.632
6.63
45
2.814
2.420
Y
25
4.224
3.328
Y
46
2.523
1.918
Y
26
3.487
3.045
Y
Source: China Industrial Firms Database, authors’ calculation.
It is worth noting that, however, as shown in Tobit model (regression 8) both Type and
productivity can promote a firm’s export value, namely, firms who have already entered into
export market, their exports will increase with the growth of their productivity.
Dummies of Region and Scale are both positively related to exporting, indicating that a firm
located in eastern region is more likely to export and to export more. Similarly, a large-sized firm
or a medium-sized firm both have greater tendency to export as well as higher exporting
capabilities. Table 6 describes the empirical results with two interaction terms included. Emphasis
of interpretations now turns to the effects of two interaction terms.
Columns (11) and (12) in Table 8 show that either in Probit model or in Tobit model, the estimated
coefficients of TFP are statistically insignificant (-0.0055 and 0.2570, respectively).Given that the
estimated coefficients are positive and statistically significant (0.3283 and 0.2048, respectively),
we can tell that TFP growth will promote a firm to export and increase its export value. The
estimated coefficients of the interaction term between TFP and Dscale are positive and statistically
significant, suggesting that productivity’s positive effect on a firm’s export behavior is more
17
significant for a firm large in size than for a small-sized firm. Columns (13) and (14) show the
results with interaction term between TFP and firm type included. The positive and statistically
significant estimated coefficients of interaction term suggest that for a foreign-funded firm,
productivity’s positive impact on a firm’s export behavior is more significant.
Table 8
lnL
lnk
DRD
DNPD
TFP
Region
Empirical Analysis of Determinants of Firms’ Export Behavior (Ⅱ)
(9)
(10)
(11)
(12)
(13)
(14)
Probit
Tobit
Probit
Tobit
Probit
Tobit
0.7608
2.3073
0.6701
2.1279
0.5859
1.9963
(70.56)***
(85.49) ***
(55.49) ***
(72.17) ***
(50.63) ***
(68.59) ***
0.1516
0.5447
0.1197
0.4640
0.0458
0.3404
(19.72) ***
(28.05) ***
(15.17) ***
(23.38) ***
(6.02) ***
(17.24) ***
0.1012
0.2608
0.0876
0.2313
0.1149
0.3043
(5.26) ***
(5.8) ***
(4.55) ***
(5.14) ***
(6.14) ***
(6.77) ***
1.3390
3.0537
1.3271
3.0208
1.3379
3.1522
(57.17) ***
(59.46) ***
(56.62) ***
(58.84) ***
(59.09) ***
(61.37) ***
0.0588
0.4571
-0.0055
0.2570
-0.2482
-0.5419
(6.92) ***
(21.85) ***
(-0.6)
(11.09) ***
(-26.31) ***
(-20.84) ***
1.9197
5.9654
1.9312
5.9892
1.4671
4.9110
(55.58) ***
(54.98) ***
(55.35) ***
(55.13) ***
(45.03) ***
(47.38) ***
0.3283
0.8003
0.3020
0.8317
(11.7) ***
(11.67) ***
(11.43) ***
(12.68) ***
0.2048
0.5523
0.1420
0.4279
(17.74) ***
(18.43) ***
(13.33) ***
(14.83) ***
0.6555
1.6758
(59.84) ***
(56.06) ***
0.5973
1.5340
(51.79) ***
(48.93) ***
Dscale1*TFP
Dscale2*TFP
Dtype1*TFP
Dtype2*TFP
industry
YES
YES
YES
YES
YES
YES
year
YES
YES
YES
YES
YES
YES
observations
244172
244172
244172
244172
244172
244172
left-censored
observations
rho
157912
0.89163
0.83958
Note:Values of z statistics in parentheses,
respectively.
157912
0.89355
0.83970
157912
0.86178
0.81095
***,** and * donate significance at the 1%,5% and 10% level
18
5. The Impact of Processing Trade on Correlation between
Productivity and Export Behavior
With the impact of Dtype controlled, we find that productivity growth does not increase a firm’s
probability of exporting. Such ―productivity paradox‖ is due to the presence of processing trade,
and foreign funded firms (including firms with funds from Hong Kong, Macao and Taiwan) who
are mostly engage in processing trade as well. To further investigate processing trade’ distortion
on correlation between productivity and export behavior, this paper attempts to separate the
influence of processing trade from our overall samples.
