Tóm tắt luận án: Knowledge spillover, sectoral innovation and firm total factor productivity The case of manufacturing industries in Vietnam.

i
MINISTRY OF EDUCATION AND TRAINING
UNIVERSITY OF ECONOMICS HO CHI MINH CITY
------------------------------------------------------

NGUYEN THI HOANG OANH
KNOWLEDGE SPILLOVER, SECTORAL INNOVATION
AND FIRM TOTAL FACTOR PRODUCTIVITY:
THE CASE OF MANUFACTURING INDUSTRIES
IN VIETNAM

Major: Development Economics
Code: 9310105

SUMMARY OF PHD THESIS
ACADEMIC ADVISORS
1. Dr. Pham Khanh Nam
2. Dr. Pham Hoang Van

HO CHI MINH CITY, 2021


ii

Dissertation is completed at:
.............................................................................................
Người hướng dẫn khoa học:
Academic supervisor 1: Dr. Pham Khanh Nam
Academic supervisor 2: Dr. Pham Hoang Van
Reviewer
1:……………………………………………………….

……………………………………………………………
………
Reviewer
2:………………………………………………………
……………………………………………………………
………
Reviewer
3:………………………………………………………
……………………………………………………………
………
Dissertation shall be defended before the economic
committee
at
the
University
level,
held
at:…………………………………………………………
………………………………………………………… .....
Time: ..................... Date ...................................................
This dissertation could be accessed at the following library:
.............................................................................................
.............................................................................................


iii

PUBLICATIONS OF RESULTS
Nguyen Thi Hoang Oanh, 2019. Determinants of Firms’ Total Factor
Productivity in Manufacturing Industry in Vietnam: An Approach of

a Cross-Classified Model. Journal of Asian Business and Economic
Studies (JABES), Volumn 26, Special Issue 01. Available from:
/>Nguyen Thi Hoang Oanh, 2018. Sector Innovation Capacity in
Vietnamese Enterprises: Spillover effects from Research and
Development (R&D), Foreign Direct Investment (FDI) and Trade.
Asian Conference on Business and Economic Studies (ACBES),
University of Economics Ho Chi Minh City, Ho Chi Minh City
Publishing House of Economics (ISBN: 978-604-922-660-1), pp. 265284


iv

Abstract
This study developed the framework of knowledge spillovers
at sector level and investigated these spillover effects of research
and development (R&D), foreign direct investment (FDI) and
trade activities on sectoral innovation by Spatial Regression
Models. Besides, the study examined the spillover effects of
sectoral innovation and provincial human resources on firms’ TFP
with 7,236 enterprises in 38 sectors of Vietnamese manufacturing
industries, located in 62 provinces by Cross-Classified Models.
By Spatial Regression Models with to 38 manufacturing sectors
in correspondence to Input/Output table from 2010 to 2014, the
intra-industry rather than inter-industry spillover effects were
found to be significant; that approved the hypothesis of MAR
rather than Jacobs externalities. In particular, only R&D and
export activities were found to have significantly positive effects
on innovation activities at sector level. In contrast, FDI and
import activities seem to have negative impact on innovation
activities. In cross-classified models, firms’ characteristics in

comparison with characteristics of sectors and provinces may
have the highest explanation on the heterogeneity in firms’ TFP.
The firm size, capital intensive and export orientation were found
to have stably significantly positive impacts on firms’ TFP. The
sectoral innovation might turn to have positive impacts on the
productivity of firms in that sector after one year. Besides, the
externalities of human resources in provinces on firms’
productivity were found to be positive.
Keywords: Knowledge Spillovers, Sectoral Innovation, TFP, Spatial
Regression Model, Cross-Classified Model


1

1. INTRODUCTION
1.1. Problem Statement
1.1.1. The importance of the topics in this thesis
It is important to investigate the role of knowledge spillovers on
innovation at sector level. As stated by Aghion and Jaravel (2015),
“innovations in one firm or one sector often build on knowledge
that was created by innovations in another firm or sector”. Mehrizi
and Ve (2008) argued that sector-level analysis enables the study
to link firm level determinants to macro-economic conditions.
Malerba (2002) also emphasized the role of sector-level analysis in
investigating innovative and production activities. According to
Padoan (1999), adopting a sectoral perspective may investigate the
knowledge accumulation and diffusion. In our knowledge, there
are few studies on the roles of channels of knowledge spillover on
sector innovation capacity. In Vietnam, there are few studies on
innovation and most of these studies focused on firm level

