The Role of R&D on Firms’ Export Performance
A case of Thailand
2
Definitions
Thai firms - Firms where 100% of its capital is held by Thai people.
Non-Thai firms - Firms report any share of foreign capital.
Exporting firms (EF) - Firms report the share of exports of total sales is greater than zero (not sold 100% on the domestic market). Large-share exporting firms (LSE)- % export in total
sale is greater than or equal to 50% Small-share exporting firms (SSE)- % export in total
sale is larger than 0% but less than 50% Non-exporting firms (NE) - Firms that not export at all
(sold 100% on the domestic market)
3
Statement of the problem
Shifting in world demand from labor-intensive to technology-intensive products (Lall, 2000) pressures manufactured firms in Thailand improve technological capabilities.
One strategy implemented to develop technological capabilities is performing research and development (R&D).
R&D expenditures in the private sector in Thailand have been increasing.
Is R&D improve export performance of firms in Thailand?
4
Statement of the problem
1996 1997 1999 2001
Total R&D expenditures 5,528
4,811
5,021 11,064
Government sector 2,708
2,666
1,833 5,019
High education sector 1,219
1,631
2,570 1,950
Business enterprise sector 1,243
466
587 4,009
Private non-profit sector 357
46
30 85
million bahtR&D Expenditures Classified by Sector
Source: Office of the National Research Council of Thailand
5
Conceptual Discussion
1) R&D and export performance However, the effects of R&D on export performance of
firms in developing countries remain ambiguous. One the one hand, some previous studies (Posner,
1961; Hirsch, 1965; Vernon, 1966) proposed that R&D determines export performance of developed countries only.
One the other hand, there are studies argued that imported technology is often costly, risky and difficult to codified (Kumar and Siddharthan, 1994; Lall, 2000; Rodriquez and Rodriquez, 2005).
There is still no conclusion.
6
2)The nature of R&D in developing countries The role and organisation of R&D in Technology-
Followers (Forbes and Wield, 2000) R&D as a complement to shop-floor innovations. R&D as the formal learning unit of the firms R&D as a measure to build an independent
product design capability R&D as a source of intangible spin-off benefits for
the firms
Conceptual Discussion
7
2)The nature of R&D in developing countries (cont.)
Industrial Structure and the role of R&D (Kumar and Siddharthan,1994; Dijk, 2002) In developing countries, R&D determines the
export performance of firms in low-technology and medium-technology industries.
However, R&D does not affect export performance of firms in the high-technology industry.
Conceptual Discussion
8
3. The Multiplant Economy of Scale and R&D Behavior of MNEs (Markusen, 1984)
Economies of multiplant operation. Explains MNEs behavior in host countries. MNEs may be composed of many plants. One
plant (or more) carries out R&D and use its output as “joint inputs” of all plant productions.
Thus, the R&D behavior of MNEs are different from the R&D behavior of local firms and should be separately analyzed.
Conceptual Discussion
9
Remarks for the study of Thailand
The role of R&D in technology follower firms may differ from technology leader firms (Forbes and Wield, 2000) .
MNE firms may behave differently in R&D activity from Thai locally owned firms. (Markusen,1984)
Effects of R&D on export performance vary across industries. (Kumar and Siddharthan,1994)
10
Objective of the study
1) To estimate the effect of R&D intensity on export performance of firms in Thailand.
2) To investigate the differences in R&D behavior between exporting firms and non-exporting firms.
11
Method of the study
To answer the 1st objective
- using econometrics model to estimate effects of R&D intensity on firm’s export performance (estimated whole sample).
- examine the effects of R&D on export performance of Thai firms and non-Thai firms.
- examine the effects of R&D on export performance across groups of industries.
To answer the 2nd objective
- using ANOVA to investigate R&D behaviors among R&D firms.
12
indifferent different
Estimating whole sample
Estimating Thai-firms and non-Thai firms separately
Estimating whole sample classified by industries
Estimating Thai-firms classified by industries
Interpretations the econometric results
Comparative studies on R&D related factors and export
(compare R&D firms in selected sample)
Interpretations the ANOVA results
Test censored tobit model against Cragg’s two stages
specification and using LR-test to find appropriated model
Check R&D behaviors of MNEs in Thailand
Examine overall effect of R&D intensity on export
Using method of estimation suggested by LR-test
Check sector variations
Investigating R&D-related factors that differentiate
firm’s export
ObjectivesInstruments
Using ANOVA
Organization of the study
The 1st objective
Estimate the effect of R&D intensity on export performance of manufacturing firms in
Thailand
Theoretical Model
From
*fjq
* max ( , , )j
kj j j j j
qq Arg pq C x m z q
* *
*
* *
( , , , )
( , , , )
f f f kj j j
j d d d kj j j
q q p x m zq
q q p x m z
is the optimal level of the firm’s total output.
is the optimal level of output sold in the domestic market
is the output sold in the foreign market or the level of the firm’s export product.
is the cost variable that is common to production for both markets
is the specific cost of each market (k= d, f) .
is a vector of the specific factors of firm j, such as the productivity, size, and ownership.