Since information on processing trade is not given by China Industrial Firms Database, and it’s
hard to tell whether exports delivered belong to processing or not, we adopt the following three
approaches to exclude the presence of processing trade.
Given the fact that foreign funded firms actually contribute a lot to processing trade in China, the
first approach is to divide firms into foreign firms (including firms with funds from Hong kong,
Macao and Taiwan included) and domestic firms, and perform respective regressions on them. By
doing so, the impact of processing trade will be removed in regression on domestic firms(refer to
the column (1) in Table 9.
The second approach is to classify sectors by the share of processing exports, and then run
regression only on firms from sectors with lower share of processing exports. Foreign Trade
Database of Development Research Center of the State Council of China provides data on
processing trade and ordinary trade at the 4-digit Harmonized System (HS) level. We assign these
disaggregated data into manufacturing sectors according to the correspondence between HS4 and
GB (T/4757-2002) released by National Bureau of China Statistics. Using data in 2007 only, we
find that for sectors of Manufacture of Non-metallic Mineral Products (31), Manufacture of
Electrical Machinery and Equipment (39), Processing of Petroleum, Coking, Processing of
Nuclear Fuel (25), Manufacture of Paper and Paper Products (22), Manufacture of Articles For
Culture, Education and Sport Activities (24), Manufacture of Measuring Instruments and
Machinery for Cultural Activity and Office Work (41) and Manufacture of Communication
Equipment, Computers and Other Electronic Equipment (40), processing exports account for over
half of the aggregate exports. Therefore, we put the rest of sectors in the group characterized by
lower share of processing exports, with regression results showed in column (2) in Table 9.
19
Table 9
Empirical Analysis of Determinants of Firms’ Export Behavior (Ⅲ)
(1)
foreign funded
firms
(2)
domestic
firms
firms from
Firms from
firms from
Firms from
industries with
industries with
industries with
industries with
lower share of
higher share of
lower share of
higher share of
processing exports
processing
processing exports
processing
Exports
lnL
0.4715
(22.54)***
0.0181
(1.38) *
0.1023
(2.9) ***
0.7019
(15.37) ***
-0.0733
(-5.33) ***
2.2398
(25.51) ***
0.3536
(2.14) **
0.4099
(6.94) ***
0. 5901
(39.54) ***
0.0276
(2.88) ***
0.1424
(6.35) ***
1.5464
(58.37) ***
0.0464
(4.35) ***
1.1381
(31.76) ***
1.4713
(13.48) ***
0.6075
(13.55) ***
0.6163
(43.95)***
0.1119
(12.64) ***
0.1147
(5.26) ***
1.436
(53.98) ***
0.0652
(6.82) ***
1.8851
(49.83) ***
1.3370
(11.86) ***
0.76763
(17.13) ***
0.6960
(24.63) ***
0.1038
(5.76) ***
-0.0106
(-0.26)
0.9192
(18.26) ***
-0.0706
(-3.62) ***
2.3288
(23.78) ***
1.9054
(9.06) ***
1.1572
(13.06) ***
Industry
Year
observations
YES
YES
65432
YES
YES
178740
YES
YES
199056
rho
0.8556
0.8710
0.8955
lnk
DRD
DNPD
TFP
Region
Dscale1
Dscale2
YES
YES
45116
0. 5474
(40.27) ***
0.0256
(2.98) ***
0.1497
(7) ***
1.4520
(55.98) ***
0.0244
(2.61) ***
1.2828
(35.59) ***
1.2682
(12.09) ***
0.5322
(13.06) ***
2.8156
(61.57) ***
2.4862
(54.47) ***
YES
YES
199056
0. 5706
(21.12) ***
0.0281
(1.63) *
0.08187
(2.03) **
1.0078
(20.65) ***
-0.1089
(-5.74) ***
1.4022
(16.3) ***
1.2355
(6.09) ***
0.6389
(8.3) ***
2.8744
(33.74) ***
2.8359
(34.8) ***
YES
YES
45116
0.8926
0.8711
0.8509
Dtype1
Dtype2
Note:Values of z statistics in parentheses,
respectively.