Therefore, the first main objective in this study is to investigate the
role of knowledge spillover on sectoral innovation through three
channels including R&D, FDI and trade activities by spatial
regression models.
It is also important to examine on heterogeneity of firms’ TFP
in considering both firms’ characteristics and spillover effects from
sectors and regions. TFP is understood as the residual of output that
is not contributed by the amount of capital and labor. In Solow
model (1956), the residual is a black box representing technical
change that leads to a sustainable development. Obviously, the


2
heterogeneity in firms’ TFP is mainly originated from the
differences in firms’ characteristics. Acemoglu (2009) stated that
“the heterogeneity in TFP are not necessarily due to technology in
the narrow sense. For instance, two firms have adopted the same
technology but make use of these techniques in different ways with
different degrees of efficiency”. However, even if these firms
adopted similar technology, they still have differences in TFP.
These differences may be originated from the characteristics of
their sectors or their location.
It is important to examine to the determinants of firms' TFP by
multileveled factors in a multilevel cross-classified model. This
model could isolate the impacts of elements at multilevel including
firm, sectoral, regional or provincial dimensions. However, most
of studies on firms’ TFP focused on the determinants as firms’
characteristics. In Vietnam, studies on TFP are still very limited
(CIEM, 2010) although TFP is recently perceived as a key role of
development quality. This study could make a contribution as a

new approach in investigating TFP in Vietnam by applying the
multileveled cross-classified model in the second objective. In
addition, the study may imply policies not only for firms but also
for sectors and regions.
1.1.2. The gaps and the new aspects in this thesis
There are three new aspects respectively on theoretical frame
work, methodology and context in this study. At theoretical
framework, the knowledge spillover at sector level was developed
by aggregating the stock of knowledge at firm level as in Cohen


3
and Levinthal (1989). Our model is new when it indicated not only
the intra-industry spillover but also the inter-industry spillover at
sector level and investigated the channel of knowledge spillover
from R&D, FDI and trade. In addition, this study revealed the
spillover effects of sectoral innovation and provincial human
capital on firms’ productivity basing on the ideas of intra-industry
economies of localization (Marshall, 1920), intra-sectoral
spillovers (Griliches, 1992) and the role of human capital spillover
on productivity (Moretti, 2004).
In regarding to the methodology, the study has two new
approaches. Innitially, the study adopted spatial regression model
in investigating sources of knowledge spillovers on sectoral
innovation. Then, a Cross-classified model was applied to make an
efficient estimate of the effects on firms’ productivity from firm
level, sectoral level and provincial level.
Besides, knowledge spillover, innovation and productivity,
integrated in this study, is a necessary topic in the context of
manufacturing sector in Vietnam. In the context of Vietnam, no

study investigated the determinants of firms’ TFP at firm, sector
and province level by Cross-classified model. Some studies have
considered such as FDI transaction (Ni et al., 2015; Vu Hoang
Duong and Le Van Hung, 2017; Khanh Le Phi Ho et al., 2018;
Nguyen, 2017) or agglomeration economies in manufacturing
industries (Francois and Nguyen, 2017; Toshitaka et al.; 2017) or
import competition in the sector (Doan et al., 2016). However,
there has been no study applying Cross- classified model. Adopting
this model in the case of 63 provinces and 38 sectors in