*jq*djq
km
jz
x
15
Theoretical Model
Focusing on optimal level of firm’s export
From Crepon, Duguet, and Mairesse (1998), R&D can improve productivity, then reduce the exporting cost, and increase export level.
R&D can also ease firms to develop new products. Applying equation (4.13) in practice, firm’s export
performance is influenced by the exporting cost and firm’s characteristics
* *( , , , )f f f kj j jq q p x m z
16
Theoretical Model
Alternatively, we use knowledge from previous literatures as proxy variables.
R&D intensity- reduces costs of exported goods and extend product ranges.
Skilled labors – can be either complementary factors for R&D activities or raising production costs.
Firm’s size – Large firms tend to have cost advantages. Empirical evidences also found inverted U-shaped relationship between firm’s size and export
Ownership – Foreign and Joint venture firms have superior knowledge, export experience, and can easier access advanced technology. Hence, costs of export market penetration are lower than local firms.
The Model0 1 2 3 4 5
6
2EXP RDINT SKILL SIZE SIZE OWN
DUMi
Variable Discription MeasurementExpected
signDependent
Export performance EXPPorportion of firm's export in total sales
Proportion
Independent
R&D intensity RDINTR&D expenditures over total employees
Proportion +/-
Skilled labors SKILLThe proportion of technicians and graduate workers in total workers
Proportion +/-
Firm size SIZE Total employees Number +SIZE2 Square of total employees Number -
Ownership OWN Share of foreign capital 1-5 Likert scale +
Industrial structure DUMDummy variable of industries (classified by Pavitt taxonomy)
Binary
18
Data descriptions
Classification All firms R&D firms % RD firms
All 1695 262 15.46Exporting firms 840 176 20.95Non-exporting firms 855 86 10.06All 553 98 17.72Exporting firms 336 63 18.75Non-exporting firms 217 35 16.13All 1142 164 14.36Exporting firms 504 114 22.62Non-exporting firms 638 50 7.84
All firms
Non-Thai firms
Thai firms
Viriables Tobit Probit Truncated
R&D intensity 0.4445*** 1.9867*** 0.3019Skilled labors 0.0684 0.0330*** -0.0279Size 0.0039*** 0.02010*** 0.0102*Size2 -0.0001*** -0.0002*** -0.0001Ownership 0.0362*** 0.1200*** 0.0754Constant -0.1980*** 0.1217Sigma 0.6039 0.5232Log-likelihood -1338.90 -1119.86 -202.21LR chi2, Wald chi2 114.43 109.91 28.12 Prob > chi2 0 0 0.0005Pseudo R2 0.041 0.0468LR test 33.66 33.66 33.66Number of Observation 1695 1695 840
All Firms
Whole sample estimated results
Viriables Tobit Probit Truncated Tobit Probit Truncated
R&D intensity 1.8746*** 3.4348*** 0.7729 0.4453 0.3849 0.2109Skilled labors 0.0159 0.0141 0.1561 0.0242 0.0423 -0.0446Size 0.0643*** 0.0488*** 0.0150* 0.0136** 0.01189*** 0.0027Size2 -0.0013*** -0.0011*** 0.0005 -0.0001** -0.0001** 0.0000Constant -0.2354*** 0.0789 0.0103 0.0466Sigma 0.6272 0.5561 0.5573 0.4768Log-likelihood -871.34 -745.57 -115.12 -453.05 -362.64 -83.91LR chi2, Wald chi2 74.79 76.24 15.93 76.24 15.53 14.62 Prob > chi2 0.00 0.00 0.0258 0.0021 0.0298 0.0412Pseudo R2 0.0412 0.0486 0.0486 0.021LR test 3.3 3.3 3.3 12.99 12.99 12.99Number of Observation 1142 1142 504 553 553 336
Thai Firms Non-Thai Firms
Estimated results by ownership
21
Notifications from estimated results
Effects of R&D intensity on export performance of non-Thai firms and Thai firms are different
Export performance of non-Thai firms is not determined by R&D intensity
The results indicate that R&D behaviors of non-Thai firms can not well explained by this model. (see Markusen,1984)
Hence, it is reasonable to consider effects of R&D intensity on export performance of Thai firms rather than whole firms.