Exports
***,** and * donate significance at the 1%,5% and 10% level
These two approaches are able to exclude the impact of processing trade to a large extent, but not
complete enough, because a proportion of domestic firms and firms from sectors with lower
processing exports share also deal with processing trade. Using the third approach below, we
consider the ratio of exports delivered to gross industrial output value (defined as &) as a proxy
for the presence of processing trade for each firm. Generally speaking, & will be greater if the
firms tend to export, and thus we can exclude firms with high percentage of processing trade to a
gradually larger extent, from 1 to 0.5, with & decreasing, more firms will be removed from our
sample. Although this approach excludes some ordinary export firms with high percentage of
exports delivered in total output, however, it mainly excludes firms dealing with processing trade
and therefore lowers the impact of processing trade effectively (see Table 10 for empirical results).
20
Table 10
Empirical Analysis of Determinants of Firms’ Export Behavior (Ⅳ)
&<1
&<=0.9
&<=0.8
&<=0.7
&<=0.6
&<=0.5
0.5796
0.5819
0.5774
0.5742
0.5712
0.5597
(46.28)***
(45.19) ***
(44.20) ***
(43.37) ***
(42.37) ***
(41.09) ***
0.0434
0.0727
0.0912
0.1058
0.1157
0.1287
(5.46) ***
(8.89) ***
(10.97) ***
(12.54) ***
(13.5) ***
(14.83) ***
0.1411
0.1534
0.1613
0.1649
0.1731
0.1828
(7.39) ***
(7.92) ***
(8.25) ***
(8.36) ***
(8.69) ***
(9.08) ***
1.3638
1.3818
1.3890
1.4054
1.4167
1.4289
(59.42) ***
(59.82) ***
(59.96) ***
(60.32) ***
(60.46) ***
(60.64) ***
0.0203
0.0328
0.0458
0.0563
0.0628
0.0674
(2.33) ***
(3.67) ***
(5.04) ***
(6.11) ***
(6.72) ***
(7.1) ***
1.2330
1.1108
1.0190
0.9480
0.8664
0.7889
(37.55) ***
(34.4) ***
(31.95) ***
(29.94) ***
(27.48) ***
(25.42) ***
1.1541
1.0568
0.9988
0.9507
0.8893
0.8700
(12.58) ***
(11.72) ***
(11.18) ***
(10.76) ***
(10.13) ***
(10.03) ***
0.5459
0.5268
0.5114
0.4862
0.4631
0.4502
(15.1) ***
(14.6) ***
(14.21) ***
(13.58) ***
(12.95) ***
(12.66) ***
2.6542
2.3969
2.2199
2.0599
1.9215
1.8012
(65.59) ***
(59.2) ***
(54.99) ***
(51.32) ***
(48.01) ***
(45.35) ***
2.3167
1.9974
1.8066
1.6575
1.5142
1.4009
(57.53) ***
(49.45) ***
(44.79) ***
(41.28) ***
(37.75) ***
(35.12) ***
industry
YES
YES
YES
YES
YES
YES
year
YES
YES
YES
YES
YES
YES
observations
229433
214842
207840
202666
198268
194056
rho
0.8609
0.8487
0.8396
0.8303
0.8211
0.8106
lnL
lnk
DRD
DNPD
TFP
Region
Dscale1
Dscale2
Dtype1
Dtype2
Note:Values of z statistics in parentheses,
respectively.
***,** and * donate significance at the 1%,5% and 10% level
As we can see in Table 9, in regressions on domestic firms and firms from sectors with a lower
share of processing exports, TFP is positively associated with a firm’s probability to export, while
TFP growth is not in favor of a firm’s exporting, and sometimes it even prevents a firm from
21
exporting for foreign funded firms and firms from sectors with a higher share of processing
exports. The estimated coefficients of TFP in Table 10 are all positive and statistically significant.