4
manufacturing industry makes this study more valuable in the
context of Vietnam.
1.2. RESEARCH OBJECTIVES
The first general objective is to investigate channels of
knowledge spillovers on sectoral innovation in manufacturing
industries in Vietnam, the study focuses on the following research
questions:
1.1. Is sectoral innovation directly affected by R&D activities
of that sector in manufacturing industries in Vietnam?
1.2. Is sectoral innovation indirectly affected by R&D
activities of other sectors in manufacturing industries in Vietnam?
1.3. Is sectoral innovation directly affected by transactions
with FDI enterprises in that sector in manufacturing industries in
Vietnam?
1.4. Is sectoral innovation indirectly affected by transactions
with FDI enterprises in other sectors in manufacturing industries
in Vietnam?
1.5. Is sectoral innovation directly affected by trade activities

in that sector in manufacturing industries in Vietnam?
1.6. Is sectoral innovation indirectly affected by trade
activities in other sectors in manufacturing industries in Vietnam?
The second objective of this study is to investigate the impacts
of characteristics at firm- level, regional and sectoral level on
firms’ total factor productivity (TFP) with the following research
questions:


5
2.1. How much heterogeneity in firms’ total factor
productivity is explained by firm-level, sector-level and provincelevel determinants?
2.2. Does firms’ size have impact on firms’ TFP in
manufacturing industries in Vietnam?
2.3. Does the capital intensity in firms have impact on their
TFP?
2.4. Is there difference in TFP of exported firms and nonexported firms?
2.5. Is firms’ TFP affected by their sectoral innovation in
manufacturing industries in Vietnam?
2.6. Does the human resource in a province have impact on
the TFP of firms in that province?
1.3. RESEARCH METHODOLOGY and RESEARCH
SCOPE
In order to investigate three channels of knowledge spillovers
on sector innovation capacity, this study applied the Spatial
Regression. Then the study applied the cross classified model to
examine the heterogeneity in firm productivity from three groups
of determinants including sector, regional and firm level. This
study made use of the data of Vietnam Enterprises Survey (VES)
and Vietnam Technology and Competitiveness Survey (TCS) in

addition to the use of Input Output (I/O) of Vietnam in 2012.
Besides, the study also used the annually surveyed data on province
of General Statistics Office (GSO).
The analysis unit in investigating the effect of R&D, FDI and
trade on sectoral innovation is sector. The sector unit is aggregated


6
from data on Vietnamese firms in manufacturing from the year of
2010 to 2014. The relations among sectors are determined basing
the intermediary transaction in the Input Output of Vietnam in
2012. By spatial regression model, the study finds the direct as well
as indirect impact of R&D, FDI and trade on sectoral innovation.
Meanwhile, firm is the analysis unit in investigating the
impacts of characteristics at firm- level, regional and sectoral level
on firms’ total factor productivity (TFP). Firms are also in
manufacturing industries in Vietnam with research period from the
year of 2011 to 2014. Using TCS and VES data, the study accesses
the characteristics at the firm level. The sectoral characteristics in
the model is also measured from these data. In addition, the annual
province data on Province Competitive Index (PCI) is also used to
determine the human resources at the province.
1.4. RESEARCH CONTRIBUTION
This study could have contributions on theoretical perspective
as well as policy implication. On theoretical perspective, this study
developed the framework and tested the hypothesis of knowledge
spillover at sector level. The study applied a new approach, Spatial
Regression Model, to investigate the knowledge spillovers among
sectors. Besides, the study tried to explore the black box of
contextual factors on firms’ TFP. In particular, the study applied

the Cross-classified Model to investigate the spillover effects of
innovation activities at sector level and human resources at
province level on firms’ TFP.
Determining the core spillover factors on sector innovation
capacity is key information for policymakers to enhance this sector


7
capacity. In addition, the Cross-classified Model also enables
policymakers to know how important are firm characteristics,
sectoral and provincial level attributed to firms’ TFP.
1.5. STRUCTURE OF THIS STUDY
This study consists of five chapters. The first chapter is the
Introduction. The second chapter is the Literature Review that
contains the Theoretical framework and Empirical Studies of two
general objectives. The next chapter, Methodology, shall illustrate
the nature of the Spatial Regression Model and the CrossClassified Model. In addition, the chapter also presents the Model
Specification, Variable measurement and the data. The two
following chapters is the chapters of Result and Discussion. One
chapter provides results and discussions on the Sectoral Innovation
and Spillover effects. The other chapter provides results and
discussions on heterogeneity in TFP of Vietnamese manufacturing
firms. The final chapter is the Conclusion and Policy Implications.
2. LITERATURE REVIEW
2.1 DEFINITION
2.1.1. Knowledge spillovers
Knowledge spillovers are generally defined as gaining
benefits from other parties’ investment in knowledge without
paying its full price since knowledge can ‘spillover’ from one agent
to another. The concept of this spillover is originated from the

public good nature of knowledge which is non-rival and nonexcludable (Arrow, 1962). Depending on types of knowledge, the
knowledge spillovers could be voluntarily or involuntarily