22
LR-test shows that tobit model is more appropriate Export performance of Thai firms is determined by R&D
intensity, firm’s size, and size2 The expected signs of all variables are correct with the
assumption. R&D intensity positively correlates with export
performance. If R&D intensity increases (decreases) one million baht, the conditional expected of export intensity increases (decreases) 1.8746 unit.
Hence, R&D intensity can improve export performance of Thai firms
Estimated results of Thai firms
23
A marginal effect of skilled labors has positive sign and is equal to 0.0159. If share of skilled labors in total employees increases (decreases) one unit, the conditional expected of export intensity increases (decreases) 0.0159 unit.
However, this variable is insignificant. Skilled labors do not affect export performance of manufacturing firms in Thailand.
Estimated results of Thai firms
24
As for firm’s size. If numbers of employees increase (decrease) by one hundred persons, the conditional expected of export intensity increases (decreases) by 0.0039.
Size2 has negative sign Firm’s size and export performance reveal the
inverted U-shaped relationship
Estimated results of Thai firms
25
Sector characteristics
IndustryTechnology
accumulationSource of
technology
Supplier dominated
Agriculture, construction, Services, Traditional manufacturing
Import technology / Little R&D
Specialized suppliers
Scale intensive
Petroleum and coal, iron and steel, motor vehicles, transportation equipments
Design, Operation , and learning by doing (Process innovation)
Internal
Specialized suppliers
Machinery building, instruments
Design & development Clients
Science basedChemical products, Electronics products, Pharmaceutical products
Academic research, Search for new technology
In-house R&D, external collaboration
The effect of R&D intensity with sector variation
26
Data descriptions
Classification All firms R&D firms % RD firms
All 1695 262 15.46Exporting firms 840 176 20.95Non-exporting firms 855 86 10.06
All 492 32 6.50Exporting firms 243 22 9.05Non-exporting firms 249 10 4.02
All 547 130 23.77Exporting firms 284 88 30.99Non-exporting firms 263 42 15.97
All 271 33 12.18Exporting firms 134 20 14.93Non-exporting firms 137 13 9.49
All 383 67 17.49Exporting firms 178 46 25.84Non-exporting firms 205 21 10.24
Specialize supplier industries
Science based industries
All sample
Supplier dominated industries
Scale intensive industries
Variables SDOM SCAI SPEC SCIBR&D intensity 2.0810** 2.7539* 0.9953 1.2946Skilled labors 0.0664 -0.1036 -0.2021 0.1775Size 0.0897*** 0.0811*** 0.1298* 0.0001Size2 -0.0020*** -0.0028*** -0.0052 0.0005Constant -0.2756*** -0.1767*** -0.1962** -0.2394***Sigma 0.6273 0.5982 0.5483 0.6777Log-likelihood -283.21 -285.75 -110.25 -177.28LR chi2, Wald chi2 51.03 26.43 11.2 6.78 Prob > chi2 0.00 0.00 0.0245 0.1482Pseudo R2 0.0826 0.0442 0.0483 0.0188
Estimated results by industries
Estimated results classified by industrial structure
R&D intensity determines export performance of firms in supplier dominated industries and scale intensive industries only.
Our results are similarly with the results of Kumar and Siddharthan (1994) for India and Dijk (2002) for Indonesia.