Moreover, with & increasing, both the estimated coefficients and Z values increase as well,
indicating that with the presence of processing trade removed more completely, the positive
impact of TFP on a firm’s exporting is more significant.
Empirical results listed in Table 9 and 10 suggest that it is the presence of processing trade that
makes firm heterogeneity model not applicable to firms in China. We have proven that, however,
with the presence of processing trade removed, firm heterogeneity model still holds, namely,
productivity growth is positively linked with a firm’s decision to export.
6. Conclusions
In the framework of firm heterogeneity model, working on data from China Industrial Firms
Database over 2001-2007 (data in 2004 not available), this paper observes characteristics of
exporters in China, and presents the differences between exporters and non-exporters, finding that
out of 39 sectors, there are 16 sectors where labor productivity of exporters is below that of
non-exporters, and 22 sectors (covering the 16 ones above) where TFP of exporters is less than
that non-exporters.
Using Probit model and Tobit model, we have empirically study the determinants of a firm’s
probability of exporting and its export capabilities. The empirical results show that both labor
input and capital input exert positive effects on a firm’s exporting, with the impact of labor more
significant, suggesting that further restructuring of industry and upgrading of factor inputs are
essential.
Both R&D inputs and new product development are positively correlated with a firm’s export
behavior, implying that capability of innovation still serves as a key determinant of a firm’s
exporting, so it is curial to increases R&D inputs and to enhance capability of innovation in the
future. Besides, a firm in eastern region is more likely to export than a firm in central or western
region. Therefore, we should improve infrastructures in central and western regions, and make
firms benefit more from exporting.
With the impact of Type controlled, estimates in Probit model show no evidence of positive
linkage between productivity and a firm’s probability of exporting. This is mainly due to the
presence of processing trade. Regression results with the absence of processing trade, however,
show that productivity is truly positively related to a firm’s probability of exporting. In addition,
results by Tobit model find that a firm’s exports increase with its productivity going up.
22
With regard to a firm’s size (Dscale) and ownership type (Dtype), it’s found that firms large in size
and foreign funded firms (including foreign firms with funds from Hong Kong, Macao and Taiwan)
are more likely to export than firms small in size and domestic firms. Moreover, the positive
impacts of productivity on a firm’s exporting are more significant for firms large in size and
foreign funded firms (including foreign firms with funds from Hong Kong, Macao and Taiwan).
References
[1] Zhang Liqing, Sun Junxin. Does Export Improve Productivity? Firm-level Evidence from
China s Manufacturing Sector [J], Nankai Economic Studies, 2010(4):110-122
[2] Li Chunding, Shi Xiaojun and Xing Chunbing. ―Exporting-Productivity Paradox‖: A Further
Investigation into China [J], Economic Perspectives, 2010(8):90-95
[3] Bao Qun, Xu Helian and Lai Minyong, How do Exports Promote Economic Growth? [J]
Shanghai Economic Review, 2003(3):3-10
[4] Peng Guohua, Empirical Research on Productivity in Chinese Provinces Using Bilateral
Gravity Model of Trade [J], Economic Research Journal, 2007(8):123-132
[5] Li Xiaoping, Lu Xianxiang and Zhu Zhongli, International Trade,Technological Progress and
Productivity Growth of Chinese Industries [J], China Economic Quarterly, 2008(2):549-564