8
transmitted between agents (Romer, 1990). Kaiser (1960) also
stated that knowledge spillovers may be originated from failures in
the protecting knowledge generated in an innovating firm. The
amount of this non- appropriable knowledge is called ‘knowledge
spillover’. Basing on these ideas, Griliches (1992) proposed that
investments in knowledge have a high propensity to spill over for
commercialization by third-party firms which do not pay for the
full cost of accessing and implementing those ideas.
2.1.2. Innovation
OECD (2005) made the definition of innovation as follows
“an innovation is the implementation of a new or significantly
improved product (good or service), or process, a new marketing
method, or a new organizational method in business practices,
workplace organization or external relations.”
Due to the trend of economic development, several studies pay
more attention development of organization and marketing terms
and base on innovation definition in OECD (2005). This manual
characterized innovation as the introduction of a new or
significantly improved product (goods or services); a new or
fundamentally improved process, a new marketing method, or a
new organization method in terms of business practice, association
of work environment.
2.1.3. Knowledge production function and the determination
of innovation in this study
This study based on the knowledge production function

(KPF), formerly proposed by Pakes and Griliches (1984), to


9
determining the innovation and its determinants in the model.
Pakes and Griliches (1984) illustrated the simplified diagram of
the knowledge production function as follows:

Figure 0. The framework of knowledge production function
Source: Pakes and Griliches (1984)
In this diagram,

is produced by a knowledge production

function (KPF) which translates past research expenditures, R, and
a disturbance term, U, into inventions. The disturbance term
reflects the combined effect of other nonformal R&D inputs and
the inherent randomness in the production of inventions.
2.1.4. Sectoral Innovation System (SIS) and its determinants
According to Malerba (2002), the founder of sectoral
innovation system, “sectors provide a key level of analysis for
economists, nosiness scholars, technologists and economic
historians in the examination of innovative and production
activities”. He proposed that a sectoral system includes products
and the set of agents which make market and non-market


10
interactions for creating, producing and selling those products. A
sectoral system has a particular knowledge base, technologies,

inputs and demand. The interactions may emerge among the agents
in a sectoral system. Agents are known as individuals and
organizations at various levels of aggregation. The interaction
among agents may be created through process of communication,
exchange, cooperation, competition and command, and these
interactions are shaped by institutions. Therefore, he suggested that
the sectoral innovation system could be used to explain the
creation, absorption, sharing and utilization of knowledge and
innovation in a sector.
2.1.5. Total Factor Productivity (TFP)
Total Factor Productivity (TFP) identifies the portion of
output not explained by traditionally measured inputs of labor and
capital. It was widely known that output is a function of the inputs
used by a firm and its productivity (Katayama, Lu and Tybout,
2009). Basically, the Cobb-Douglas production function is used to
measure TFP.
The choice of measurement methods on TFP in this study
based on the comparison of four principal methods including Fixed
effects, Instrumental variables and GMM, the semi parametric
estimation algorithm developed by Olley and Pakes (1996) and the
semi parametric estimation algorithm developed by Levinsohn and
Petrin (2003).
2.2 THEORETICAL FRAMEWORK
2.2.1. Developing model on Knowledge Spillovers at sector-level


11
We began our construction of functional form of the equation
that may connect these variables in our data at sector level from
the equation at firm level of Cohen and Levinthal (1989).