29
The 2nd objective
To investigate the differences in R&D behavior between exporting firms and non-
exporting firms
30
Based on Forbes and Wield (2000) and other previous literatures, R&D behaviors can be represented by
Analytical framework
Criteria Measurement
A presence of in-house R&D
% distribution of in-house R&D activities (in all R&D activities)
Numbers of scientists and engineers
The number of engineers and scientists in total employees
The numbers of own design product
1)%Products designed by firms according to customer requirements in total sales
2)%Products designed by firms and sold under own brand
Types of R&D % distribution of R&D expenditures on Basic, applied, experimental
Objectives of R&D R&D outputs (new or improved product/ process)
Level of external collaboration in R&D
1-5 Likert scale (evaluated the intensity of collaboration)
31
Analytical Results
All Firmswith R&D
NE SSE LSE
%Export =0
(0<%Export <50%)
Export≥ 50%
The in-house R&D0.90
(0.26)0.93
(0.21)0.88
(0.28)0.90
(0.26)
Numbers of scientists and engineers
0.07(0.10)
0.06(0.09)
0.10(0.12)
0.06(0.07)
Share of products designed by firms according to customer requirements in total sale*
0.15(0.29)
0.15(0.29)
0.10(0.21)
0.22(0.35)
Share of products designed and sell under own brand in total sale*
0.35(0.43)
0.26(0.41)
0.51(0.45)
0.25(0.39)
N 164 51 63 50
32
Analytical Results
All Firms
NE SSE LSE
%Export =0
(0<%Export<50%)
Export≥ 50%
Basic Research0.64
(0.31)0.59
(0.31)0.63
(0.31)0.70
(0.32)
Applied Research0.28
(0.30)0.31
(0.29)0.29
(0.31)0.25
(0.29)
Experimental Development
0.08(0.11)
0.10(0.12)
0.08(0.12)
0.05(0.09)
N 164 51 63 50
33
Analytical Results
All Firms
NE SSE LSE
%Export =0
(0<%Export<5
0%)
Export≥ 50%
Improve production process
0.39(0.49)
0.50(0.51)
0.29(0.46)
0.40(0.49)
Develop new production process
0.32(0.47)
0.29(0.46)
0.33(0.48)
0.34(0.48)
Improve quality of existing products
0.62(0.49)
0.75(0.44)
0.56(0.50)
0.58(0.50)
Develop new products0.67
(0.61)0.61
(0.49)0.76
(0.76)0.62
(0.49)
Granted patents0.09
(0.28)0.10
(0.30)0.05
(0.21)0.12
(0.33)
N 164 51 63 50
Non-exporter
ExporterMajor-
exporter
%Export =0 0<% Export <50%
Export≥ 50%
Customers, buyers 88 3.65 3.47 3.84 3.55Locally-owned suppliers 83 3.02 2.90 3.23 2.84Foreign-owned suppliers 71 2.80 2.60 2.93 2.88 Parent/associate company overseas 44 2.80 2.31 3.00 3.00Public research institutes 77 2.43 2.56 2.32 2.44Private non-profit 50 1.92 1.76 1.89 2.14Universities 74 2.62 2.39 2.76 2.67Business Service Providers 59 2.10 1.80 2.21 2.33Technical Service Providers 63 2.25 2.05 2.42 2.27Other government agencies 35 2.20 2.00 2.47 1.88Competitors 60 2.47 2.29 2.32 3.00
All FirmsN
External collaboration in R&D
35
Average % of in-house R&D of NE, SSE, LSE are not statistically different.
On average, SSE have more numbers of scientists and engineer than other sub-samples, but the distinction is not statistically different.
Design capabilities of LSE, SSE, and NFE are statistically different.
LSE typically design and produce products according to customer requirements.
SSE design and produce products, and then sell under their own brands.
Analytical Results
36
The results roughly present that exporting firms (both SSE and LSE) have design capabilities more than non-exporting firms.
On average, non-exporting firms mostly perform experimental R&D, and also successfully improve efficiency of production process as well as improve quality of existing products.
One possible explanation is that non-exporters carry out experimental R&D and conduct R&D with the purpose of succeed in the domestic markets, not for export markets.
Analytical Results
37
On average, exporting firms (SSE and LSE) cooperate with the external agents more than non-exporting firms.
Non-exporting firms intensely collaborate with public agents
However, the degrees of external collaborations in R&D do not statistically different among three sub-samples.
Analytical Results
38
Summary of the Finding
R&D intensity positively determines export performance of Thai firms, but not influence export performance of non-Thai firms.
The results support the explanation on R&D behaviors of MNEs by Markusen (1984).
Regarding Thai firms, R&D intensity affect export performance of firms in supplier dominated industries and scale intensive industries in positive way.
For specialize supplier and science based industries, R&D intensity has no effects on export performance.
39
Summary of the Finding
So far we can finish both objectives, 1st objective - R&D intensity positively influences
an export performance of manufactured firms in Thailand, but its impacts vary by industrial structures and ownership.
2nd objective - Thai R&D firms that export report higher designed capabilities than other firms.
Hence, apart from R&D intensity, it can be roughly said that designed capabilities are important.
40
Limitations and Suggestions for further studies
Restrictions of the data. It is difficult to measure R&D output. This study use
R&D input instead. The data is cross-section data, which do not allow us to
examine the effects of R&D more than one reference year. According to the fact that R&D require some period of time to yield benefits, studying on one year may too short.
In addition, many previous studies propose that, alternatively, export also causes R&D. Using cross-section data restrict to effectively analyze in dual ways.
41
Limitations and Suggestions for further studies
Limitations of quantitative analysis.
It does not give much information on firm’s R&D behaviors and export decision. In-depth study should be continued henceforth.
Furthermore, non-formal R&D also important for developing countries, further studies should draw attention on non-formal R&D activities as well.
42
End of presentation