[6] Xue Mantian, Zhao Shudong, Imports’ Impact on Productivity in China [J], Modern
Management Science, 2009(1):92-94
[7] Xi Guo, Chen Gang. Foreign Direct Investment, International Trade and Productivity Growth
of China: An Empirical Study Based on Nonparametric Malmquist Index [J], Journal of
International Trade, 2008(6 ):89-97
[8] Zhangjie, Liyong and Liu Zhibiao. Exporting and Productivity of Domestic Firms: An
Empirical Study Based on Manufacturing Firms in Jiangsu Province [J], Management World, 2008
(11):50-64
[9] Li Chunding, Yin Xiangshuo. Chinese Export Firms ―Productivity Paradox‖ and its
Explanations [J], Finance & Trade Economics, 2009(11):84-90
[10] Liu Yaojia. International Trade, FDI and Total Factor Productivity Weakening: DEA and
Cointegration based on Panel Data from 1952~2006 [J] The Journal of Quantitative & Technical
Economics, 2008(11):28-39
23
[11]Yi Jingtao, Firm Heterogeneity, Sunk Costs, Spillovers and Export Participation [J], Economic
Research Journal, 2009(9):106-115
[12] Tang Yihong, Lin Faqin. An Empirical Test: The Relevancy of the Firm-Heterogeneity-Model
to Firms Exports in China [J], Nankai Economic Studies, 2009(6):88-99
[13] Ye Zhen. Why do Chinese Exporting Enterprises Have Higher Productivity?——Evidence
from Jiangsu Province [J], Finance & Trade Economics, 2010(5):77-81
[14] Li Chunding. Does ―Productivity Paradox‖ Exist in China’s Exporting Firms: A Test Based
on Manufacturing Firms in China [J], The Journal of World Economy, 2010(7):64-81
[15] Aw, B. Chung, S. ,Roberts, M. 2000 Productivity and Turnover Patterns in the Export
Market:Firm Level Evidence from Taiwan and South Korea [J]. World Bank Economic Review,
2000(14) : 65-90
[16] Arnold, J. M. and Hussinger, K. Export Behavior and Firm Productivity in German
Manufacturing: A Firm-Level Analysis [J]. Review of Economics, 2005(2): 119- 243.
[17] Bernard, A. B. and Jensen, J. B. Exceptional Exporter Performance: Cause Effect, or Both?
[J]. Journal of International Economics, 1999(1): 1-25
[18] Bernard, A., Jensen, J. B., Lawrence,R. Z. Exporters,Jobs and Wages in U.S. Manufacturing:
1976-1987 [J]. Brookings Papers on Economic Activity, Microeconomics, 1995: 67-119
[19] Bernard, A. B., J. Wagner. Export Entry and Exit by German Firms [J]. Review of World
Economics, 2001(137): 105-123
[20] Bernard, A., Eaton, J., Jensen, J. B. et al. Plants and Productivity in International Trade [J].
American Economic Review, 2003(4): 1268-1290
[21] Bernard, A. B. and Jensen, J. B., Why some Firms Export? [J]. The Review of Economics and
Statistics, 2004(2): 561-569
[22] Bernard, A. B., Jensen, J. B. and Schott, P. K., Trade Costs, Firms and Productivity [J].
Journal of Monetary Economics, 2006(53): 917-937
[23] Bigsten A,Collier P,Dercon S,et al. Do African Manufacturing Firms Learn from Exporting?
[J], Journal of Development Studies, 2004 (3): 115-141
[24] Bladwin, J., Gu, W. Trade Liberalization: Export-Market Participation,Productivity Growth
and Innovation [J], Oxford Review of Economic Policy, 2004 (20): 372-92
[25] Castellani , D. Export Behavior and Productivity Growth: Evidence from Italian
24
Manufacturing Firm [J], Review of World Economics, 2002(4): 605-628
[26] Clerides, S., Lauch, S., Tybout, J. Is Learning by Exporting Important? Micro-Dynamic
Evidence from Colombia, Mexico, and Morocco [J]. Quarterly Journal of Economics, 1998 (3):
903-947
[27] Crespi, G., Criscuolo,C., J. Haskel. Productivity, Exporting, and the Learning-by-Exporting
Hypothesis: Direct Evidence from UK Firms [J], Canadian Journal of Economics, 2008 (2):
619-638
[28] Das, S.,M. J.Roberts, J. R. Tybout .Market Entry Costs, Producer Heterogeneity and Export
Dynamics[J], Econometrica, 2007 (3): 837-872
[29] Delgado, M., Farinas, J., Ruano, S. Firm Productivity and Export Markets:A Non-Parametric