Cohen and Levinthal (1989) constructed a model of firm’s
stock knowledge as follow:
=

( ∑

+

Where

+ T)

(2.19)

the firm’s i stock of technological and scientific
is a firm’s investment in R&D;

knowledge;

is the fraction

of knowledge in the public domain that the firm is able to
assimilate and exploit and represents the firm’s absorptive
capacity;

is the degree of intra-industry spillovers and T is the

level of extra-industry knowledge. Other firm’s investment in
research and development is

for j≠i also contribute to . This

model implies the intra as well as inter sectoral knowledge
spillover among sectors.
We defined Zs to be the total output of knowledge in the
sector s: Zs =∑

.

(2.20)

Similarly, Ms is the total input of knowledge in the sector s:
=∑

.

(2.21)

After the transformation, we have:
 .
+∑
=
+ ( − 1).
.∑

.

( ≠ !) +

. .∑

" ∑

+
(2.38)

In order to express the inter-industry knowledge spillover, the
study made a basis on the ideas of Griliches (1992).
The amount of aggregate knowledge borrowed by the ith
industry from all available sources was expressed by Griliches
(1992) as follows:


12
= ∑$

(2.39)

Finally, we have:
.
∑

=

+(

− 1).

. ∑ $% . &% (' ≠ () +

. .∑

+

.

( ≠ !) -

.∑

∑

(2.41)

Following the knowledge production function of Griliches
and Pakes (1984), the study constructed the model of the sources
of knowledge spillover on sectoral innovation.
2.2.2. Channels of knowledge spillovers and the research
hypothesis of the first objective
Griliches (1979) argued that the level of knowledge in any
sector or industry not only is derived from "own" research and
development investments but also is affected by the knowledge
borrowed or stolen from other sectors or industries. Thus, the
productivity of industry i will depend also on the research and
development investments of industries j and h, among others.
Basing on this proposition, the study tests the hypothesis on the
direct impact of R&D on innovation within a sector and the
indirect impact of R&D in other sectors on innovation of a sector
as follows:
H11: The research and development (R&D) in the sector i may

have positive impact on its sectoral innovation.
H12: The sectoral innovation in the sector i may be positively
affected by the R&D from other related sectors.
Hofmann and Wan (2013) suggested that the horizontal
externalities from FDI may have direct or indirect on domestic


13
firms in its same industry by four channels including competition,
imitation and adoption, labor turnover and second round effects
through input suppliers. Markusen and Venables (1997) provided
an analytical framework which can assess the effects of the
industrial linkages. They proposed that at the industry level, the
presence of FDI may change supplies and demands in a number
of related industries.
On the basis of the potential backward and forward
externalities from FDI suggested by Markusen and Venables
(1997), Hofmann and Wan (2013), the study had the following
hypothesis:
H13: The transaction with FDI enterprises in the sector i may
enhance its sectoral innovation.
H14: The sectoral innovation in the sector i may be affected by
the transaction with FDI enterprises in other related sectors.
Grossman and Helpman (1991), henceforth GH who
formulate a theoretical model where the foreign contribution to the
local knowledge capital stock increases with the number of
commercial interactions between domestic and foreign agents.
Basing on the assumption of Grossman and Helpman (1991) that
the number of commercial interactions between domestic and
foreign agents may increase the local knowledge capital stock, the

study had the hypothesis on both the direct and indirect impacts of
exports and imports as follows:
H15a: The export of the sector i may upgrade the innovation
capacity of its sectoral innovation.


14
H15b: The input import of the sector i may upgrade the
innovation capacity of its sectoral innovation.
H16a: The sectoral innovation of the sector i may be affected
by export of other related sectors.
H16b: The sectoral innovation of the sector i may be affected
by input import of other related sectors.
2.2.3. Theoretical framework of knowledge spillovers to firms
2.2.3.1. Debates on knowledge spillover of intra- sector to firms
Griliches (1991) verified the intra-industry spillover effects by
measuring total factor productivity as follows:
A = Y/X = ). * +,-. .