Approach [J], Journal of International Economics, 2002 (57): 397-422
[30] Feenstra, R., H.L. Kee, Export Variety and Country Productivity: Estimating the
Monopolistic Competition Model with Endogenous Productivity [J], Journal of International
Economics, 2008(2): 500-518
[31] Fu X. Exports,Technical Progress and Productivity Growth in a Transition Economy: a NonParametric Approach for China [J], Applied Economics, 2005(7): 725-739
[32] Girma, S., Greenaway, D., Kneller, R. Does Exporting Increase Productivity? A
Micro-econometric Analysis of Matched Firms [J], Review of International Economics, 2004 (5):
855-866
[33] Helpman, Elhanan, Marc J. Melitz and Stephen R. Yeaple, Export Versus FDI with
Heterogeneous Firms [J], American Economic Review, 2004(94): 300- 316
[34] Isgut , A. E. What's Different About Exporters? Evidence from Colombian Manufacturing [J],
The Journal of Development Studies, 2001(5): 57-82
[35] Krugman, P. R. Scale Economies, Product Differentiation and the Pattern of Trade [J],
American Economic Review, 1980 (70): 950-959
[36] Loecker, J. D., Do Exports Generate Higher Productivity? Evidence from Slovenia [J],
Journal of International Economics, 2007(73): 69- 98
[37] Lileeva, A., D. Trefler. Improved Access to Foreign Markets Raises Plant-Level Productivity
for Some Plants [J], Quarterly Journal of Economics, 2010(3): 1051-1099
[38]Melitz, M. The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
25
Productivity [J], Econometrica, 2003 (6): 1695-1725
[39] Park A., Yang D., Shi X., et al. Exporting and Firm Performance: Chinese Exporters and the
Asian Financial Crisis [J], Review of Economics and Statistics, 2010(4): 822-842
[40] Roberts, M. J., Tybout J. R.. The Decision to Export in Colombia: An Empirical Model of
Entry with Sunk Costs [J], American Economic Review, 1997(87): 545–564
[41] Van Biesebroeck, J. Export Raises Productivity in Sub-Saharan African Manufacturing Firms
[J], Journal of International Economics, 2005 (2): 373-391
Appendix
Codes by industrial sector:
(6)Mining and Washing of Coal
(7)Extraction of Petroleum and Natural Gas
(8)Mining and Processing of Ferrous Metal Ores
(9)Mining and Processing of Non-Ferrous Metal Ores
(10)Mining and Processing of Nonmetal Ores
(11)Mining of Other Ores
(12)Processing of Food from Agricultural Products
(13)Manufacture of Foods
(14)Manufacture of Beverages
(15)Manufacture of Tobacco
(16)Manufacture of Textile
(17)Manufacture of Textile Wearing Apparel, Footwear and Caps
(18)Manufacture of Leather, Fur, Feather and Related Products
(19)Processing of Timber, Manufacture of Wood,
(20)Bamboo, Rattan, Palm and Straw Products
(21)Manufacture of Furniture
(22)Manufacture of Paper and Paper Products
(23)Printing, Reproduction of Recording Media
(24)Manufacture of Articles For Culture,
Education and Sport Activities
(25)Processing of Petroleum, Coking, Processing of Nuclear Fuel
(26)Manufacture of Raw Chemical Materials and Chemical Products
(27)Manufacture of Medicines
26
(28)Manufacture of Chemical Fibers
(29)Manufacture of Rubber
(30)Manufacture of Plastics
(31)Manufacture of Non-metallic Mineral Products
(32)Smelting and Pressing of Ferrous Metals
(33)Smelting and Pressing of Non-ferrous Metals
(34)Manufacture of Metal Products
(35)Manufacture of General Purpose Machinery
(36)Manufacture of Special Purpose Machinery
(37)Manufacture of Transport Equipment
(39)Manufacture of Electrical Machinery
and Equipment
(40)Manufacture of Communication Equipment,
Computers and Other Electronic Equipment
(41)Manufacture of Measuring Instruments and Machinery for Cultural Activity and Office Work
(42)Manufacture of Artwork and Other Manufacturing
(43)Recycling and Disposal of Waste
(44)Production and Supply of Electric Power And Heat Power
(45)Production and Supply of Gas
(46)Production and Supply of Water
27