/

. 0 12,3

(2.48)

As presented, the TFP depends not only on the conventional
inputs or research capital but also the contribution of the trend t in
the other unmeasured factors. The unmeasured factors were not
specified in the framework of Griliches (1991). We argue that
these unmeasured factors may include context determinants

relating to sector-level and province-level. The context
determinants were suggested in this study to be innovation
activities in the sector and level of human resources in the
province.
2.2.3.2. Human capital externalities from the province to firms
Moretti (2004) is one of first studies that built the framework
and directly estimated the human capital externalities on the
productivity of manufacturing plants. In order to illustrate the
nature of a spatial equilibrium in the presence of human capital
spillover, Moretti (2004) built a general equilibrium framework.
Considering two cities and two types of labor, educated and


15
uneducated workers, he assumed that there are two types of goods,
a composite good y- national traded and land h-locally traded.
Using a Cobb-Douglas function, each city is a competitive
economy with the production of firms as following:
Y = A 454 656 78

(2.49)

where H and L are the hours worked by skilled and unskilled
workers, respectively, and K is capital. In order to find the
possibility of human capital externalities, he allowed the
productivity of plants in a city to depend on the aggregate level of
human capital in the city: A = f(9̅) in which 9̅ is the fraction of
college-educated workers in the city, outside the firm.
2.2.4. Multilevel modeling on firms’ total factor productivity and
the research hypothesis of the second objective

Basing on theory on internal economies of scale (Silberston A.,
1972), we test the following hypothesis:
H21: The firm size may have positive effect on the firm’s
productivity.
In decomposing the components of firms’ TFP in the United
States, Solow (1962) found that the firms’ TFP is most affected by
firms’ technology in comparison with capital and labor. Basing on
this, the study test the following hypothesis:
H22: Firms with have higher capital per worker may have
higher productivity.
Basing on the idea of learning by exporting (LBE) began to be
discussed and studied (empirically and theoretically) in the mid-80s
with Rhee et al. (1984), Westphal et al. (1984) and in the 90s with


16
Grossman and Helpman (1991) and the World Bank (1993), the study
test the following hypothesis:
H23: Firms with exports activities may have higher
productivity than firms without exports.
Basing on Griliches (1992) about the intra-industry spillover
effects to firms, this study tests the hypothesis that firms located in
the high localized sector may have higher TFP than the others as
follows:
H24: The sectoral innovation may have positive spillover effect
on firms’ productivity in that sector in the same year.
H25: The sectoral innovation in the previous year may have
positive spillover effect on firms’ productivity in that sector in the
current year.
Basing on the framework of Moretti (2004), this study tried to

explore the human capital externalities of the province to firms’
productivity in that province by the following hypothesis:
H26: The vocational human resource in a province may have
positive effects on firms’ productivity in that province.

2.3 EMPERICAL STUDIES
In this section, the study reviewed the empirical studies relating
to three following topics. The first one is the empirical studies on
determinants of sectoral innovation. The next is the empirical studies
on channels of knowledge spillover and applications of Spatial
Regression Model. And the final one is the empirical studies on TFP.


17
Figure 0. Theoretical framework of the study

Source: By author’s own


18
3. RESEARCH METHODOLOGY
3.1 The research model on Sectoral Innovation
This study follows Spatial Durbin Model (SDM) which is an
appropriate approach to investigate the externalities (Beer and
Riedl, 2010) as follows:
;2 =δ∑

ᵂ

;2+∑

ᵂ

*%

2

θk + *% 2 =% + Zkit γk +εit

(*) (t=1……T, i=1…n)

(3.1)

In which the dependent variable ; 2 in (1) is respectively
measured by S_modified and S_Innovation. The interaction
weighted

repressors,ᵂ

S_FDI_Supplierit,

*% 2 ,

namely

S_FDI_Customerit,

S_RD_meanit,
S_exportit

and

S_InputImportit. In this model, yit is the innovation activity of
sector i in the period t. Wij yjt is the interaction weighted dependent
variable, ᵂ

*%

2

is the interaction weighted regressors and Zkit

are control variables. The description of variables in the models are
as in the Table 3.1.
The sector of the firm is determined basing on its principal 4digit VSIC sector. The principal sector herein is meant to be the
sector in that firms have highest value of production or sales or
use highest number of employees. In order to make it
correspondent with the division of sector in Input/Output Table,
the study makes group of manufacturing sectors in VSIC to 38
manufacturing sectors in correspondence to Input/Output table (as
detail in The Appendix 1, page xx). This generates a panel data of
190 observations from the year of 2010 to 2014.


19
3.2 The research on firms’ TFP heterogeneity by CrossClassified Model
In order to investigate the separate effect of firm
characteristics, sector and region, the simple model without any
independent variables is firstly estimated as follow:
;(

)

=

>>>

Where ; (
the region j,

)

+ ? + ? +? +0(

(3.29)

)

is TFP of the firm in the sector s and located in

>>>

is the mean TFP across all sectors and all

regions. ? is the effect of firm i’s sector. ? is the effect of firm
i’s region. 0 (

)

is the firm level residual error term. Besides, this

model includes a random interaction effects between sector and
region, ? .
The estimation result of the equation (3.29) provides how
much firm characteristics, sector and region are attributed to the
heterogeneity firms ’TFP. Basing on this result, the study
considers to add variables on firm characteristics, sector and
region specific into the following model:
; ( ) = >>> + ∑@
@ *@ ( ) +
%
∑C BC . 9C + ? + ? + 0 ( )

∑A

=A . &A

+
(3.30)

Where y is the TFP of the firm i (in logs) belonging to sector
s and established in region j, X is a vector of m firm-level variables
which may be important determinants of TFP, Z presents the
variables at the regional level, S is the variables at the sectoral
level. This study considered the result estimation of the equation
(3.30) to include the number of variables in each investigated
level.


20

This study controlled the issue of omitted variable bias and
endogeneity by make comparison the results of the Crossclassified Model with the ones of the Fixed Effects model. In
addition, the study also applied the Hausman – Taylor method to
obtain a more efficient estimator and perform valid statistical
inference using that estimator.
The description of variables in the models are as in the Table
3.2. The VES data is used to measure the firm’s TFP from the year
2011 to 2015. In addition to the VES data, this study also makes
use of the data on province of General Statistics Office (GSO). All
enterprises exists during the period from 2011 to 2014 are
combined into a balance panel data with the number of
observations to be 7,236 enterprises per year and the total number
of observations to be 28,944. This data covers 38 sectors of firms
located in 62 provinces.
4. SECTORAL INNOVATION AND SPILLOVER
EFFECTS: RESULTS FROM SPATIAL REGRESSION
MODELS AND DISCUSSIONS
The results of the models in the section 4.3 have been mostly
consistent. The positive direct impacts of R&D activities on both
innovation activities and modification activities in a sector
approved the hypothesis of Cohen and Levinthal (1989) and
Griliches (1992). However, there was no evidence of indirect
effects of R&D activities on the other sectors’ innovation activities
or modification activities. On regarding to the impact of FDI, the
study has not found any positive impact as stated in the hypothesis
of Hofmann and Wan (2013) and Markusen and Venables (1997).


21
In contrast, this study found the negative impact of the number of

FDI suppliers in a sector on its innovation or modification
activities. In respect of exports, this study found the positive
impacts of exports on innovation activities or modification
activities in a sector. This result approved the assumption of
Grossman and Helpman (1991). In contrast, the effects of imported
inputs on was negative in innovation activities or modification
activities in this study.
5. HETEROGENEITY IN TFP OF VIETNAMESE
MANUFACTURING FIRMS: RESULTS FROM CROSSCLASSIFIED MODELS AND DISCUSSIONS
The estimation results on the determinants of firms’ TFP has
been consistent across the models. Basing on the results of LR
tests in the Table A5 and Table A6 (Appendix, page xxiv), the
cross-classified model was confirmed to be better fit to the data.
This study carefully made the comparison between the model with
random effects in both sector level and province level and the
other model with fixed effects. The consistence in results of
random effects and fixed effects model may be a good sign to
apply random effects model (Bell and Jones, 2015).
This study provided evidences on determinants of firms’ TFP.
The positive effect of firm size on the firm’s TFP approved the
hypothesis of economies of scale (Silberston A., 1972).
Additionally, the study also approved the positive contribution of
capital to productivity as in Solow (1962). The positive impacts of
export orientation on TFP found in this study also approved for